Add files using upload-large-folder tool

Qwen3-30B-A3B-Instruct-2507-sft-fusang-swa=2k_sink=0/checkpoint-870/adapter_config.json

@@ -0,0 +1,38 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "/share/models/Qwen3-30B-A3B-Instruct-2507",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": true
+}

Qwen3-30B-A3B-Instruct-2507-sft-fusang-swa=2k_sink=0_falayer=25_from0_step2_fadec/checkpoint-994/README.md

@@ -0,0 +1,207 @@
+---
+base_model: /share/models/Qwen3-30B-A3B-Instruct-2507
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:/share/models/Qwen3-30B-A3B-Instruct-2507
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.17.1

Qwen3-30B-A3B-Instruct-2507-sft-fusang-swa=2k_sink=0_falayer=25_from0_step2_fadec/checkpoint-994/trainer_state.json

@@ -0,0 +1,1420 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 994,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.005033345916698125,
+      "grad_norm": 0.14038996398448944,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.6828,
+      "step": 5
+    },
+    {
+      "epoch": 0.01006669183339625,
+      "grad_norm": 0.1702621877193451,
+      "learning_rate": 1.8e-05,
+      "loss": 0.6646,
+      "step": 10
+    },
+    {
+      "epoch": 0.015100037750094376,
+      "grad_norm": 0.13121849298477173,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.568,
+      "step": 15
+    },
+    {
+      "epoch": 0.0201333836667925,
+      "grad_norm": 0.11524524539709091,
+      "learning_rate": 3.8e-05,
+      "loss": 0.6609,
+      "step": 20
+    },
+    {
+      "epoch": 0.025166729583490626,
+      "grad_norm": 0.1256837546825409,
+      "learning_rate": 4.8e-05,
+      "loss": 0.6599,
+      "step": 25
+    },
+    {
+      "epoch": 0.03020007550018875,
+      "grad_norm": 0.11717186123132706,
+      "learning_rate": 5.8e-05,
+      "loss": 0.6276,
+      "step": 30
+    },
+    {
+      "epoch": 0.03523342141688687,
+      "grad_norm": 0.1608511060476303,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.6536,
+      "step": 35
+    },
+    {
+      "epoch": 0.040266767333585,
+      "grad_norm": 0.14913535118103027,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.6647,
+      "step": 40
+    },
+    {
+      "epoch": 0.045300113250283124,
+      "grad_norm": 0.10683944076299667,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.6059,
+      "step": 45
+    },
+    {
+      "epoch": 0.05033345916698125,
+      "grad_norm": 0.10477400571107864,
+      "learning_rate": 9.8e-05,
+      "loss": 0.6717,
+      "step": 50
+    },
+    {
+      "epoch": 0.055366805083679374,
+      "grad_norm": 0.1359020173549652,
+      "learning_rate": 9.957627118644068e-05,
+      "loss": 0.6392,
+      "step": 55
+    },
+    {
+      "epoch": 0.0604001510003775,
+      "grad_norm": 0.09668928384780884,
+      "learning_rate": 9.904661016949153e-05,
+      "loss": 0.6378,
+      "step": 60
+    },
+    {
+      "epoch": 0.06543349691707563,
+      "grad_norm": 0.10658483952283859,
+      "learning_rate": 9.851694915254239e-05,
+      "loss": 0.5784,
+      "step": 65
+    },
+    {
+      "epoch": 0.07046684283377375,
+      "grad_norm": 0.10898926854133606,
+      "learning_rate": 9.798728813559322e-05,
+      "loss": 0.609,
+      "step": 70
+    },
+    {
+      "epoch": 0.07550018875047187,
+      "grad_norm": 0.08570314198732376,
+      "learning_rate": 9.745762711864407e-05,
+      "loss": 0.6271,
+      "step": 75
+    },
+    {
+      "epoch": 0.08053353466717,
+      "grad_norm": 0.09355496615171432,
+      "learning_rate": 9.692796610169492e-05,
+      "loss": 0.5937,
+      "step": 80
+    },
+    {
+      "epoch": 0.08556688058386813,
+      "grad_norm": 0.09191977977752686,
+      "learning_rate": 9.639830508474576e-05,
+      "loss": 0.6183,
+      "step": 85
+    },
+    {
+      "epoch": 0.09060022650056625,
+      "grad_norm": 0.09747638553380966,
+      "learning_rate": 9.586864406779662e-05,
+      "loss": 0.6345,
+      "step": 90
+    },
+    {
+      "epoch": 0.09563357241726438,
+      "grad_norm": 0.09679915755987167,
+      "learning_rate": 9.533898305084746e-05,
+      "loss": 0.5907,
+      "step": 95
+    },
+    {
+      "epoch": 0.1006669183339625,
+      "grad_norm": 0.1033581793308258,
+      "learning_rate": 9.48093220338983e-05,
+      "loss": 0.5628,
+      "step": 100
+    },
+    {
+      "epoch": 0.10570026425066063,
+      "grad_norm": 0.09648433327674866,
+      "learning_rate": 9.427966101694917e-05,
+      "loss": 0.6457,
+      "step": 105
+    },
+    {
+      "epoch": 0.11073361016735875,
+      "grad_norm": 0.09600954502820969,
+      "learning_rate": 9.375e-05,
+      "loss": 0.6206,
+      "step": 110
+    },
+    {
+      "epoch": 0.11576695608405688,
+      "grad_norm": 0.09261015802621841,
+      "learning_rate": 9.322033898305085e-05,
+      "loss": 0.5643,
+      "step": 115
+    },
+    {
+      "epoch": 0.120800302000755,
+      "grad_norm": 0.08777470141649246,
+      "learning_rate": 9.26906779661017e-05,
+      "loss": 0.6377,
+      "step": 120
+    },
+    {
+      "epoch": 0.12583364791745313,
+      "grad_norm": 0.07885514199733734,
+      "learning_rate": 9.216101694915254e-05,
+      "loss": 0.6263,
+      "step": 125
+    },
+    {
+      "epoch": 0.13086699383415126,
+      "grad_norm": 0.146609365940094,
+      "learning_rate": 9.163135593220339e-05,
+      "loss": 0.5661,
+      "step": 130
+    },
+    {
+      "epoch": 0.13590033975084936,
+      "grad_norm": 0.07068520039319992,
+      "learning_rate": 9.110169491525424e-05,
+      "loss": 0.5898,
+      "step": 135
+    },
+    {
+      "epoch": 0.1409336856675475,
+      "grad_norm": 0.0955682322382927,
+      "learning_rate": 9.057203389830509e-05,
+      "loss": 0.6161,
+      "step": 140
+    },
+    {
+      "epoch": 0.14596703158424562,
+      "grad_norm": 0.09020308405160904,
+      "learning_rate": 9.004237288135594e-05,
+      "loss": 0.6149,
+      "step": 145
+    },
+    {
+      "epoch": 0.15100037750094375,
+      "grad_norm": 0.08094290643930435,
+      "learning_rate": 8.951271186440678e-05,
+      "loss": 0.5707,
+      "step": 150
+    },
+    {
+      "epoch": 0.15603372341764188,
+      "grad_norm": 0.08532893657684326,
+      "learning_rate": 8.898305084745763e-05,
+      "loss": 0.625,
+      "step": 155
+    },
+    {
+      "epoch": 0.16106706933434,
+      "grad_norm": 0.4635956585407257,
+      "learning_rate": 8.845338983050848e-05,
+      "loss": 0.6457,
+      "step": 160
+    },
+    {
+      "epoch": 0.16610041525103814,
+      "grad_norm": 0.11454236507415771,
+      "learning_rate": 8.792372881355933e-05,
+      "loss": 0.5587,
+      "step": 165
+    },
+    {
+      "epoch": 0.17113376116773626,
+      "grad_norm": 0.09114626795053482,
+      "learning_rate": 8.739406779661017e-05,
+      "loss": 0.6043,
+      "step": 170
+    },
+    {
+      "epoch": 0.17616710708443437,
+      "grad_norm": 0.14178045094013214,
+      "learning_rate": 8.686440677966102e-05,
+      "loss": 0.6199,
+      "step": 175
+    },
+    {
+      "epoch": 0.1812004530011325,
+      "grad_norm": 0.08791183680295944,
+      "learning_rate": 8.633474576271187e-05,
+      "loss": 0.593,
+      "step": 180
+    },
+    {
+      "epoch": 0.18623379891783062,
+      "grad_norm": 0.12179736793041229,
+      "learning_rate": 8.580508474576272e-05,
+      "loss": 0.6315,
+      "step": 185
+    },
+    {
+      "epoch": 0.19126714483452875,
+      "grad_norm": 0.07060039043426514,
+      "learning_rate": 8.527542372881356e-05,
+      "loss": 0.6341,
+      "step": 190
+    },
+    {
+      "epoch": 0.19630049075122688,
+      "grad_norm": 0.08048715442419052,
+      "learning_rate": 8.474576271186441e-05,
+      "loss": 0.6035,
+      "step": 195
+    },
+    {
+      "epoch": 0.201333836667925,
+      "grad_norm": 0.07930552959442139,
+      "learning_rate": 8.421610169491526e-05,
+      "loss": 0.5854,
+      "step": 200
+    },
+    {
+      "epoch": 0.20636718258462314,
+      "grad_norm": 0.07271519303321838,
+      "learning_rate": 8.368644067796611e-05,
+      "loss": 0.6244,
+      "step": 205
+    },
+    {
+      "epoch": 0.21140052850132127,
+      "grad_norm": 0.12227227538824081,
+      "learning_rate": 8.315677966101694e-05,
+      "loss": 0.6368,
+      "step": 210
+    },
+    {
+      "epoch": 0.21643387441801937,
+      "grad_norm": 0.09962992370128632,
+      "learning_rate": 8.26271186440678e-05,
+      "loss": 0.5928,
+      "step": 215
+    },
+    {
+      "epoch": 0.2214672203347175,
+      "grad_norm": 0.07313612848520279,
+      "learning_rate": 8.209745762711865e-05,
+      "loss": 0.6272,
+      "step": 220
+    },
+    {
+      "epoch": 0.22650056625141562,
+      "grad_norm": 0.0753592848777771,
+      "learning_rate": 8.15677966101695e-05,
+      "loss": 0.6263,
+      "step": 225
+    },
+    {
+      "epoch": 0.23153391216811375,
+      "grad_norm": 0.09605638682842255,
+      "learning_rate": 8.103813559322035e-05,
+      "loss": 0.5929,
+      "step": 230
+    },
+    {
+      "epoch": 0.23656725808481188,
+      "grad_norm": 0.06890146434307098,
+      "learning_rate": 8.050847457627118e-05,
+      "loss": 0.6074,
+      "step": 235
+    },
+    {
+      "epoch": 0.24160060400151,
+      "grad_norm": 0.07414574921131134,
+      "learning_rate": 7.997881355932204e-05,
+      "loss": 0.6538,
+      "step": 240
+    },
+    {
+      "epoch": 0.24663394991820814,
+      "grad_norm": 0.07699766010046005,
+      "learning_rate": 7.944915254237289e-05,
+      "loss": 0.5952,
+      "step": 245
+    },
+    {
+      "epoch": 0.25166729583490627,
+      "grad_norm": 0.07343696802854538,
+      "learning_rate": 7.891949152542372e-05,
+      "loss": 0.5924,
+      "step": 250
+    },
+    {
+      "epoch": 0.2567006417516044,
+      "grad_norm": 0.11014200001955032,
+      "learning_rate": 7.838983050847458e-05,
+      "loss": 0.6456,
+      "step": 255
+    },
+    {
+      "epoch": 0.2617339876683025,
+      "grad_norm": 0.09955993294715881,
+      "learning_rate": 7.786016949152542e-05,
+      "loss": 0.6015,
+      "step": 260
+    },
+    {
+      "epoch": 0.26676733358500065,
+      "grad_norm": 0.08594182878732681,
+      "learning_rate": 7.733050847457628e-05,
+      "loss": 0.5748,
+      "step": 265
+    },
+    {
+      "epoch": 0.2718006795016987,
+      "grad_norm": 0.08562752604484558,
+      "learning_rate": 7.680084745762713e-05,
+      "loss": 0.6403,
+      "step": 270
+    },
+    {
+      "epoch": 0.27683402541839686,
+      "grad_norm": 0.07596726715564728,
+      "learning_rate": 7.627118644067796e-05,
+      "loss": 0.6365,
+      "step": 275
+    },
+    {
+      "epoch": 0.281867371335095,
+      "grad_norm": 0.08623526990413666,
+      "learning_rate": 7.574152542372882e-05,
+      "loss": 0.5757,
+      "step": 280
+    },
+    {
+      "epoch": 0.2869007172517931,
+      "grad_norm": 0.06453239172697067,
+      "learning_rate": 7.521186440677967e-05,
+      "loss": 0.6051,
+      "step": 285
+    },
+    {
+      "epoch": 0.29193406316849124,
+      "grad_norm": 0.0846736952662468,
+      "learning_rate": 7.46822033898305e-05,
+      "loss": 0.5988,
+      "step": 290
+    },
+    {
+      "epoch": 0.29696740908518937,
+      "grad_norm": 0.07134438306093216,
+      "learning_rate": 7.415254237288137e-05,
+      "loss": 0.5929,
+      "step": 295
+    },
+    {
+      "epoch": 0.3020007550018875,
+      "grad_norm": 0.08004368841648102,
+      "learning_rate": 7.36228813559322e-05,
+      "loss": 0.5852,
+      "step": 300
+    },
+    {
+      "epoch": 0.30703410091858563,
+      "grad_norm": 0.07338687032461166,
+      "learning_rate": 7.309322033898306e-05,
+      "loss": 0.64,
+      "step": 305
+    },
+    {
+      "epoch": 0.31206744683528376,
+      "grad_norm": 0.07914911955595016,
+      "learning_rate": 7.256355932203391e-05,
+      "loss": 0.5951,
+      "step": 310
+    },
+    {
+      "epoch": 0.3171007927519819,
+      "grad_norm": 0.10672074556350708,
+      "learning_rate": 7.203389830508474e-05,
+      "loss": 0.5861,
+      "step": 315
+    },
+    {
+      "epoch": 0.32213413866868,
+      "grad_norm": 0.07805804908275604,
+      "learning_rate": 7.15042372881356e-05,
+      "loss": 0.6347,
+      "step": 320
+    },
+    {
+      "epoch": 0.32716748458537814,
+      "grad_norm": 0.06919122487306595,
+      "learning_rate": 7.097457627118644e-05,
+      "loss": 0.6273,
+      "step": 325
+    },
+    {
+      "epoch": 0.33220083050207627,
+      "grad_norm": 0.09325078129768372,
+      "learning_rate": 7.044491525423729e-05,
+      "loss": 0.5759,
+      "step": 330
+    },
+    {
+      "epoch": 0.3372341764187744,
+      "grad_norm": 0.0745769739151001,
+      "learning_rate": 6.991525423728815e-05,
+      "loss": 0.6059,
+      "step": 335
+    },
+    {
+      "epoch": 0.34226752233547253,
+      "grad_norm": 0.07978899031877518,
+      "learning_rate": 6.938559322033898e-05,
+      "loss": 0.6229,
+      "step": 340
+    },
+    {
+      "epoch": 0.34730086825217066,
+      "grad_norm": 0.07445425540208817,
+      "learning_rate": 6.885593220338984e-05,
+      "loss": 0.5845,
+      "step": 345
+    },
+    {
+      "epoch": 0.35233421416886873,
+      "grad_norm": 0.07330293208360672,
+      "learning_rate": 6.832627118644068e-05,
+      "loss": 0.5974,
+      "step": 350
+    },
+    {
+      "epoch": 0.35736756008556686,
+      "grad_norm": 0.0678195133805275,
+      "learning_rate": 6.779661016949152e-05,
+      "loss": 0.6298,
+      "step": 355
+    },
+    {
+      "epoch": 0.362400906002265,
+      "grad_norm": 0.0872490331530571,
+      "learning_rate": 6.726694915254239e-05,
+      "loss": 0.6035,
+      "step": 360
+    },
+    {
+      "epoch": 0.3674342519189631,
+      "grad_norm": 0.0971163958311081,
+      "learning_rate": 6.673728813559322e-05,
+      "loss": 0.5686,
+      "step": 365
+    },
+    {
+      "epoch": 0.37246759783566125,
+      "grad_norm": 0.07296817749738693,
+      "learning_rate": 6.620762711864407e-05,
+      "loss": 0.6126,
+      "step": 370
+    },
+    {
+      "epoch": 0.3775009437523594,
+      "grad_norm": 0.08710750937461853,
+      "learning_rate": 6.567796610169492e-05,
+      "loss": 0.6106,
+      "step": 375
+    },
+    {
+      "epoch": 0.3825342896690575,
+      "grad_norm": 0.07178603857755661,
+      "learning_rate": 6.514830508474576e-05,
+      "loss": 0.5929,
+      "step": 380
+    },
+    {
+      "epoch": 0.38756763558575563,
+      "grad_norm": 0.06608158349990845,
+      "learning_rate": 6.461864406779662e-05,
+      "loss": 0.5926,
+      "step": 385
+    },
+    {
+      "epoch": 0.39260098150245376,
+      "grad_norm": 0.07164452224969864,
+      "learning_rate": 6.408898305084746e-05,
+      "loss": 0.6491,
+      "step": 390
+    },
+    {
+      "epoch": 0.3976343274191519,
+      "grad_norm": 0.08054110407829285,
+      "learning_rate": 6.35593220338983e-05,
+      "loss": 0.587,
+      "step": 395
+    },
+    {
+      "epoch": 0.40266767333585,
+      "grad_norm": 0.07114928960800171,
+      "learning_rate": 6.302966101694915e-05,
+      "loss": 0.6163,
+      "step": 400
+    },
+    {
+      "epoch": 0.40770101925254815,
+      "grad_norm": 0.07621515542268753,
+      "learning_rate": 6.25e-05,
+      "loss": 0.6253,
+      "step": 405
+    },
+    {
+      "epoch": 0.4127343651692463,
+      "grad_norm": 0.07434070110321045,
+      "learning_rate": 6.197033898305085e-05,
+      "loss": 0.6034,
+      "step": 410
+    },
+    {
+      "epoch": 0.4177677110859444,
+      "grad_norm": 0.09286583214998245,
+      "learning_rate": 6.14406779661017e-05,
+      "loss": 0.5656,
+      "step": 415
+    },
+    {
+      "epoch": 0.42280105700264253,
+      "grad_norm": 0.08313080668449402,
+      "learning_rate": 6.0911016949152544e-05,
+      "loss": 0.583,
+      "step": 420
+    },
+    {
+      "epoch": 0.4278344029193406,
+      "grad_norm": 0.07711467146873474,
+      "learning_rate": 6.0381355932203385e-05,
+      "loss": 0.6026,
+      "step": 425
+    },
+    {
+      "epoch": 0.43286774883603873,
+      "grad_norm": 0.07896324247121811,
+      "learning_rate": 5.985169491525424e-05,
+      "loss": 0.5873,
+      "step": 430
+    },
+    {
+      "epoch": 0.43790109475273686,
+      "grad_norm": 0.06385936588048935,
+      "learning_rate": 5.932203389830509e-05,
+      "loss": 0.5904,
+      "step": 435
+    },
+    {
+      "epoch": 0.442934440669435,
+      "grad_norm": 0.09176833182573318,
+      "learning_rate": 5.879237288135594e-05,
+      "loss": 0.6538,
+      "step": 440
+    },
+    {
+      "epoch": 0.4479677865861331,
+      "grad_norm": 0.0740937814116478,
+      "learning_rate": 5.826271186440678e-05,
+      "loss": 0.5782,
+      "step": 445
+    },
+    {
+      "epoch": 0.45300113250283125,
+      "grad_norm": 0.07935595512390137,
+      "learning_rate": 5.7733050847457624e-05,
+      "loss": 0.5867,
+      "step": 450
+    },
+    {
+      "epoch": 0.4580344784195294,
+      "grad_norm": 0.09611266851425171,
+      "learning_rate": 5.720338983050848e-05,
+      "loss": 0.6471,
+      "step": 455
+    },
+    {
+      "epoch": 0.4630678243362275,
+      "grad_norm": 0.07880236953496933,
+      "learning_rate": 5.6673728813559326e-05,
+      "loss": 0.5638,
+      "step": 460
+    },
+    {
+      "epoch": 0.46810117025292564,
+      "grad_norm": 0.0920519232749939,
+      "learning_rate": 5.614406779661017e-05,
+      "loss": 0.5652,
+      "step": 465
+    },
+    {
+      "epoch": 0.47313451616962376,
+      "grad_norm": 0.08813466876745224,
+      "learning_rate": 5.561440677966102e-05,
+      "loss": 0.6037,
+      "step": 470
+    },
+    {
+      "epoch": 0.4781678620863219,
+      "grad_norm": 0.08962405472993851,
+      "learning_rate": 5.508474576271186e-05,
+      "loss": 0.6192,
+      "step": 475
+    },
+    {
+      "epoch": 0.48320120800302,
+      "grad_norm": 0.06809104233980179,
+      "learning_rate": 5.4555084745762716e-05,
+      "loss": 0.5393,
+      "step": 480
+    },
+    {
+      "epoch": 0.48823455391971815,
+      "grad_norm": 0.06712453067302704,
+      "learning_rate": 5.4025423728813564e-05,
+      "loss": 0.5962,
+      "step": 485
+    },
+    {
+      "epoch": 0.4932678998364163,
+      "grad_norm": 0.08403871953487396,
+      "learning_rate": 5.3495762711864405e-05,
+      "loss": 0.6284,
+      "step": 490
+    },
+    {
+      "epoch": 0.4983012457531144,
+      "grad_norm": 0.06908678263425827,
+      "learning_rate": 5.296610169491526e-05,
+      "loss": 0.6207,
+      "step": 495
+    },
+    {
+      "epoch": 0.5033345916698125,
+      "grad_norm": 0.06774463504552841,
+      "learning_rate": 5.24364406779661e-05,
+      "loss": 0.5825,
+      "step": 500
+    },
+    {
+      "epoch": 0.5083679375865107,
+      "grad_norm": 0.07778752595186234,
+      "learning_rate": 5.190677966101695e-05,
+      "loss": 0.6044,
+      "step": 505
+    },
+    {
+      "epoch": 0.5134012835032088,
+      "grad_norm": 0.08649144321680069,
+      "learning_rate": 5.13771186440678e-05,
+      "loss": 0.6504,
+      "step": 510
+    },
+    {
+      "epoch": 0.5184346294199069,
+      "grad_norm": 0.0770857185125351,
+      "learning_rate": 5.0847457627118643e-05,
+      "loss": 0.5725,
+      "step": 515
+    },
+    {
+      "epoch": 0.523467975336605,
+      "grad_norm": 0.0756506696343422,
+      "learning_rate": 5.03177966101695e-05,
+      "loss": 0.6352,
+      "step": 520
+    },
+    {
+      "epoch": 0.5285013212533032,
+      "grad_norm": 0.0720185860991478,
+      "learning_rate": 4.978813559322034e-05,
+      "loss": 0.6112,
+      "step": 525
+    },
+    {
+      "epoch": 0.5335346671700013,
+      "grad_norm": 0.08800795674324036,
+      "learning_rate": 4.925847457627119e-05,
+      "loss": 0.5996,
+      "step": 530
+    },
+    {
+      "epoch": 0.5385680130866994,
+      "grad_norm": 0.06613479554653168,
+      "learning_rate": 4.8728813559322034e-05,
+      "loss": 0.612,
+      "step": 535
+    },
+    {
+      "epoch": 0.5436013590033975,
+      "grad_norm": 0.07253889739513397,
+      "learning_rate": 4.819915254237288e-05,
+      "loss": 0.6257,
+      "step": 540
+    },
+    {
+      "epoch": 0.5486347049200956,
+      "grad_norm": 0.0724099799990654,
+      "learning_rate": 4.766949152542373e-05,
+      "loss": 0.5961,
+      "step": 545
+    },
+    {
+      "epoch": 0.5536680508367937,
+      "grad_norm": 0.08074811100959778,
+      "learning_rate": 4.7139830508474584e-05,
+      "loss": 0.5899,
+      "step": 550
+    },
+    {
+      "epoch": 0.5587013967534918,
+      "grad_norm": 0.07143207639455795,
+      "learning_rate": 4.6610169491525425e-05,
+      "loss": 0.6223,
+      "step": 555
+    },
+    {
+      "epoch": 0.56373474267019,
+      "grad_norm": 0.07310809195041656,
+      "learning_rate": 4.608050847457627e-05,
+      "loss": 0.6294,
+      "step": 560
+    },
+    {
+      "epoch": 0.5687680885868881,
+      "grad_norm": 0.07864546775817871,
+      "learning_rate": 4.555084745762712e-05,
+      "loss": 0.5814,
+      "step": 565
+    },
+    {
+      "epoch": 0.5738014345035862,
+      "grad_norm": 0.07621514797210693,
+      "learning_rate": 4.502118644067797e-05,
+      "loss": 0.6135,
+      "step": 570
+    },
+    {
+      "epoch": 0.5788347804202844,
+      "grad_norm": 0.07107630372047424,
+      "learning_rate": 4.4491525423728816e-05,
+      "loss": 0.6159,
+      "step": 575
+    },
+    {
+      "epoch": 0.5838681263369825,
+      "grad_norm": 0.06919348239898682,
+      "learning_rate": 4.396186440677966e-05,
+      "loss": 0.6059,
+      "step": 580
+    },
+    {
+      "epoch": 0.5889014722536806,
+      "grad_norm": 0.0646335557103157,
+      "learning_rate": 4.343220338983051e-05,
+      "loss": 0.6195,
+      "step": 585
+    },
+    {
+      "epoch": 0.5939348181703787,
+      "grad_norm": 0.07596474885940552,
+      "learning_rate": 4.290254237288136e-05,
+      "loss": 0.6258,
+      "step": 590
+    },
+    {
+      "epoch": 0.5989681640870769,
+      "grad_norm": 0.06612690538167953,
+      "learning_rate": 4.2372881355932206e-05,
+      "loss": 0.6078,
+      "step": 595
+    },
+    {
+      "epoch": 0.604001510003775,
+      "grad_norm": 0.06783054769039154,
+      "learning_rate": 4.1843220338983054e-05,
+      "loss": 0.5683,
+      "step": 600
+    },
+    {
+      "epoch": 0.6090348559204731,
+      "grad_norm": 0.08887500315904617,
+      "learning_rate": 4.13135593220339e-05,
+      "loss": 0.6774,
+      "step": 605
+    },
+    {
+      "epoch": 0.6140682018371713,
+      "grad_norm": 0.07525520026683807,
+      "learning_rate": 4.078389830508475e-05,
+      "loss": 0.6314,
+      "step": 610
+    },
+    {
+      "epoch": 0.6191015477538694,
+      "grad_norm": 0.23945115506649017,
+      "learning_rate": 4.025423728813559e-05,
+      "loss": 0.5712,
+      "step": 615
+    },
+    {
+      "epoch": 0.6241348936705675,
+      "grad_norm": 0.07753656059503555,
+      "learning_rate": 3.9724576271186445e-05,
+      "loss": 0.6107,
+      "step": 620
+    },
+    {
+      "epoch": 0.6291682395872656,
+      "grad_norm": 0.08550898730754852,
+      "learning_rate": 3.919491525423729e-05,
+      "loss": 0.6209,
+      "step": 625
+    },
+    {
+      "epoch": 0.6342015855039638,
+      "grad_norm": 0.09466369450092316,
+      "learning_rate": 3.866525423728814e-05,
+      "loss": 0.594,
+      "step": 630
+    },
+    {
+      "epoch": 0.6392349314206619,
+      "grad_norm": 0.06254788488149643,
+      "learning_rate": 3.813559322033898e-05,
+      "loss": 0.5985,
+      "step": 635
+    },
+    {
+      "epoch": 0.64426827733736,
+      "grad_norm": 0.06980835646390915,
+      "learning_rate": 3.7605932203389835e-05,
+      "loss": 0.6518,
+      "step": 640
+    },
+    {
+      "epoch": 0.6493016232540582,
+      "grad_norm": 0.07291670143604279,
+      "learning_rate": 3.707627118644068e-05,
+      "loss": 0.6192,
+      "step": 645
+    },
+    {
+      "epoch": 0.6543349691707563,
+      "grad_norm": 0.07008092850446701,
+      "learning_rate": 3.654661016949153e-05,
+      "loss": 0.6011,
+      "step": 650
+    },
+    {
+      "epoch": 0.6593683150874544,
+      "grad_norm": 0.07765305787324905,
+      "learning_rate": 3.601694915254237e-05,
+      "loss": 0.6317,
+      "step": 655
+    },
+    {
+      "epoch": 0.6644016610041525,
+      "grad_norm": 0.07724044471979141,
+      "learning_rate": 3.548728813559322e-05,
+      "loss": 0.6362,
+      "step": 660
+    },
+    {
+      "epoch": 0.6694350069208507,
+      "grad_norm": 0.08179567009210587,
+      "learning_rate": 3.4957627118644074e-05,
+      "loss": 0.5665,
+      "step": 665
+    },
+    {
+      "epoch": 0.6744683528375488,
+      "grad_norm": 0.0660698190331459,
+      "learning_rate": 3.442796610169492e-05,
+      "loss": 0.5832,
+      "step": 670
+    },
+    {
+      "epoch": 0.6795016987542469,
+      "grad_norm": 0.07155987620353699,
+      "learning_rate": 3.389830508474576e-05,
+      "loss": 0.639,
+      "step": 675
+    },
+    {
+      "epoch": 0.6845350446709451,
+      "grad_norm": 0.07769577950239182,
+      "learning_rate": 3.336864406779661e-05,
+      "loss": 0.5643,
+      "step": 680
+    },
+    {
+      "epoch": 0.6895683905876432,
+      "grad_norm": 0.06444139033555984,
+      "learning_rate": 3.283898305084746e-05,
+      "loss": 0.5761,
+      "step": 685
+    },
+    {
+      "epoch": 0.6946017365043413,
+      "grad_norm": 0.08081527799367905,
+      "learning_rate": 3.230932203389831e-05,
+      "loss": 0.6574,
+      "step": 690
+    },
+    {
+      "epoch": 0.6996350824210393,
+      "grad_norm": 0.062211159616708755,
+      "learning_rate": 3.177966101694915e-05,
+      "loss": 0.6195,
+      "step": 695
+    },
+    {
+      "epoch": 0.7046684283377375,
+      "grad_norm": 0.06285897642374039,
+      "learning_rate": 3.125e-05,
+      "loss": 0.6026,
+      "step": 700
+    },
+    {
+      "epoch": 0.7097017742544356,
+      "grad_norm": 0.07187480479478836,
+      "learning_rate": 3.072033898305085e-05,
+      "loss": 0.6168,
+      "step": 705
+    },
+    {
+      "epoch": 0.7147351201711337,
+      "grad_norm": 0.08164852857589722,
+      "learning_rate": 3.0190677966101693e-05,
+      "loss": 0.5604,
+      "step": 710
+    },
+    {
+      "epoch": 0.7197684660878318,
+      "grad_norm": 0.09309656918048859,
+      "learning_rate": 2.9661016949152544e-05,
+      "loss": 0.5902,
+      "step": 715
+    },
+    {
+      "epoch": 0.72480181200453,
+      "grad_norm": 0.07180369645357132,
+      "learning_rate": 2.913135593220339e-05,
+      "loss": 0.6083,
+      "step": 720
+    },
+    {
+      "epoch": 0.7298351579212281,
+      "grad_norm": 0.06690961867570877,
+      "learning_rate": 2.860169491525424e-05,
+      "loss": 0.6064,
+      "step": 725
+    },
+    {
+      "epoch": 0.7348685038379262,
+      "grad_norm": 0.07759429514408112,
+      "learning_rate": 2.8072033898305083e-05,
+      "loss": 0.6078,
+      "step": 730
+    },
+    {
+      "epoch": 0.7399018497546244,
+      "grad_norm": 0.09760767966508865,
+      "learning_rate": 2.754237288135593e-05,
+      "loss": 0.5839,
+      "step": 735
+    },
+    {
+      "epoch": 0.7449351956713225,
+      "grad_norm": 0.06190226599574089,
+      "learning_rate": 2.7012711864406782e-05,
+      "loss": 0.6424,
+      "step": 740
+    },
+    {
+      "epoch": 0.7499685415880206,
+      "grad_norm": 0.07761067897081375,
+      "learning_rate": 2.648305084745763e-05,
+      "loss": 0.6109,
+      "step": 745
+    },
+    {
+      "epoch": 0.7550018875047187,
+      "grad_norm": 0.07072454690933228,
+      "learning_rate": 2.5953389830508474e-05,
+      "loss": 0.5638,
+      "step": 750
+    },
+    {
+      "epoch": 0.7600352334214169,
+      "grad_norm": 0.07209552079439163,
+      "learning_rate": 2.5423728813559322e-05,
+      "loss": 0.6586,
+      "step": 755
+    },
+    {
+      "epoch": 0.765068579338115,
+      "grad_norm": 0.07083916664123535,
+      "learning_rate": 2.489406779661017e-05,
+      "loss": 0.5989,
+      "step": 760
+    },
+    {
+      "epoch": 0.7701019252548131,
+      "grad_norm": 0.07853751629590988,
+      "learning_rate": 2.4364406779661017e-05,
+      "loss": 0.5734,
+      "step": 765
+    },
+    {
+      "epoch": 0.7751352711715113,
+      "grad_norm": 0.06931478530168533,
+      "learning_rate": 2.3834745762711865e-05,
+      "loss": 0.6101,
+      "step": 770
+    },
+    {
+      "epoch": 0.7801686170882094,
+      "grad_norm": 0.07923394441604614,
+      "learning_rate": 2.3305084745762712e-05,
+      "loss": 0.6291,
+      "step": 775
+    },
+    {
+      "epoch": 0.7852019630049075,
+      "grad_norm": 0.08270952105522156,
+      "learning_rate": 2.277542372881356e-05,
+      "loss": 0.5892,
+      "step": 780
+    },
+    {
+      "epoch": 0.7902353089216057,
+      "grad_norm": 0.06102617457509041,
+      "learning_rate": 2.2245762711864408e-05,
+      "loss": 0.618,
+      "step": 785
+    },
+    {
+      "epoch": 0.7952686548383038,
+      "grad_norm": 0.09832928329706192,
+      "learning_rate": 2.1716101694915255e-05,
+      "loss": 0.6427,
+      "step": 790
+    },
+    {
+      "epoch": 0.8003020007550019,
+      "grad_norm": 0.06228268891572952,
+      "learning_rate": 2.1186440677966103e-05,
+      "loss": 0.6312,
+      "step": 795
+    },
+    {
+      "epoch": 0.8053353466717,
+      "grad_norm": 0.071562260389328,
+      "learning_rate": 2.065677966101695e-05,
+      "loss": 0.5636,
+      "step": 800
+    },
+    {
+      "epoch": 0.8103686925883982,
+      "grad_norm": 0.1984732449054718,
+      "learning_rate": 2.0127118644067795e-05,
+      "loss": 0.6223,
+      "step": 805
+    },
+    {
+      "epoch": 0.8154020385050963,
+      "grad_norm": 0.07434993982315063,
+      "learning_rate": 1.9597457627118646e-05,
+      "loss": 0.6222,
+      "step": 810
+    },
+    {
+      "epoch": 0.8204353844217944,
+      "grad_norm": 0.09027515351772308,
+      "learning_rate": 1.906779661016949e-05,
+      "loss": 0.5831,
+      "step": 815
+    },
+    {
+      "epoch": 0.8254687303384926,
+      "grad_norm": 0.07995504140853882,
+      "learning_rate": 1.853813559322034e-05,
+      "loss": 0.625,
+      "step": 820
+    },
+    {
+      "epoch": 0.8305020762551907,
+      "grad_norm": 0.06889399886131287,
+      "learning_rate": 1.8008474576271186e-05,
+      "loss": 0.639,
+      "step": 825
+    },
+    {
+      "epoch": 0.8355354221718888,
+      "grad_norm": 0.08385035395622253,
+      "learning_rate": 1.7478813559322037e-05,
+      "loss": 0.5871,
+      "step": 830
+    },
+    {
+      "epoch": 0.8405687680885869,
+      "grad_norm": 0.05659618228673935,
+      "learning_rate": 1.694915254237288e-05,
+      "loss": 0.6023,
+      "step": 835
+    },
+    {
+      "epoch": 0.8456021140052851,
+      "grad_norm": 0.07665451616048813,
+      "learning_rate": 1.641949152542373e-05,
+      "loss": 0.6434,
+      "step": 840
+    },
+    {
+      "epoch": 0.8506354599219831,
+      "grad_norm": 0.16329655051231384,
+      "learning_rate": 1.5889830508474576e-05,
+      "loss": 0.6251,
+      "step": 845
+    },
+    {
+      "epoch": 0.8556688058386812,
+      "grad_norm": 0.06561240553855896,
+      "learning_rate": 1.5360169491525424e-05,
+      "loss": 0.5535,
+      "step": 850
+    },
+    {
+      "epoch": 0.8607021517553793,
+      "grad_norm": 0.08012928068637848,
+      "learning_rate": 1.4830508474576272e-05,
+      "loss": 0.6364,
+      "step": 855
+    },
+    {
+      "epoch": 0.8657354976720775,
+      "grad_norm": 0.07773120701313019,
+      "learning_rate": 1.430084745762712e-05,
+      "loss": 0.6329,
+      "step": 860
+    },
+    {
+      "epoch": 0.8707688435887756,
+      "grad_norm": 0.08004656434059143,
+      "learning_rate": 1.3771186440677965e-05,
+      "loss": 0.5577,
+      "step": 865
+    },
+    {
+      "epoch": 0.8758021895054737,
+      "grad_norm": 0.0730496346950531,
+      "learning_rate": 1.3241525423728815e-05,
+      "loss": 0.6426,
+      "step": 870
+    },
+    {
+      "epoch": 0.8808355354221719,
+      "grad_norm": 0.07398191839456558,
+      "learning_rate": 1.2711864406779661e-05,
+      "loss": 0.624,
+      "step": 875
+    },
+    {
+      "epoch": 0.88586888133887,
+      "grad_norm": 0.07571957260370255,
+      "learning_rate": 1.2182203389830509e-05,
+      "loss": 0.5746,
+      "step": 880
+    },
+    {
+      "epoch": 0.8909022272555681,
+      "grad_norm": 0.0590958334505558,
+      "learning_rate": 1.1652542372881356e-05,
+      "loss": 0.5958,
+      "step": 885
+    },
+    {
+      "epoch": 0.8959355731722662,
+      "grad_norm": 0.0737120732665062,
+      "learning_rate": 1.1122881355932204e-05,
+      "loss": 0.6305,
+      "step": 890
+    },
+    {
+      "epoch": 0.9009689190889644,
+      "grad_norm": 0.06610913574695587,
+      "learning_rate": 1.0593220338983052e-05,
+      "loss": 0.6023,
+      "step": 895
+    },
+    {
+      "epoch": 0.9060022650056625,
+      "grad_norm": 0.07495182752609253,
+      "learning_rate": 1.0063559322033898e-05,
+      "loss": 0.574,
+      "step": 900
+    },
+    {
+      "epoch": 0.9110356109223606,
+      "grad_norm": 0.07383362948894501,
+      "learning_rate": 9.533898305084745e-06,
+      "loss": 0.6506,
+      "step": 905
+    },
+    {
+      "epoch": 0.9160689568390588,
+      "grad_norm": 0.06826779246330261,
+      "learning_rate": 9.004237288135593e-06,
+      "loss": 0.638,
+      "step": 910
+    },
+    {
+      "epoch": 0.9211023027557569,
+      "grad_norm": 0.08567427098751068,
+      "learning_rate": 8.47457627118644e-06,
+      "loss": 0.562,
+      "step": 915
+    },
+    {
+      "epoch": 0.926135648672455,
+      "grad_norm": 0.06812965124845505,
+      "learning_rate": 7.944915254237288e-06,
+      "loss": 0.62,
+      "step": 920
+    },
+    {
+      "epoch": 0.9311689945891531,
+      "grad_norm": 0.07216673344373703,
+      "learning_rate": 7.415254237288136e-06,
+      "loss": 0.6251,
+      "step": 925
+    },
+    {
+      "epoch": 0.9362023405058513,
+      "grad_norm": 0.06816910952329636,
+      "learning_rate": 6.885593220338983e-06,
+      "loss": 0.5744,
+      "step": 930
+    },
+    {
+      "epoch": 0.9412356864225494,
+      "grad_norm": 0.06639246642589569,
+      "learning_rate": 6.3559322033898304e-06,
+      "loss": 0.5761,
+      "step": 935
+    },
+    {
+      "epoch": 0.9462690323392475,
+      "grad_norm": 0.06687802076339722,
+      "learning_rate": 5.826271186440678e-06,
+      "loss": 0.6376,
+      "step": 940
+    },
+    {
+      "epoch": 0.9513023782559457,
+      "grad_norm": 0.07500467449426651,
+      "learning_rate": 5.296610169491526e-06,
+      "loss": 0.592,
+      "step": 945
+    },
+    {
+      "epoch": 0.9563357241726438,
+      "grad_norm": 0.06670086830854416,
+      "learning_rate": 4.766949152542373e-06,
+      "loss": 0.5681,
+      "step": 950
+    },
+    {
+      "epoch": 0.9613690700893419,
+      "grad_norm": 0.08027223497629166,
+      "learning_rate": 4.23728813559322e-06,
+      "loss": 0.6497,
+      "step": 955
+    },
+    {
+      "epoch": 0.96640241600604,
+      "grad_norm": 0.06293675303459167,
+      "learning_rate": 3.707627118644068e-06,
+      "loss": 0.6116,
+      "step": 960
+    },
+    {
+      "epoch": 0.9714357619227382,
+      "grad_norm": 0.08093003183603287,
+      "learning_rate": 3.1779661016949152e-06,
+      "loss": 0.5685,
+      "step": 965
+    },
+    {
+      "epoch": 0.9764691078394363,
+      "grad_norm": 0.08290575444698334,
+      "learning_rate": 2.648305084745763e-06,
+      "loss": 0.6192,
+      "step": 970
+    },
+    {
+      "epoch": 0.9815024537561344,
+      "grad_norm": 0.07009682059288025,
+      "learning_rate": 2.11864406779661e-06,
+      "loss": 0.6249,
+      "step": 975
+    },
+    {
+      "epoch": 0.9865357996728326,
+      "grad_norm": 0.08704908192157745,
+      "learning_rate": 1.5889830508474576e-06,
+      "loss": 0.5714,
+      "step": 980
+    },
+    {
+      "epoch": 0.9915691455895307,
+      "grad_norm": 0.08047839254140854,
+      "learning_rate": 1.059322033898305e-06,
+      "loss": 0.5998,
+      "step": 985
+    },
+    {
+      "epoch": 0.9966024915062288,
+      "grad_norm": 0.07949598878622055,
+      "learning_rate": 5.296610169491525e-07,
+      "loss": 0.6314,
+      "step": 990
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 994,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.1407902519689898e+20,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

Qwen3-30B-A3B-Instruct-2507-sft-fusang-swa=2k_sink=100_fadec/checkpoint-870/README.md

@@ -0,0 +1,207 @@
+---
+base_model: /share/models/Qwen3-30B-A3B-Instruct-2507
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:/share/models/Qwen3-30B-A3B-Instruct-2507
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.17.1

Qwen3-30B-A3B-Thinking-2507-sft-fusang-swa=2k_sink=0/checkpoint-862/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

Qwen3-30B-A3B-Thinking-2507-sft-fusang-swa=2k_sink=0/checkpoint-862/chat_template.jinja

@@ -0,0 +1,86 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n<think>\n' }}
+{%- endif %}

Qwen3-30B-A3B-Thinking-2507-sft-fusang-swa=2k_sink=0_falayer=25_from0_step2/checkpoint-862/README.md

@@ -0,0 +1,207 @@
+---
+base_model: /share/models/Qwen3-30B-A3B-Thinking-2507
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:/share/models/Qwen3-30B-A3B-Thinking-2507
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.17.1

Qwen3-30B-A3B-Thinking-2507-sft-fusang-swa=2k_sink=0_falayer=25_from0_step2/checkpoint-862/adapter_config.json

@@ -0,0 +1,38 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "/share/models/Qwen3-30B-A3B-Thinking-2507",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "k_proj",
+    "v_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": true
+}

Qwen3-30B-A3B-Thinking-2507-sft-fusang-swa=2k_sink=0_falayer=25_from0_step2/checkpoint-862/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

Qwen3-30B-A3B-Thinking-2507-sft-fusang-swa=2k_sink=0_falayer=25_from0_step2/checkpoint-862/chat_template.jinja

@@ -0,0 +1,86 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n<think>\n' }}
+{%- endif %}

Qwen3-30B-A3B-Thinking-2507-sft-fusang-swa=2k_sink=0_falayer=25_from0_step2/checkpoint-862/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

Qwen3-30B-A3B-Thinking-2507-sft-fusang-swa=2k_sink=0_falayer=25_from0_step2/checkpoint-862/tokenizer_config.json

@@ -0,0 +1,239 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 1010000,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

Qwen3-30B-A3B-Thinking-2507-sft-fusang-swa=2k_sink=100/checkpoint-862/README.md

@@ -0,0 +1,207 @@
+---
+base_model: /share/models/Qwen3-30B-A3B-Thinking-2507
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:/share/models/Qwen3-30B-A3B-Thinking-2507
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.17.1

Qwen3-30B-A3B-Thinking-2507-sft-fusang-swa=2k_sink=100/checkpoint-862/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

Qwen3-30B-A3B-Thinking-2507-sft-fusang-swa=2k_sink=100/checkpoint-862/tokenizer_config.json

@@ -0,0 +1,239 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 1010000,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

Qwen3-30B-A3B-Thinking-2507-sft-fusang-swa=2k_sink=100/checkpoint-862/trainer_state.json

@@ -0,0 +1,1238 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 862,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.005803830528148578,
+      "grad_norm": 0.07806325703859329,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.4424,
+      "step": 5
+    },
+    {
+      "epoch": 0.011607661056297156,
+      "grad_norm": 0.06576955318450928,
+      "learning_rate": 2.0454545454545457e-05,
+      "loss": 0.4676,
+      "step": 10
+    },
+    {
+      "epoch": 0.017411491584445733,
+      "grad_norm": 0.058958571404218674,
+      "learning_rate": 3.181818181818182e-05,
+      "loss": 0.4786,
+      "step": 15
+    },
+    {
+      "epoch": 0.023215322112594312,
+      "grad_norm": 0.05414086580276489,
+      "learning_rate": 4.318181818181819e-05,
+      "loss": 0.4724,
+      "step": 20
+    },
+    {
+      "epoch": 0.02901915264074289,
+      "grad_norm": 0.07120595872402191,
+      "learning_rate": 5.4545454545454546e-05,
+      "loss": 0.4654,
+      "step": 25
+    },
+    {
+      "epoch": 0.034822983168891465,
+      "grad_norm": 0.05655818060040474,
+      "learning_rate": 6.59090909090909e-05,
+      "loss": 0.4489,
+      "step": 30
+    },
+    {
+      "epoch": 0.04062681369704005,
+      "grad_norm": 0.04869989678263664,
+      "learning_rate": 7.727272727272727e-05,
+      "loss": 0.4552,
+      "step": 35
+    },
+    {
+      "epoch": 0.046430644225188625,
+      "grad_norm": 0.04493036866188049,
+      "learning_rate": 8.863636363636364e-05,
+      "loss": 0.4674,
+      "step": 40
+    },
+    {
+      "epoch": 0.0522344747533372,
+      "grad_norm": 0.05274348706007004,
+      "learning_rate": 0.0001,
+      "loss": 0.4678,
+      "step": 45
+    },
+    {
+      "epoch": 0.05803830528148578,
+      "grad_norm": 0.057375144213438034,
+      "learning_rate": 9.938875305623473e-05,
+      "loss": 0.4739,
+      "step": 50
+    },
+    {
+      "epoch": 0.06384213580963435,
+      "grad_norm": 0.060564618557691574,
+      "learning_rate": 9.877750611246945e-05,
+      "loss": 0.4673,
+      "step": 55
+    },
+    {
+      "epoch": 0.06964596633778293,
+      "grad_norm": 0.04939981922507286,
+      "learning_rate": 9.816625916870417e-05,
+      "loss": 0.4679,
+      "step": 60
+    },
+    {
+      "epoch": 0.07544979686593152,
+      "grad_norm": 0.05144249647855759,
+      "learning_rate": 9.755501222493889e-05,
+      "loss": 0.4555,
+      "step": 65
+    },
+    {
+      "epoch": 0.0812536273940801,
+      "grad_norm": 0.05333191528916359,
+      "learning_rate": 9.69437652811736e-05,
+      "loss": 0.4515,
+      "step": 70
+    },
+    {
+      "epoch": 0.08705745792222867,
+      "grad_norm": 0.05306636914610863,
+      "learning_rate": 9.633251833740831e-05,
+      "loss": 0.4903,
+      "step": 75
+    },
+    {
+      "epoch": 0.09286128845037725,
+      "grad_norm": 0.0424158088862896,
+      "learning_rate": 9.572127139364305e-05,
+      "loss": 0.442,
+      "step": 80
+    },
+    {
+      "epoch": 0.09866511897852583,
+      "grad_norm": 0.04384077340364456,
+      "learning_rate": 9.511002444987775e-05,
+      "loss": 0.4478,
+      "step": 85
+    },
+    {
+      "epoch": 0.1044689495066744,
+      "grad_norm": 0.0483214296400547,
+      "learning_rate": 9.449877750611247e-05,
+      "loss": 0.4532,
+      "step": 90
+    },
+    {
+      "epoch": 0.11027278003482298,
+      "grad_norm": 0.06844743341207504,
+      "learning_rate": 9.38875305623472e-05,
+      "loss": 0.4533,
+      "step": 95
+    },
+    {
+      "epoch": 0.11607661056297155,
+      "grad_norm": 0.05522039905190468,
+      "learning_rate": 9.327628361858191e-05,
+      "loss": 0.4663,
+      "step": 100
+    },
+    {
+      "epoch": 0.12188044109112015,
+      "grad_norm": 0.04972909390926361,
+      "learning_rate": 9.266503667481663e-05,
+      "loss": 0.4512,
+      "step": 105
+    },
+    {
+      "epoch": 0.1276842716192687,
+      "grad_norm": 0.05091628432273865,
+      "learning_rate": 9.205378973105135e-05,
+      "loss": 0.4533,
+      "step": 110
+    },
+    {
+      "epoch": 0.1334881021474173,
+      "grad_norm": 0.06282547116279602,
+      "learning_rate": 9.144254278728606e-05,
+      "loss": 0.4668,
+      "step": 115
+    },
+    {
+      "epoch": 0.13929193267556586,
+      "grad_norm": 0.0661332905292511,
+      "learning_rate": 9.08312958435208e-05,
+      "loss": 0.471,
+      "step": 120
+    },
+    {
+      "epoch": 0.14509576320371445,
+      "grad_norm": 0.04994620010256767,
+      "learning_rate": 9.02200488997555e-05,
+      "loss": 0.4463,
+      "step": 125
+    },
+    {
+      "epoch": 0.15089959373186304,
+      "grad_norm": 0.04718010500073433,
+      "learning_rate": 8.960880195599024e-05,
+      "loss": 0.4602,
+      "step": 130
+    },
+    {
+      "epoch": 0.1567034242600116,
+      "grad_norm": 0.03880172222852707,
+      "learning_rate": 8.899755501222494e-05,
+      "loss": 0.4682,
+      "step": 135
+    },
+    {
+      "epoch": 0.1625072547881602,
+      "grad_norm": 0.04132480546832085,
+      "learning_rate": 8.838630806845966e-05,
+      "loss": 0.481,
+      "step": 140
+    },
+    {
+      "epoch": 0.16831108531630876,
+      "grad_norm": 0.037531778216362,
+      "learning_rate": 8.777506112469438e-05,
+      "loss": 0.4741,
+      "step": 145
+    },
+    {
+      "epoch": 0.17411491584445735,
+      "grad_norm": 0.051236528903245926,
+      "learning_rate": 8.71638141809291e-05,
+      "loss": 0.4628,
+      "step": 150
+    },
+    {
+      "epoch": 0.1799187463726059,
+      "grad_norm": 0.041979845613241196,
+      "learning_rate": 8.655256723716382e-05,
+      "loss": 0.4387,
+      "step": 155
+    },
+    {
+      "epoch": 0.1857225769007545,
+      "grad_norm": 0.08546081930398941,
+      "learning_rate": 8.594132029339854e-05,
+      "loss": 0.46,
+      "step": 160
+    },
+    {
+      "epoch": 0.1915264074289031,
+      "grad_norm": 0.042775433510541916,
+      "learning_rate": 8.533007334963325e-05,
+      "loss": 0.4603,
+      "step": 165
+    },
+    {
+      "epoch": 0.19733023795705165,
+      "grad_norm": 0.03881630674004555,
+      "learning_rate": 8.471882640586798e-05,
+      "loss": 0.4604,
+      "step": 170
+    },
+    {
+      "epoch": 0.20313406848520024,
+      "grad_norm": 0.04963105544447899,
+      "learning_rate": 8.410757946210269e-05,
+      "loss": 0.4647,
+      "step": 175
+    },
+    {
+      "epoch": 0.2089378990133488,
+      "grad_norm": 0.04456959664821625,
+      "learning_rate": 8.349633251833741e-05,
+      "loss": 0.4649,
+      "step": 180
+    },
+    {
+      "epoch": 0.2147417295414974,
+      "grad_norm": 0.047044992446899414,
+      "learning_rate": 8.288508557457213e-05,
+      "loss": 0.4649,
+      "step": 185
+    },
+    {
+      "epoch": 0.22054556006964596,
+      "grad_norm": 0.034605883061885834,
+      "learning_rate": 8.227383863080685e-05,
+      "loss": 0.4587,
+      "step": 190
+    },
+    {
+      "epoch": 0.22634939059779455,
+      "grad_norm": 0.0324706956744194,
+      "learning_rate": 8.166259168704157e-05,
+      "loss": 0.4362,
+      "step": 195
+    },
+    {
+      "epoch": 0.2321532211259431,
+      "grad_norm": 0.04107559472322464,
+      "learning_rate": 8.105134474327629e-05,
+      "loss": 0.4535,
+      "step": 200
+    },
+    {
+      "epoch": 0.2379570516540917,
+      "grad_norm": 0.04205545783042908,
+      "learning_rate": 8.044009779951101e-05,
+      "loss": 0.448,
+      "step": 205
+    },
+    {
+      "epoch": 0.2437608821822403,
+      "grad_norm": 0.038843851536512375,
+      "learning_rate": 7.982885085574573e-05,
+      "loss": 0.4842,
+      "step": 210
+    },
+    {
+      "epoch": 0.24956471271038885,
+      "grad_norm": 0.09389880299568176,
+      "learning_rate": 7.921760391198044e-05,
+      "loss": 0.4561,
+      "step": 215
+    },
+    {
+      "epoch": 0.2553685432385374,
+      "grad_norm": 0.03246712312102318,
+      "learning_rate": 7.860635696821517e-05,
+      "loss": 0.4429,
+      "step": 220
+    },
+    {
+      "epoch": 0.261172373766686,
+      "grad_norm": 0.04052925854921341,
+      "learning_rate": 7.799511002444988e-05,
+      "loss": 0.4637,
+      "step": 225
+    },
+    {
+      "epoch": 0.2669762042948346,
+      "grad_norm": 0.04661140590906143,
+      "learning_rate": 7.73838630806846e-05,
+      "loss": 0.4515,
+      "step": 230
+    },
+    {
+      "epoch": 0.2727800348229832,
+      "grad_norm": 0.034557487815618515,
+      "learning_rate": 7.677261613691932e-05,
+      "loss": 0.4778,
+      "step": 235
+    },
+    {
+      "epoch": 0.2785838653511317,
+      "grad_norm": 0.04210180044174194,
+      "learning_rate": 7.616136919315404e-05,
+      "loss": 0.464,
+      "step": 240
+    },
+    {
+      "epoch": 0.2843876958792803,
+      "grad_norm": 0.034747764468193054,
+      "learning_rate": 7.555012224938876e-05,
+      "loss": 0.4536,
+      "step": 245
+    },
+    {
+      "epoch": 0.2901915264074289,
+      "grad_norm": 0.04132502153515816,
+      "learning_rate": 7.493887530562348e-05,
+      "loss": 0.4777,
+      "step": 250
+    },
+    {
+      "epoch": 0.2959953569355775,
+      "grad_norm": 0.04702381044626236,
+      "learning_rate": 7.432762836185819e-05,
+      "loss": 0.4561,
+      "step": 255
+    },
+    {
+      "epoch": 0.3017991874637261,
+      "grad_norm": 0.035738810896873474,
+      "learning_rate": 7.371638141809292e-05,
+      "loss": 0.4788,
+      "step": 260
+    },
+    {
+      "epoch": 0.3076030179918746,
+      "grad_norm": 0.06460477411746979,
+      "learning_rate": 7.310513447432763e-05,
+      "loss": 0.4452,
+      "step": 265
+    },
+    {
+      "epoch": 0.3134068485200232,
+      "grad_norm": 0.03836427628993988,
+      "learning_rate": 7.249388753056235e-05,
+      "loss": 0.4481,
+      "step": 270
+    },
+    {
+      "epoch": 0.3192106790481718,
+      "grad_norm": 0.04127519577741623,
+      "learning_rate": 7.188264058679707e-05,
+      "loss": 0.4624,
+      "step": 275
+    },
+    {
+      "epoch": 0.3250145095763204,
+      "grad_norm": 0.040822580456733704,
+      "learning_rate": 7.127139364303179e-05,
+      "loss": 0.466,
+      "step": 280
+    },
+    {
+      "epoch": 0.3308183401044689,
+      "grad_norm": 0.10596367716789246,
+      "learning_rate": 7.066014669926651e-05,
+      "loss": 0.4776,
+      "step": 285
+    },
+    {
+      "epoch": 0.3366221706326175,
+      "grad_norm": 0.04359729588031769,
+      "learning_rate": 7.004889975550123e-05,
+      "loss": 0.4589,
+      "step": 290
+    },
+    {
+      "epoch": 0.3424260011607661,
+      "grad_norm": 0.03602125123143196,
+      "learning_rate": 6.943765281173595e-05,
+      "loss": 0.4582,
+      "step": 295
+    },
+    {
+      "epoch": 0.3482298316889147,
+      "grad_norm": 0.04396088048815727,
+      "learning_rate": 6.882640586797067e-05,
+      "loss": 0.4614,
+      "step": 300
+    },
+    {
+      "epoch": 0.3540336622170633,
+      "grad_norm": 0.03952578455209732,
+      "learning_rate": 6.821515892420538e-05,
+      "loss": 0.4452,
+      "step": 305
+    },
+    {
+      "epoch": 0.3598374927452118,
+      "grad_norm": 0.04135408252477646,
+      "learning_rate": 6.760391198044011e-05,
+      "loss": 0.4745,
+      "step": 310
+    },
+    {
+      "epoch": 0.3656413232733604,
+      "grad_norm": 0.0472245030105114,
+      "learning_rate": 6.699266503667482e-05,
+      "loss": 0.458,
+      "step": 315
+    },
+    {
+      "epoch": 0.371445153801509,
+      "grad_norm": 0.032031141221523285,
+      "learning_rate": 6.638141809290954e-05,
+      "loss": 0.4482,
+      "step": 320
+    },
+    {
+      "epoch": 0.3772489843296576,
+      "grad_norm": 0.04314447194337845,
+      "learning_rate": 6.577017114914426e-05,
+      "loss": 0.4737,
+      "step": 325
+    },
+    {
+      "epoch": 0.3830528148578062,
+      "grad_norm": 0.04295211285352707,
+      "learning_rate": 6.515892420537898e-05,
+      "loss": 0.4488,
+      "step": 330
+    },
+    {
+      "epoch": 0.3888566453859547,
+      "grad_norm": 0.039687275886535645,
+      "learning_rate": 6.45476772616137e-05,
+      "loss": 0.462,
+      "step": 335
+    },
+    {
+      "epoch": 0.3946604759141033,
+      "grad_norm": 0.03278418257832527,
+      "learning_rate": 6.393643031784842e-05,
+      "loss": 0.4532,
+      "step": 340
+    },
+    {
+      "epoch": 0.4004643064422519,
+      "grad_norm": 0.0360940545797348,
+      "learning_rate": 6.332518337408312e-05,
+      "loss": 0.4637,
+      "step": 345
+    },
+    {
+      "epoch": 0.4062681369704005,
+      "grad_norm": 0.0391172431409359,
+      "learning_rate": 6.271393643031786e-05,
+      "loss": 0.4418,
+      "step": 350
+    },
+    {
+      "epoch": 0.412071967498549,
+      "grad_norm": 0.03997889161109924,
+      "learning_rate": 6.210268948655256e-05,
+      "loss": 0.4546,
+      "step": 355
+    },
+    {
+      "epoch": 0.4178757980266976,
+      "grad_norm": 0.03979405760765076,
+      "learning_rate": 6.14914425427873e-05,
+      "loss": 0.4784,
+      "step": 360
+    },
+    {
+      "epoch": 0.4236796285548462,
+      "grad_norm": 0.03965720161795616,
+      "learning_rate": 6.0880195599022005e-05,
+      "loss": 0.4527,
+      "step": 365
+    },
+    {
+      "epoch": 0.4294834590829948,
+      "grad_norm": 0.03643997758626938,
+      "learning_rate": 6.026894865525673e-05,
+      "loss": 0.4469,
+      "step": 370
+    },
+    {
+      "epoch": 0.4352872896111434,
+      "grad_norm": 0.043169762939214706,
+      "learning_rate": 5.9657701711491446e-05,
+      "loss": 0.4751,
+      "step": 375
+    },
+    {
+      "epoch": 0.4410911201392919,
+      "grad_norm": 0.03882328048348427,
+      "learning_rate": 5.9046454767726166e-05,
+      "loss": 0.4442,
+      "step": 380
+    },
+    {
+      "epoch": 0.4468949506674405,
+      "grad_norm": 0.04037743806838989,
+      "learning_rate": 5.843520782396088e-05,
+      "loss": 0.4351,
+      "step": 385
+    },
+    {
+      "epoch": 0.4526987811955891,
+      "grad_norm": 0.039136238396167755,
+      "learning_rate": 5.7823960880195606e-05,
+      "loss": 0.4604,
+      "step": 390
+    },
+    {
+      "epoch": 0.4585026117237377,
+      "grad_norm": 0.03826816380023956,
+      "learning_rate": 5.721271393643032e-05,
+      "loss": 0.4467,
+      "step": 395
+    },
+    {
+      "epoch": 0.4643064422518862,
+      "grad_norm": 0.044455818831920624,
+      "learning_rate": 5.660146699266504e-05,
+      "loss": 0.4548,
+      "step": 400
+    },
+    {
+      "epoch": 0.4701102727800348,
+      "grad_norm": 0.03864174708724022,
+      "learning_rate": 5.5990220048899754e-05,
+      "loss": 0.4523,
+      "step": 405
+    },
+    {
+      "epoch": 0.4759141033081834,
+      "grad_norm": 0.053677916526794434,
+      "learning_rate": 5.537897310513448e-05,
+      "loss": 0.4387,
+      "step": 410
+    },
+    {
+      "epoch": 0.481717933836332,
+      "grad_norm": 0.03247923031449318,
+      "learning_rate": 5.4767726161369194e-05,
+      "loss": 0.4601,
+      "step": 415
+    },
+    {
+      "epoch": 0.4875217643644806,
+      "grad_norm": 0.03868545591831207,
+      "learning_rate": 5.4156479217603914e-05,
+      "loss": 0.4512,
+      "step": 420
+    },
+    {
+      "epoch": 0.4933255948926291,
+      "grad_norm": 0.04135012626647949,
+      "learning_rate": 5.354523227383863e-05,
+      "loss": 0.4592,
+      "step": 425
+    },
+    {
+      "epoch": 0.4991294254207777,
+      "grad_norm": 0.042650043964385986,
+      "learning_rate": 5.2933985330073355e-05,
+      "loss": 0.4468,
+      "step": 430
+    },
+    {
+      "epoch": 0.5049332559489262,
+      "grad_norm": 0.05115941911935806,
+      "learning_rate": 5.232273838630807e-05,
+      "loss": 0.4705,
+      "step": 435
+    },
+    {
+      "epoch": 0.5107370864770748,
+      "grad_norm": 0.04721362888813019,
+      "learning_rate": 5.1711491442542795e-05,
+      "loss": 0.4677,
+      "step": 440
+    },
+    {
+      "epoch": 0.5165409170052234,
+      "grad_norm": 0.03605891764163971,
+      "learning_rate": 5.110024449877751e-05,
+      "loss": 0.4683,
+      "step": 445
+    },
+    {
+      "epoch": 0.522344747533372,
+      "grad_norm": 0.04193539544939995,
+      "learning_rate": 5.048899755501223e-05,
+      "loss": 0.477,
+      "step": 450
+    },
+    {
+      "epoch": 0.5281485780615206,
+      "grad_norm": 0.03736807778477669,
+      "learning_rate": 4.987775061124695e-05,
+      "loss": 0.4456,
+      "step": 455
+    },
+    {
+      "epoch": 0.5339524085896692,
+      "grad_norm": 0.03756151720881462,
+      "learning_rate": 4.926650366748167e-05,
+      "loss": 0.4636,
+      "step": 460
+    },
+    {
+      "epoch": 0.5397562391178178,
+      "grad_norm": 0.03452502563595772,
+      "learning_rate": 4.865525672371638e-05,
+      "loss": 0.4673,
+      "step": 465
+    },
+    {
+      "epoch": 0.5455600696459664,
+      "grad_norm": 0.03386940434575081,
+      "learning_rate": 4.80440097799511e-05,
+      "loss": 0.4669,
+      "step": 470
+    },
+    {
+      "epoch": 0.551363900174115,
+      "grad_norm": 0.041803594678640366,
+      "learning_rate": 4.743276283618582e-05,
+      "loss": 0.443,
+      "step": 475
+    },
+    {
+      "epoch": 0.5571677307022634,
+      "grad_norm": 0.04277484491467476,
+      "learning_rate": 4.6821515892420544e-05,
+      "loss": 0.4322,
+      "step": 480
+    },
+    {
+      "epoch": 0.562971561230412,
+      "grad_norm": 0.03855544701218605,
+      "learning_rate": 4.6210268948655264e-05,
+      "loss": 0.4607,
+      "step": 485
+    },
+    {
+      "epoch": 0.5687753917585606,
+      "grad_norm": 0.03208433464169502,
+      "learning_rate": 4.559902200488998e-05,
+      "loss": 0.4648,
+      "step": 490
+    },
+    {
+      "epoch": 0.5745792222867092,
+      "grad_norm": 0.0425882525742054,
+      "learning_rate": 4.49877750611247e-05,
+      "loss": 0.4453,
+      "step": 495
+    },
+    {
+      "epoch": 0.5803830528148578,
+      "grad_norm": 0.03916308283805847,
+      "learning_rate": 4.437652811735942e-05,
+      "loss": 0.4543,
+      "step": 500
+    },
+    {
+      "epoch": 0.5861868833430064,
+      "grad_norm": 0.03801513463258743,
+      "learning_rate": 4.376528117359414e-05,
+      "loss": 0.459,
+      "step": 505
+    },
+    {
+      "epoch": 0.591990713871155,
+      "grad_norm": 0.03717612475156784,
+      "learning_rate": 4.315403422982885e-05,
+      "loss": 0.4646,
+      "step": 510
+    },
+    {
+      "epoch": 0.5977945443993036,
+      "grad_norm": 0.037225909531116486,
+      "learning_rate": 4.254278728606357e-05,
+      "loss": 0.4631,
+      "step": 515
+    },
+    {
+      "epoch": 0.6035983749274522,
+      "grad_norm": 0.03405251353979111,
+      "learning_rate": 4.193154034229829e-05,
+      "loss": 0.4698,
+      "step": 520
+    },
+    {
+      "epoch": 0.6094022054556006,
+      "grad_norm": 0.04223908483982086,
+      "learning_rate": 4.132029339853301e-05,
+      "loss": 0.4611,
+      "step": 525
+    },
+    {
+      "epoch": 0.6152060359837492,
+      "grad_norm": 0.05366433411836624,
+      "learning_rate": 4.070904645476773e-05,
+      "loss": 0.4559,
+      "step": 530
+    },
+    {
+      "epoch": 0.6210098665118978,
+      "grad_norm": 0.04716875031590462,
+      "learning_rate": 4.0097799511002446e-05,
+      "loss": 0.464,
+      "step": 535
+    },
+    {
+      "epoch": 0.6268136970400464,
+      "grad_norm": 0.038778018206357956,
+      "learning_rate": 3.9486552567237166e-05,
+      "loss": 0.4494,
+      "step": 540
+    },
+    {
+      "epoch": 0.632617527568195,
+      "grad_norm": 0.03278287500143051,
+      "learning_rate": 3.8875305623471886e-05,
+      "loss": 0.4398,
+      "step": 545
+    },
+    {
+      "epoch": 0.6384213580963436,
+      "grad_norm": 0.04271184280514717,
+      "learning_rate": 3.8264058679706607e-05,
+      "loss": 0.4816,
+      "step": 550
+    },
+    {
+      "epoch": 0.6442251886244922,
+      "grad_norm": 0.04001977667212486,
+      "learning_rate": 3.765281173594133e-05,
+      "loss": 0.4452,
+      "step": 555
+    },
+    {
+      "epoch": 0.6500290191526408,
+      "grad_norm": 0.034920286387205124,
+      "learning_rate": 3.704156479217604e-05,
+      "loss": 0.4518,
+      "step": 560
+    },
+    {
+      "epoch": 0.6558328496807894,
+      "grad_norm": 0.030343499034643173,
+      "learning_rate": 3.643031784841076e-05,
+      "loss": 0.4676,
+      "step": 565
+    },
+    {
+      "epoch": 0.6616366802089378,
+      "grad_norm": 0.040206462144851685,
+      "learning_rate": 3.581907090464548e-05,
+      "loss": 0.4537,
+      "step": 570
+    },
+    {
+      "epoch": 0.6674405107370864,
+      "grad_norm": 0.04354901611804962,
+      "learning_rate": 3.52078239608802e-05,
+      "loss": 0.4586,
+      "step": 575
+    },
+    {
+      "epoch": 0.673244341265235,
+      "grad_norm": 0.049322061240673065,
+      "learning_rate": 3.4596577017114914e-05,
+      "loss": 0.4548,
+      "step": 580
+    },
+    {
+      "epoch": 0.6790481717933836,
+      "grad_norm": 0.03693823888897896,
+      "learning_rate": 3.3985330073349635e-05,
+      "loss": 0.4679,
+      "step": 585
+    },
+    {
+      "epoch": 0.6848520023215322,
+      "grad_norm": 0.03417356684803963,
+      "learning_rate": 3.3374083129584355e-05,
+      "loss": 0.4676,
+      "step": 590
+    },
+    {
+      "epoch": 0.6906558328496808,
+      "grad_norm": 0.034008949995040894,
+      "learning_rate": 3.2762836185819075e-05,
+      "loss": 0.4581,
+      "step": 595
+    },
+    {
+      "epoch": 0.6964596633778294,
+      "grad_norm": 0.040350962430238724,
+      "learning_rate": 3.2151589242053795e-05,
+      "loss": 0.4714,
+      "step": 600
+    },
+    {
+      "epoch": 0.702263493905978,
+      "grad_norm": 0.03704742714762688,
+      "learning_rate": 3.154034229828851e-05,
+      "loss": 0.4435,
+      "step": 605
+    },
+    {
+      "epoch": 0.7080673244341266,
+      "grad_norm": 0.036646343767642975,
+      "learning_rate": 3.092909535452323e-05,
+      "loss": 0.4721,
+      "step": 610
+    },
+    {
+      "epoch": 0.7138711549622752,
+      "grad_norm": 0.034359563142061234,
+      "learning_rate": 3.031784841075795e-05,
+      "loss": 0.4632,
+      "step": 615
+    },
+    {
+      "epoch": 0.7196749854904236,
+      "grad_norm": 0.037978023290634155,
+      "learning_rate": 2.9706601466992666e-05,
+      "loss": 0.4498,
+      "step": 620
+    },
+    {
+      "epoch": 0.7254788160185722,
+      "grad_norm": 0.041571393609046936,
+      "learning_rate": 2.9095354523227386e-05,
+      "loss": 0.4366,
+      "step": 625
+    },
+    {
+      "epoch": 0.7312826465467208,
+      "grad_norm": 0.04250132292509079,
+      "learning_rate": 2.8484107579462103e-05,
+      "loss": 0.4545,
+      "step": 630
+    },
+    {
+      "epoch": 0.7370864770748694,
+      "grad_norm": 0.03715914487838745,
+      "learning_rate": 2.7872860635696824e-05,
+      "loss": 0.4503,
+      "step": 635
+    },
+    {
+      "epoch": 0.742890307603018,
+      "grad_norm": 0.03634462505578995,
+      "learning_rate": 2.7261613691931544e-05,
+      "loss": 0.4694,
+      "step": 640
+    },
+    {
+      "epoch": 0.7486941381311666,
+      "grad_norm": 0.03091687522828579,
+      "learning_rate": 2.665036674816626e-05,
+      "loss": 0.4802,
+      "step": 645
+    },
+    {
+      "epoch": 0.7544979686593152,
+      "grad_norm": 0.03848150745034218,
+      "learning_rate": 2.603911980440098e-05,
+      "loss": 0.4611,
+      "step": 650
+    },
+    {
+      "epoch": 0.7603017991874638,
+      "grad_norm": 0.04081200435757637,
+      "learning_rate": 2.5427872860635698e-05,
+      "loss": 0.4399,
+      "step": 655
+    },
+    {
+      "epoch": 0.7661056297156124,
+      "grad_norm": 0.033600542694330215,
+      "learning_rate": 2.4816625916870418e-05,
+      "loss": 0.4511,
+      "step": 660
+    },
+    {
+      "epoch": 0.7719094602437608,
+      "grad_norm": 0.038769450038671494,
+      "learning_rate": 2.4205378973105135e-05,
+      "loss": 0.5382,
+      "step": 665
+    },
+    {
+      "epoch": 0.7777132907719094,
+      "grad_norm": 0.034687142819166183,
+      "learning_rate": 2.3594132029339855e-05,
+      "loss": 0.4387,
+      "step": 670
+    },
+    {
+      "epoch": 0.783517121300058,
+      "grad_norm": 0.042531248182058334,
+      "learning_rate": 2.2982885085574575e-05,
+      "loss": 0.4295,
+      "step": 675
+    },
+    {
+      "epoch": 0.7893209518282066,
+      "grad_norm": 0.04859015345573425,
+      "learning_rate": 2.2371638141809292e-05,
+      "loss": 0.4528,
+      "step": 680
+    },
+    {
+      "epoch": 0.7951247823563552,
+      "grad_norm": 0.03576788306236267,
+      "learning_rate": 2.1760391198044012e-05,
+      "loss": 0.4714,
+      "step": 685
+    },
+    {
+      "epoch": 0.8009286128845038,
+      "grad_norm": 0.03525792807340622,
+      "learning_rate": 2.114914425427873e-05,
+      "loss": 0.4466,
+      "step": 690
+    },
+    {
+      "epoch": 0.8067324434126524,
+      "grad_norm": 0.0542202927172184,
+      "learning_rate": 2.053789731051345e-05,
+      "loss": 0.4706,
+      "step": 695
+    },
+    {
+      "epoch": 0.812536273940801,
+      "grad_norm": 0.042330674827098846,
+      "learning_rate": 1.9926650366748166e-05,
+      "loss": 0.4739,
+      "step": 700
+    },
+    {
+      "epoch": 0.8183401044689496,
+      "grad_norm": 0.060287829488515854,
+      "learning_rate": 1.9315403422982887e-05,
+      "loss": 0.4451,
+      "step": 705
+    },
+    {
+      "epoch": 0.824143934997098,
+      "grad_norm": 0.03659594804048538,
+      "learning_rate": 1.8704156479217603e-05,
+      "loss": 0.4488,
+      "step": 710
+    },
+    {
+      "epoch": 0.8299477655252466,
+      "grad_norm": 0.03669734671711922,
+      "learning_rate": 1.8092909535452324e-05,
+      "loss": 0.4645,
+      "step": 715
+    },
+    {
+      "epoch": 0.8357515960533952,
+      "grad_norm": 0.03742624819278717,
+      "learning_rate": 1.7481662591687044e-05,
+      "loss": 0.4745,
+      "step": 720
+    },
+    {
+      "epoch": 0.8415554265815438,
+      "grad_norm": 0.04164464771747589,
+      "learning_rate": 1.687041564792176e-05,
+      "loss": 0.4573,
+      "step": 725
+    },
+    {
+      "epoch": 0.8473592571096924,
+      "grad_norm": 0.04054637998342514,
+      "learning_rate": 1.625916870415648e-05,
+      "loss": 0.4485,
+      "step": 730
+    },
+    {
+      "epoch": 0.853163087637841,
+      "grad_norm": 0.03948818892240524,
+      "learning_rate": 1.5647921760391198e-05,
+      "loss": 0.4802,
+      "step": 735
+    },
+    {
+      "epoch": 0.8589669181659896,
+      "grad_norm": 0.053339846432209015,
+      "learning_rate": 1.5036674816625918e-05,
+      "loss": 0.4619,
+      "step": 740
+    },
+    {
+      "epoch": 0.8647707486941382,
+      "grad_norm": 0.03236434608697891,
+      "learning_rate": 1.4425427872860637e-05,
+      "loss": 0.4488,
+      "step": 745
+    },
+    {
+      "epoch": 0.8705745792222868,
+      "grad_norm": 0.03988557681441307,
+      "learning_rate": 1.3814180929095355e-05,
+      "loss": 0.4524,
+      "step": 750
+    },
+    {
+      "epoch": 0.8763784097504352,
+      "grad_norm": 0.03730703890323639,
+      "learning_rate": 1.3202933985330074e-05,
+      "loss": 0.4265,
+      "step": 755
+    },
+    {
+      "epoch": 0.8821822402785838,
+      "grad_norm": 0.0383509136736393,
+      "learning_rate": 1.2591687041564792e-05,
+      "loss": 0.4571,
+      "step": 760
+    },
+    {
+      "epoch": 0.8879860708067324,
+      "grad_norm": 0.036255862563848495,
+      "learning_rate": 1.198044009779951e-05,
+      "loss": 0.5154,
+      "step": 765
+    },
+    {
+      "epoch": 0.893789901334881,
+      "grad_norm": 0.05152517929673195,
+      "learning_rate": 1.136919315403423e-05,
+      "loss": 0.4509,
+      "step": 770
+    },
+    {
+      "epoch": 0.8995937318630296,
+      "grad_norm": 0.038996148854494095,
+      "learning_rate": 1.075794621026895e-05,
+      "loss": 0.4536,
+      "step": 775
+    },
+    {
+      "epoch": 0.9053975623911782,
+      "grad_norm": 0.03999714180827141,
+      "learning_rate": 1.0146699266503668e-05,
+      "loss": 0.4392,
+      "step": 780
+    },
+    {
+      "epoch": 0.9112013929193268,
+      "grad_norm": 0.034626469016075134,
+      "learning_rate": 9.535452322738387e-06,
+      "loss": 0.4614,
+      "step": 785
+    },
+    {
+      "epoch": 0.9170052234474754,
+      "grad_norm": 0.03376827761530876,
+      "learning_rate": 8.924205378973105e-06,
+      "loss": 0.4479,
+      "step": 790
+    },
+    {
+      "epoch": 0.922809053975624,
+      "grad_norm": 0.08973196148872375,
+      "learning_rate": 8.312958435207824e-06,
+      "loss": 0.4598,
+      "step": 795
+    },
+    {
+      "epoch": 0.9286128845037724,
+      "grad_norm": 0.03864937275648117,
+      "learning_rate": 7.701711491442542e-06,
+      "loss": 0.4469,
+      "step": 800
+    },
+    {
+      "epoch": 0.934416715031921,
+      "grad_norm": 0.038818832486867905,
+      "learning_rate": 7.090464547677262e-06,
+      "loss": 0.4173,
+      "step": 805
+    },
+    {
+      "epoch": 0.9402205455600696,
+      "grad_norm": 0.03407546505331993,
+      "learning_rate": 6.47921760391198e-06,
+      "loss": 0.4399,
+      "step": 810
+    },
+    {
+      "epoch": 0.9460243760882182,
+      "grad_norm": 0.03433331474661827,
+      "learning_rate": 5.8679706601467e-06,
+      "loss": 0.4811,
+      "step": 815
+    },
+    {
+      "epoch": 0.9518282066163668,
+      "grad_norm": 0.03800525888800621,
+      "learning_rate": 5.256723716381418e-06,
+      "loss": 0.4504,
+      "step": 820
+    },
+    {
+      "epoch": 0.9576320371445154,
+      "grad_norm": 0.041561078280210495,
+      "learning_rate": 4.645476772616137e-06,
+      "loss": 0.4563,
+      "step": 825
+    },
+    {
+      "epoch": 0.963435867672664,
+      "grad_norm": 0.04036097601056099,
+      "learning_rate": 4.034229828850856e-06,
+      "loss": 0.4296,
+      "step": 830
+    },
+    {
+      "epoch": 0.9692396982008126,
+      "grad_norm": 0.03923596069216728,
+      "learning_rate": 3.4229828850855746e-06,
+      "loss": 0.4604,
+      "step": 835
+    },
+    {
+      "epoch": 0.9750435287289612,
+      "grad_norm": 0.033339060842990875,
+      "learning_rate": 2.8117359413202936e-06,
+      "loss": 0.4565,
+      "step": 840
+    },
+    {
+      "epoch": 0.9808473592571096,
+      "grad_norm": 0.034090008586645126,
+      "learning_rate": 2.2004889975550126e-06,
+      "loss": 0.4601,
+      "step": 845
+    },
+    {
+      "epoch": 0.9866511897852582,
+      "grad_norm": 0.04013431444764137,
+      "learning_rate": 1.5892420537897311e-06,
+      "loss": 0.4687,
+      "step": 850
+    },
+    {
+      "epoch": 0.9924550203134068,
+      "grad_norm": 0.03881672024726868,
+      "learning_rate": 9.779951100244499e-07,
+      "loss": 0.422,
+      "step": 855
+    },
+    {
+      "epoch": 0.9982588508415554,
+      "grad_norm": 0.047759708017110825,
+      "learning_rate": 3.667481662591687e-07,
+      "loss": 0.4518,
+      "step": 860
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 862,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.192258616665068e+20,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

Qwen3-30B-A3B-Thinking-2507-sft-fusang/checkpoint-862/README.md

@@ -0,0 +1,207 @@
+---
+base_model: /share/models/Qwen3-30B-A3B-Thinking-2507
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:/share/models/Qwen3-30B-A3B-Thinking-2507
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.17.1

Qwen3-30B-A3B-Thinking-2507-sft-fusang/checkpoint-862/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

Qwen3-30B-A3B-Thinking-2507-sft-fusang/checkpoint-862/trainer_state.json

@@ -0,0 +1,1238 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 862,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.005803830528148578,
+      "grad_norm": 0.11167649179697037,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.4179,
+      "step": 5
+    },
+    {
+      "epoch": 0.011607661056297156,
+      "grad_norm": 0.05878744646906853,
+      "learning_rate": 2.0454545454545457e-05,
+      "loss": 0.4451,
+      "step": 10
+    },
+    {
+      "epoch": 0.017411491584445733,
+      "grad_norm": 0.04675213620066643,
+      "learning_rate": 3.181818181818182e-05,
+      "loss": 0.4582,
+      "step": 15
+    },
+    {
+      "epoch": 0.023215322112594312,
+      "grad_norm": 0.047889795154333115,
+      "learning_rate": 4.318181818181819e-05,
+      "loss": 0.4506,
+      "step": 20
+    },
+    {
+      "epoch": 0.02901915264074289,
+      "grad_norm": 0.05356425419449806,
+      "learning_rate": 5.4545454545454546e-05,
+      "loss": 0.4452,
+      "step": 25
+    },
+    {
+      "epoch": 0.034822983168891465,
+      "grad_norm": 0.05251117795705795,
+      "learning_rate": 6.59090909090909e-05,
+      "loss": 0.4305,
+      "step": 30
+    },
+    {
+      "epoch": 0.04062681369704005,
+      "grad_norm": 0.055615171790122986,
+      "learning_rate": 7.727272727272727e-05,
+      "loss": 0.4372,
+      "step": 35
+    },
+    {
+      "epoch": 0.046430644225188625,
+      "grad_norm": 0.03562075272202492,
+      "learning_rate": 8.863636363636364e-05,
+      "loss": 0.4492,
+      "step": 40
+    },
+    {
+      "epoch": 0.0522344747533372,
+      "grad_norm": 0.04553038999438286,
+      "learning_rate": 0.0001,
+      "loss": 0.4499,
+      "step": 45
+    },
+    {
+      "epoch": 0.05803830528148578,
+      "grad_norm": 0.05148487165570259,
+      "learning_rate": 9.938875305623473e-05,
+      "loss": 0.4576,
+      "step": 50
+    },
+    {
+      "epoch": 0.06384213580963435,
+      "grad_norm": 0.04853189364075661,
+      "learning_rate": 9.877750611246945e-05,
+      "loss": 0.4502,
+      "step": 55
+    },
+    {
+      "epoch": 0.06964596633778293,
+      "grad_norm": 0.045064426958560944,
+      "learning_rate": 9.816625916870417e-05,
+      "loss": 0.4499,
+      "step": 60
+    },
+    {
+      "epoch": 0.07544979686593152,
+      "grad_norm": 0.04164193943142891,
+      "learning_rate": 9.755501222493889e-05,
+      "loss": 0.4395,
+      "step": 65
+    },
+    {
+      "epoch": 0.0812536273940801,
+      "grad_norm": 0.03383349999785423,
+      "learning_rate": 9.69437652811736e-05,
+      "loss": 0.4371,
+      "step": 70
+    },
+    {
+      "epoch": 0.08705745792222867,
+      "grad_norm": 0.04373278468847275,
+      "learning_rate": 9.633251833740831e-05,
+      "loss": 0.4755,
+      "step": 75
+    },
+    {
+      "epoch": 0.09286128845037725,
+      "grad_norm": 0.04336105287075043,
+      "learning_rate": 9.572127139364305e-05,
+      "loss": 0.4265,
+      "step": 80
+    },
+    {
+      "epoch": 0.09866511897852583,
+      "grad_norm": 0.034007731825113297,
+      "learning_rate": 9.511002444987775e-05,
+      "loss": 0.4328,
+      "step": 85
+    },
+    {
+      "epoch": 0.1044689495066744,
+      "grad_norm": 0.03617757931351662,
+      "learning_rate": 9.449877750611247e-05,
+      "loss": 0.438,
+      "step": 90
+    },
+    {
+      "epoch": 0.11027278003482298,
+      "grad_norm": 0.0359223447740078,
+      "learning_rate": 9.38875305623472e-05,
+      "loss": 0.4366,
+      "step": 95
+    },
+    {
+      "epoch": 0.11607661056297155,
+      "grad_norm": 0.042958397418260574,
+      "learning_rate": 9.327628361858191e-05,
+      "loss": 0.4523,
+      "step": 100
+    },
+    {
+      "epoch": 0.12188044109112015,
+      "grad_norm": 0.04804977774620056,
+      "learning_rate": 9.266503667481663e-05,
+      "loss": 0.4356,
+      "step": 105
+    },
+    {
+      "epoch": 0.1276842716192687,
+      "grad_norm": 0.03226548433303833,
+      "learning_rate": 9.205378973105135e-05,
+      "loss": 0.4381,
+      "step": 110
+    },
+    {
+      "epoch": 0.1334881021474173,
+      "grad_norm": 0.032912448048591614,
+      "learning_rate": 9.144254278728606e-05,
+      "loss": 0.451,
+      "step": 115
+    },
+    {
+      "epoch": 0.13929193267556586,
+      "grad_norm": 0.0320778451859951,
+      "learning_rate": 9.08312958435208e-05,
+      "loss": 0.4554,
+      "step": 120
+    },
+    {
+      "epoch": 0.14509576320371445,
+      "grad_norm": 0.03928998112678528,
+      "learning_rate": 9.02200488997555e-05,
+      "loss": 0.432,
+      "step": 125
+    },
+    {
+      "epoch": 0.15089959373186304,
+      "grad_norm": 0.04315907880663872,
+      "learning_rate": 8.960880195599024e-05,
+      "loss": 0.4454,
+      "step": 130
+    },
+    {
+      "epoch": 0.1567034242600116,
+      "grad_norm": 0.035480398684740067,
+      "learning_rate": 8.899755501222494e-05,
+      "loss": 0.4537,
+      "step": 135
+    },
+    {
+      "epoch": 0.1625072547881602,
+      "grad_norm": 0.03198500722646713,
+      "learning_rate": 8.838630806845966e-05,
+      "loss": 0.4658,
+      "step": 140
+    },
+    {
+      "epoch": 0.16831108531630876,
+      "grad_norm": 0.03255561366677284,
+      "learning_rate": 8.777506112469438e-05,
+      "loss": 0.4593,
+      "step": 145
+    },
+    {
+      "epoch": 0.17411491584445735,
+      "grad_norm": 0.03970634937286377,
+      "learning_rate": 8.71638141809291e-05,
+      "loss": 0.4495,
+      "step": 150
+    },
+    {
+      "epoch": 0.1799187463726059,
+      "grad_norm": 0.03526106849312782,
+      "learning_rate": 8.655256723716382e-05,
+      "loss": 0.424,
+      "step": 155
+    },
+    {
+      "epoch": 0.1857225769007545,
+      "grad_norm": 0.051440611481666565,
+      "learning_rate": 8.594132029339854e-05,
+      "loss": 0.4457,
+      "step": 160
+    },
+    {
+      "epoch": 0.1915264074289031,
+      "grad_norm": 0.032669343054294586,
+      "learning_rate": 8.533007334963325e-05,
+      "loss": 0.4457,
+      "step": 165
+    },
+    {
+      "epoch": 0.19733023795705165,
+      "grad_norm": 0.03135131299495697,
+      "learning_rate": 8.471882640586798e-05,
+      "loss": 0.445,
+      "step": 170
+    },
+    {
+      "epoch": 0.20313406848520024,
+      "grad_norm": 0.03780507668852806,
+      "learning_rate": 8.410757946210269e-05,
+      "loss": 0.4516,
+      "step": 175
+    },
+    {
+      "epoch": 0.2089378990133488,
+      "grad_norm": 0.039683420211076736,
+      "learning_rate": 8.349633251833741e-05,
+      "loss": 0.4502,
+      "step": 180
+    },
+    {
+      "epoch": 0.2147417295414974,
+      "grad_norm": 0.03234437108039856,
+      "learning_rate": 8.288508557457213e-05,
+      "loss": 0.4509,
+      "step": 185
+    },
+    {
+      "epoch": 0.22054556006964596,
+      "grad_norm": 0.03360700234770775,
+      "learning_rate": 8.227383863080685e-05,
+      "loss": 0.4457,
+      "step": 190
+    },
+    {
+      "epoch": 0.22634939059779455,
+      "grad_norm": 0.03230682387948036,
+      "learning_rate": 8.166259168704157e-05,
+      "loss": 0.4228,
+      "step": 195
+    },
+    {
+      "epoch": 0.2321532211259431,
+      "grad_norm": 0.03940160945057869,
+      "learning_rate": 8.105134474327629e-05,
+      "loss": 0.4415,
+      "step": 200
+    },
+    {
+      "epoch": 0.2379570516540917,
+      "grad_norm": 0.03374328091740608,
+      "learning_rate": 8.044009779951101e-05,
+      "loss": 0.4356,
+      "step": 205
+    },
+    {
+      "epoch": 0.2437608821822403,
+      "grad_norm": 0.031871989369392395,
+      "learning_rate": 7.982885085574573e-05,
+      "loss": 0.4696,
+      "step": 210
+    },
+    {
+      "epoch": 0.24956471271038885,
+      "grad_norm": 0.03276577964425087,
+      "learning_rate": 7.921760391198044e-05,
+      "loss": 0.4427,
+      "step": 215
+    },
+    {
+      "epoch": 0.2553685432385374,
+      "grad_norm": 0.02574090287089348,
+      "learning_rate": 7.860635696821517e-05,
+      "loss": 0.4291,
+      "step": 220
+    },
+    {
+      "epoch": 0.261172373766686,
+      "grad_norm": 0.036154329776763916,
+      "learning_rate": 7.799511002444988e-05,
+      "loss": 0.4509,
+      "step": 225
+    },
+    {
+      "epoch": 0.2669762042948346,
+      "grad_norm": 0.04011650010943413,
+      "learning_rate": 7.73838630806846e-05,
+      "loss": 0.4358,
+      "step": 230
+    },
+    {
+      "epoch": 0.2727800348229832,
+      "grad_norm": 0.030674200505018234,
+      "learning_rate": 7.677261613691932e-05,
+      "loss": 0.4637,
+      "step": 235
+    },
+    {
+      "epoch": 0.2785838653511317,
+      "grad_norm": 0.032683223485946655,
+      "learning_rate": 7.616136919315404e-05,
+      "loss": 0.4516,
+      "step": 240
+    },
+    {
+      "epoch": 0.2843876958792803,
+      "grad_norm": 0.027314957231283188,
+      "learning_rate": 7.555012224938876e-05,
+      "loss": 0.4402,
+      "step": 245
+    },
+    {
+      "epoch": 0.2901915264074289,
+      "grad_norm": 0.034763675183057785,
+      "learning_rate": 7.493887530562348e-05,
+      "loss": 0.4643,
+      "step": 250
+    },
+    {
+      "epoch": 0.2959953569355775,
+      "grad_norm": 0.04508906230330467,
+      "learning_rate": 7.432762836185819e-05,
+      "loss": 0.4428,
+      "step": 255
+    },
+    {
+      "epoch": 0.3017991874637261,
+      "grad_norm": 0.034660518169403076,
+      "learning_rate": 7.371638141809292e-05,
+      "loss": 0.4656,
+      "step": 260
+    },
+    {
+      "epoch": 0.3076030179918746,
+      "grad_norm": 0.02968621253967285,
+      "learning_rate": 7.310513447432763e-05,
+      "loss": 0.4309,
+      "step": 265
+    },
+    {
+      "epoch": 0.3134068485200232,
+      "grad_norm": 0.03332359343767166,
+      "learning_rate": 7.249388753056235e-05,
+      "loss": 0.435,
+      "step": 270
+    },
+    {
+      "epoch": 0.3192106790481718,
+      "grad_norm": 0.03404337912797928,
+      "learning_rate": 7.188264058679707e-05,
+      "loss": 0.4499,
+      "step": 275
+    },
+    {
+      "epoch": 0.3250145095763204,
+      "grad_norm": 0.036635175347328186,
+      "learning_rate": 7.127139364303179e-05,
+      "loss": 0.4516,
+      "step": 280
+    },
+    {
+      "epoch": 0.3308183401044689,
+      "grad_norm": 0.030087586492300034,
+      "learning_rate": 7.066014669926651e-05,
+      "loss": 0.4642,
+      "step": 285
+    },
+    {
+      "epoch": 0.3366221706326175,
+      "grad_norm": 0.029073191806674004,
+      "learning_rate": 7.004889975550123e-05,
+      "loss": 0.4457,
+      "step": 290
+    },
+    {
+      "epoch": 0.3424260011607661,
+      "grad_norm": 0.026812931522727013,
+      "learning_rate": 6.943765281173595e-05,
+      "loss": 0.4451,
+      "step": 295
+    },
+    {
+      "epoch": 0.3482298316889147,
+      "grad_norm": 0.034956712275743484,
+      "learning_rate": 6.882640586797067e-05,
+      "loss": 0.4481,
+      "step": 300
+    },
+    {
+      "epoch": 0.3540336622170633,
+      "grad_norm": 0.03261638805270195,
+      "learning_rate": 6.821515892420538e-05,
+      "loss": 0.4313,
+      "step": 305
+    },
+    {
+      "epoch": 0.3598374927452118,
+      "grad_norm": 0.03206435590982437,
+      "learning_rate": 6.760391198044011e-05,
+      "loss": 0.4628,
+      "step": 310
+    },
+    {
+      "epoch": 0.3656413232733604,
+      "grad_norm": 0.02776835672557354,
+      "learning_rate": 6.699266503667482e-05,
+      "loss": 0.443,
+      "step": 315
+    },
+    {
+      "epoch": 0.371445153801509,
+      "grad_norm": 0.024607423692941666,
+      "learning_rate": 6.638141809290954e-05,
+      "loss": 0.4351,
+      "step": 320
+    },
+    {
+      "epoch": 0.3772489843296576,
+      "grad_norm": 0.03608020395040512,
+      "learning_rate": 6.577017114914426e-05,
+      "loss": 0.4612,
+      "step": 325
+    },
+    {
+      "epoch": 0.3830528148578062,
+      "grad_norm": 0.030558280646800995,
+      "learning_rate": 6.515892420537898e-05,
+      "loss": 0.4346,
+      "step": 330
+    },
+    {
+      "epoch": 0.3888566453859547,
+      "grad_norm": 0.038997165858745575,
+      "learning_rate": 6.45476772616137e-05,
+      "loss": 0.4484,
+      "step": 335
+    },
+    {
+      "epoch": 0.3946604759141033,
+      "grad_norm": 0.027846684679389,
+      "learning_rate": 6.393643031784842e-05,
+      "loss": 0.4389,
+      "step": 340
+    },
+    {
+      "epoch": 0.4004643064422519,
+      "grad_norm": 0.02644030563533306,
+      "learning_rate": 6.332518337408312e-05,
+      "loss": 0.4496,
+      "step": 345
+    },
+    {
+      "epoch": 0.4062681369704005,
+      "grad_norm": 0.033678922802209854,
+      "learning_rate": 6.271393643031786e-05,
+      "loss": 0.4302,
+      "step": 350
+    },
+    {
+      "epoch": 0.412071967498549,
+      "grad_norm": 0.03207817301154137,
+      "learning_rate": 6.210268948655256e-05,
+      "loss": 0.4404,
+      "step": 355
+    },
+    {
+      "epoch": 0.4178757980266976,
+      "grad_norm": 0.03058437816798687,
+      "learning_rate": 6.14914425427873e-05,
+      "loss": 0.4654,
+      "step": 360
+    },
+    {
+      "epoch": 0.4236796285548462,
+      "grad_norm": 0.04575405269861221,
+      "learning_rate": 6.0880195599022005e-05,
+      "loss": 0.4386,
+      "step": 365
+    },
+    {
+      "epoch": 0.4294834590829948,
+      "grad_norm": 0.03139202296733856,
+      "learning_rate": 6.026894865525673e-05,
+      "loss": 0.434,
+      "step": 370
+    },
+    {
+      "epoch": 0.4352872896111434,
+      "grad_norm": 0.03597712516784668,
+      "learning_rate": 5.9657701711491446e-05,
+      "loss": 0.462,
+      "step": 375
+    },
+    {
+      "epoch": 0.4410911201392919,
+      "grad_norm": 0.029897456988692284,
+      "learning_rate": 5.9046454767726166e-05,
+      "loss": 0.4312,
+      "step": 380
+    },
+    {
+      "epoch": 0.4468949506674405,
+      "grad_norm": 0.04324301332235336,
+      "learning_rate": 5.843520782396088e-05,
+      "loss": 0.4221,
+      "step": 385
+    },
+    {
+      "epoch": 0.4526987811955891,
+      "grad_norm": 0.027129394933581352,
+      "learning_rate": 5.7823960880195606e-05,
+      "loss": 0.4476,
+      "step": 390
+    },
+    {
+      "epoch": 0.4585026117237377,
+      "grad_norm": 0.03885333240032196,
+      "learning_rate": 5.721271393643032e-05,
+      "loss": 0.4329,
+      "step": 395
+    },
+    {
+      "epoch": 0.4643064422518862,
+      "grad_norm": 0.03425256907939911,
+      "learning_rate": 5.660146699266504e-05,
+      "loss": 0.443,
+      "step": 400
+    },
+    {
+      "epoch": 0.4701102727800348,
+      "grad_norm": 0.032868240028619766,
+      "learning_rate": 5.5990220048899754e-05,
+      "loss": 0.4394,
+      "step": 405
+    },
+    {
+      "epoch": 0.4759141033081834,
+      "grad_norm": 0.030719870701432228,
+      "learning_rate": 5.537897310513448e-05,
+      "loss": 0.4265,
+      "step": 410
+    },
+    {
+      "epoch": 0.481717933836332,
+      "grad_norm": 0.027947254478931427,
+      "learning_rate": 5.4767726161369194e-05,
+      "loss": 0.4471,
+      "step": 415
+    },
+    {
+      "epoch": 0.4875217643644806,
+      "grad_norm": 0.06247246265411377,
+      "learning_rate": 5.4156479217603914e-05,
+      "loss": 0.438,
+      "step": 420
+    },
+    {
+      "epoch": 0.4933255948926291,
+      "grad_norm": 0.03340250998735428,
+      "learning_rate": 5.354523227383863e-05,
+      "loss": 0.4452,
+      "step": 425
+    },
+    {
+      "epoch": 0.4991294254207777,
+      "grad_norm": 0.038149379193782806,
+      "learning_rate": 5.2933985330073355e-05,
+      "loss": 0.4332,
+      "step": 430
+    },
+    {
+      "epoch": 0.5049332559489262,
+      "grad_norm": 0.02780376374721527,
+      "learning_rate": 5.232273838630807e-05,
+      "loss": 0.4574,
+      "step": 435
+    },
+    {
+      "epoch": 0.5107370864770748,
+      "grad_norm": 0.031863778829574585,
+      "learning_rate": 5.1711491442542795e-05,
+      "loss": 0.4547,
+      "step": 440
+    },
+    {
+      "epoch": 0.5165409170052234,
+      "grad_norm": 0.030303943902254105,
+      "learning_rate": 5.110024449877751e-05,
+      "loss": 0.4547,
+      "step": 445
+    },
+    {
+      "epoch": 0.522344747533372,
+      "grad_norm": 0.03468239679932594,
+      "learning_rate": 5.048899755501223e-05,
+      "loss": 0.4657,
+      "step": 450
+    },
+    {
+      "epoch": 0.5281485780615206,
+      "grad_norm": 0.031695377081632614,
+      "learning_rate": 4.987775061124695e-05,
+      "loss": 0.4329,
+      "step": 455
+    },
+    {
+      "epoch": 0.5339524085896692,
+      "grad_norm": 0.034550122916698456,
+      "learning_rate": 4.926650366748167e-05,
+      "loss": 0.4511,
+      "step": 460
+    },
+    {
+      "epoch": 0.5397562391178178,
+      "grad_norm": 0.025001615285873413,
+      "learning_rate": 4.865525672371638e-05,
+      "loss": 0.4549,
+      "step": 465
+    },
+    {
+      "epoch": 0.5455600696459664,
+      "grad_norm": 0.028805740177631378,
+      "learning_rate": 4.80440097799511e-05,
+      "loss": 0.4543,
+      "step": 470
+    },
+    {
+      "epoch": 0.551363900174115,
+      "grad_norm": 0.03354053199291229,
+      "learning_rate": 4.743276283618582e-05,
+      "loss": 0.432,
+      "step": 475
+    },
+    {
+      "epoch": 0.5571677307022634,
+      "grad_norm": 0.030400967225432396,
+      "learning_rate": 4.6821515892420544e-05,
+      "loss": 0.4188,
+      "step": 480
+    },
+    {
+      "epoch": 0.562971561230412,
+      "grad_norm": 0.0275269765406847,
+      "learning_rate": 4.6210268948655264e-05,
+      "loss": 0.4491,
+      "step": 485
+    },
+    {
+      "epoch": 0.5687753917585606,
+      "grad_norm": 0.025978457182645798,
+      "learning_rate": 4.559902200488998e-05,
+      "loss": 0.4532,
+      "step": 490
+    },
+    {
+      "epoch": 0.5745792222867092,
+      "grad_norm": 0.029258042573928833,
+      "learning_rate": 4.49877750611247e-05,
+      "loss": 0.4325,
+      "step": 495
+    },
+    {
+      "epoch": 0.5803830528148578,
+      "grad_norm": 0.032496389001607895,
+      "learning_rate": 4.437652811735942e-05,
+      "loss": 0.4422,
+      "step": 500
+    },
+    {
+      "epoch": 0.5861868833430064,
+      "grad_norm": 0.03094470128417015,
+      "learning_rate": 4.376528117359414e-05,
+      "loss": 0.4462,
+      "step": 505
+    },
+    {
+      "epoch": 0.591990713871155,
+      "grad_norm": 0.028828352689743042,
+      "learning_rate": 4.315403422982885e-05,
+      "loss": 0.4516,
+      "step": 510
+    },
+    {
+      "epoch": 0.5977945443993036,
+      "grad_norm": 0.033804845064878464,
+      "learning_rate": 4.254278728606357e-05,
+      "loss": 0.4502,
+      "step": 515
+    },
+    {
+      "epoch": 0.6035983749274522,
+      "grad_norm": 0.025966903194785118,
+      "learning_rate": 4.193154034229829e-05,
+      "loss": 0.4564,
+      "step": 520
+    },
+    {
+      "epoch": 0.6094022054556006,
+      "grad_norm": 0.03651871904730797,
+      "learning_rate": 4.132029339853301e-05,
+      "loss": 0.4484,
+      "step": 525
+    },
+    {
+      "epoch": 0.6152060359837492,
+      "grad_norm": 0.030410734936594963,
+      "learning_rate": 4.070904645476773e-05,
+      "loss": 0.4432,
+      "step": 530
+    },
+    {
+      "epoch": 0.6210098665118978,
+      "grad_norm": 0.03372242674231529,
+      "learning_rate": 4.0097799511002446e-05,
+      "loss": 0.4512,
+      "step": 535
+    },
+    {
+      "epoch": 0.6268136970400464,
+      "grad_norm": 0.028306983411312103,
+      "learning_rate": 3.9486552567237166e-05,
+      "loss": 0.4364,
+      "step": 540
+    },
+    {
+      "epoch": 0.632617527568195,
+      "grad_norm": 0.03793470188975334,
+      "learning_rate": 3.8875305623471886e-05,
+      "loss": 0.4279,
+      "step": 545
+    },
+    {
+      "epoch": 0.6384213580963436,
+      "grad_norm": 0.03674202412366867,
+      "learning_rate": 3.8264058679706607e-05,
+      "loss": 0.4692,
+      "step": 550
+    },
+    {
+      "epoch": 0.6442251886244922,
+      "grad_norm": 0.03083740547299385,
+      "learning_rate": 3.765281173594133e-05,
+      "loss": 0.4326,
+      "step": 555
+    },
+    {
+      "epoch": 0.6500290191526408,
+      "grad_norm": 0.03499336168169975,
+      "learning_rate": 3.704156479217604e-05,
+      "loss": 0.4399,
+      "step": 560
+    },
+    {
+      "epoch": 0.6558328496807894,
+      "grad_norm": 0.023755021393299103,
+      "learning_rate": 3.643031784841076e-05,
+      "loss": 0.4552,
+      "step": 565
+    },
+    {
+      "epoch": 0.6616366802089378,
+      "grad_norm": 0.025819219648838043,
+      "learning_rate": 3.581907090464548e-05,
+      "loss": 0.4404,
+      "step": 570
+    },
+    {
+      "epoch": 0.6674405107370864,
+      "grad_norm": 0.033186931163072586,
+      "learning_rate": 3.52078239608802e-05,
+      "loss": 0.4472,
+      "step": 575
+    },
+    {
+      "epoch": 0.673244341265235,
+      "grad_norm": 0.033613502979278564,
+      "learning_rate": 3.4596577017114914e-05,
+      "loss": 0.4421,
+      "step": 580
+    },
+    {
+      "epoch": 0.6790481717933836,
+      "grad_norm": 0.02944432757794857,
+      "learning_rate": 3.3985330073349635e-05,
+      "loss": 0.4554,
+      "step": 585
+    },
+    {
+      "epoch": 0.6848520023215322,
+      "grad_norm": 0.03536716476082802,
+      "learning_rate": 3.3374083129584355e-05,
+      "loss": 0.4549,
+      "step": 590
+    },
+    {
+      "epoch": 0.6906558328496808,
+      "grad_norm": 0.024904152378439903,
+      "learning_rate": 3.2762836185819075e-05,
+      "loss": 0.445,
+      "step": 595
+    },
+    {
+      "epoch": 0.6964596633778294,
+      "grad_norm": 0.03327803313732147,
+      "learning_rate": 3.2151589242053795e-05,
+      "loss": 0.4601,
+      "step": 600
+    },
+    {
+      "epoch": 0.702263493905978,
+      "grad_norm": 0.04129062220454216,
+      "learning_rate": 3.154034229828851e-05,
+      "loss": 0.4304,
+      "step": 605
+    },
+    {
+      "epoch": 0.7080673244341266,
+      "grad_norm": 0.03460316359996796,
+      "learning_rate": 3.092909535452323e-05,
+      "loss": 0.4591,
+      "step": 610
+    },
+    {
+      "epoch": 0.7138711549622752,
+      "grad_norm": 0.03884468227624893,
+      "learning_rate": 3.031784841075795e-05,
+      "loss": 0.4509,
+      "step": 615
+    },
+    {
+      "epoch": 0.7196749854904236,
+      "grad_norm": 0.027258818969130516,
+      "learning_rate": 2.9706601466992666e-05,
+      "loss": 0.4375,
+      "step": 620
+    },
+    {
+      "epoch": 0.7254788160185722,
+      "grad_norm": 0.0322805680334568,
+      "learning_rate": 2.9095354523227386e-05,
+      "loss": 0.4245,
+      "step": 625
+    },
+    {
+      "epoch": 0.7312826465467208,
+      "grad_norm": 0.03353621065616608,
+      "learning_rate": 2.8484107579462103e-05,
+      "loss": 0.4416,
+      "step": 630
+    },
+    {
+      "epoch": 0.7370864770748694,
+      "grad_norm": 0.03136138245463371,
+      "learning_rate": 2.7872860635696824e-05,
+      "loss": 0.4376,
+      "step": 635
+    },
+    {
+      "epoch": 0.742890307603018,
+      "grad_norm": 0.02590339072048664,
+      "learning_rate": 2.7261613691931544e-05,
+      "loss": 0.4554,
+      "step": 640
+    },
+    {
+      "epoch": 0.7486941381311666,
+      "grad_norm": 0.023068683221936226,
+      "learning_rate": 2.665036674816626e-05,
+      "loss": 0.4679,
+      "step": 645
+    },
+    {
+      "epoch": 0.7544979686593152,
+      "grad_norm": 0.030758408829569817,
+      "learning_rate": 2.603911980440098e-05,
+      "loss": 0.4504,
+      "step": 650
+    },
+    {
+      "epoch": 0.7603017991874638,
+      "grad_norm": 0.03439256548881531,
+      "learning_rate": 2.5427872860635698e-05,
+      "loss": 0.4263,
+      "step": 655
+    },
+    {
+      "epoch": 0.7661056297156124,
+      "grad_norm": 0.027209492400288582,
+      "learning_rate": 2.4816625916870418e-05,
+      "loss": 0.4377,
+      "step": 660
+    },
+    {
+      "epoch": 0.7719094602437608,
+      "grad_norm": 0.028269799426198006,
+      "learning_rate": 2.4205378973105135e-05,
+      "loss": 0.5256,
+      "step": 665
+    },
+    {
+      "epoch": 0.7777132907719094,
+      "grad_norm": 0.027192717418074608,
+      "learning_rate": 2.3594132029339855e-05,
+      "loss": 0.4263,
+      "step": 670
+    },
+    {
+      "epoch": 0.783517121300058,
+      "grad_norm": 0.03299058973789215,
+      "learning_rate": 2.2982885085574575e-05,
+      "loss": 0.4184,
+      "step": 675
+    },
+    {
+      "epoch": 0.7893209518282066,
+      "grad_norm": 0.03457274287939072,
+      "learning_rate": 2.2371638141809292e-05,
+      "loss": 0.4398,
+      "step": 680
+    },
+    {
+      "epoch": 0.7951247823563552,
+      "grad_norm": 0.026992570608854294,
+      "learning_rate": 2.1760391198044012e-05,
+      "loss": 0.4589,
+      "step": 685
+    },
+    {
+      "epoch": 0.8009286128845038,
+      "grad_norm": 0.028259936720132828,
+      "learning_rate": 2.114914425427873e-05,
+      "loss": 0.4338,
+      "step": 690
+    },
+    {
+      "epoch": 0.8067324434126524,
+      "grad_norm": 0.026307212188839912,
+      "learning_rate": 2.053789731051345e-05,
+      "loss": 0.4583,
+      "step": 695
+    },
+    {
+      "epoch": 0.812536273940801,
+      "grad_norm": 0.03430171683430672,
+      "learning_rate": 1.9926650366748166e-05,
+      "loss": 0.4627,
+      "step": 700
+    },
+    {
+      "epoch": 0.8183401044689496,
+      "grad_norm": 0.03142637386918068,
+      "learning_rate": 1.9315403422982887e-05,
+      "loss": 0.4324,
+      "step": 705
+    },
+    {
+      "epoch": 0.824143934997098,
+      "grad_norm": 0.02762400545179844,
+      "learning_rate": 1.8704156479217603e-05,
+      "loss": 0.4366,
+      "step": 710
+    },
+    {
+      "epoch": 0.8299477655252466,
+      "grad_norm": 0.026468684896826744,
+      "learning_rate": 1.8092909535452324e-05,
+      "loss": 0.4527,
+      "step": 715
+    },
+    {
+      "epoch": 0.8357515960533952,
+      "grad_norm": 0.026880772784352303,
+      "learning_rate": 1.7481662591687044e-05,
+      "loss": 0.4621,
+      "step": 720
+    },
+    {
+      "epoch": 0.8415554265815438,
+      "grad_norm": 0.03330186754465103,
+      "learning_rate": 1.687041564792176e-05,
+      "loss": 0.4454,
+      "step": 725
+    },
+    {
+      "epoch": 0.8473592571096924,
+      "grad_norm": 0.03175678476691246,
+      "learning_rate": 1.625916870415648e-05,
+      "loss": 0.4359,
+      "step": 730
+    },
+    {
+      "epoch": 0.853163087637841,
+      "grad_norm": 0.03173214942216873,
+      "learning_rate": 1.5647921760391198e-05,
+      "loss": 0.4677,
+      "step": 735
+    },
+    {
+      "epoch": 0.8589669181659896,
+      "grad_norm": 0.02613997459411621,
+      "learning_rate": 1.5036674816625918e-05,
+      "loss": 0.4488,
+      "step": 740
+    },
+    {
+      "epoch": 0.8647707486941382,
+      "grad_norm": 0.02250559628009796,
+      "learning_rate": 1.4425427872860637e-05,
+      "loss": 0.4363,
+      "step": 745
+    },
+    {
+      "epoch": 0.8705745792222868,
+      "grad_norm": 0.03180099278688431,
+      "learning_rate": 1.3814180929095355e-05,
+      "loss": 0.4411,
+      "step": 750
+    },
+    {
+      "epoch": 0.8763784097504352,
+      "grad_norm": 0.02956099808216095,
+      "learning_rate": 1.3202933985330074e-05,
+      "loss": 0.4147,
+      "step": 755
+    },
+    {
+      "epoch": 0.8821822402785838,
+      "grad_norm": 0.027272948995232582,
+      "learning_rate": 1.2591687041564792e-05,
+      "loss": 0.4444,
+      "step": 760
+    },
+    {
+      "epoch": 0.8879860708067324,
+      "grad_norm": 0.034130338579416275,
+      "learning_rate": 1.198044009779951e-05,
+      "loss": 0.5034,
+      "step": 765
+    },
+    {
+      "epoch": 0.893789901334881,
+      "grad_norm": 0.03460566699504852,
+      "learning_rate": 1.136919315403423e-05,
+      "loss": 0.439,
+      "step": 770
+    },
+    {
+      "epoch": 0.8995937318630296,
+      "grad_norm": 0.0311946552246809,
+      "learning_rate": 1.075794621026895e-05,
+      "loss": 0.4429,
+      "step": 775
+    },
+    {
+      "epoch": 0.9053975623911782,
+      "grad_norm": 0.03752993047237396,
+      "learning_rate": 1.0146699266503668e-05,
+      "loss": 0.4265,
+      "step": 780
+    },
+    {
+      "epoch": 0.9112013929193268,
+      "grad_norm": 0.026637699455022812,
+      "learning_rate": 9.535452322738387e-06,
+      "loss": 0.4489,
+      "step": 785
+    },
+    {
+      "epoch": 0.9170052234474754,
+      "grad_norm": 0.026025403290987015,
+      "learning_rate": 8.924205378973105e-06,
+      "loss": 0.4357,
+      "step": 790
+    },
+    {
+      "epoch": 0.922809053975624,
+      "grad_norm": 0.029870592057704926,
+      "learning_rate": 8.312958435207824e-06,
+      "loss": 0.4478,
+      "step": 795
+    },
+    {
+      "epoch": 0.9286128845037724,
+      "grad_norm": 0.031670063734054565,
+      "learning_rate": 7.701711491442542e-06,
+      "loss": 0.4348,
+      "step": 800
+    },
+    {
+      "epoch": 0.934416715031921,
+      "grad_norm": 0.032493527978658676,
+      "learning_rate": 7.090464547677262e-06,
+      "loss": 0.4051,
+      "step": 805
+    },
+    {
+      "epoch": 0.9402205455600696,
+      "grad_norm": 0.02890474908053875,
+      "learning_rate": 6.47921760391198e-06,
+      "loss": 0.4272,
+      "step": 810
+    },
+    {
+      "epoch": 0.9460243760882182,
+      "grad_norm": 0.02616616152226925,
+      "learning_rate": 5.8679706601467e-06,
+      "loss": 0.4687,
+      "step": 815
+    },
+    {
+      "epoch": 0.9518282066163668,
+      "grad_norm": 0.02924773097038269,
+      "learning_rate": 5.256723716381418e-06,
+      "loss": 0.439,
+      "step": 820
+    },
+    {
+      "epoch": 0.9576320371445154,
+      "grad_norm": 0.03177264705300331,
+      "learning_rate": 4.645476772616137e-06,
+      "loss": 0.4451,
+      "step": 825
+    },
+    {
+      "epoch": 0.963435867672664,
+      "grad_norm": 0.02940208837389946,
+      "learning_rate": 4.034229828850856e-06,
+      "loss": 0.4177,
+      "step": 830
+    },
+    {
+      "epoch": 0.9692396982008126,
+      "grad_norm": 0.028753064572811127,
+      "learning_rate": 3.4229828850855746e-06,
+      "loss": 0.447,
+      "step": 835
+    },
+    {
+      "epoch": 0.9750435287289612,
+      "grad_norm": 0.035569071769714355,
+      "learning_rate": 2.8117359413202936e-06,
+      "loss": 0.4451,
+      "step": 840
+    },
+    {
+      "epoch": 0.9808473592571096,
+      "grad_norm": 0.02479061856865883,
+      "learning_rate": 2.2004889975550126e-06,
+      "loss": 0.4484,
+      "step": 845
+    },
+    {
+      "epoch": 0.9866511897852582,
+      "grad_norm": 0.031199142336845398,
+      "learning_rate": 1.5892420537897311e-06,
+      "loss": 0.4575,
+      "step": 850
+    },
+    {
+      "epoch": 0.9924550203134068,
+      "grad_norm": 0.03143865615129471,
+      "learning_rate": 9.779951100244499e-07,
+      "loss": 0.409,
+      "step": 855
+    },
+    {
+      "epoch": 0.9982588508415554,
+      "grad_norm": 0.029200268909335136,
+      "learning_rate": 3.667481662591687e-07,
+      "loss": 0.4396,
+      "step": 860
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 862,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.192258616665068e+20,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=0/checkpoint-1736/README.md

@@ -0,0 +1,207 @@
+---
+base_model: /share/models/Qwen3-4B-Instruct-2507
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:/share/models/Qwen3-4B-Instruct-2507
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.17.1

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=0/checkpoint-1736/chat_template.jinja

@@ -0,0 +1,61 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=0_fadec/checkpoint-1760/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=0_fadec/checkpoint-1760/chat_template.jinja

@@ -0,0 +1,61 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=0_fadec/checkpoint-1760/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=0_falayer=25_from1_step2/checkpoint-921/README.md

@@ -0,0 +1,207 @@
+---
+base_model: /share/models/Qwen3-4B-Instruct-2507
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:/share/models/Qwen3-4B-Instruct-2507
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.17.1

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=0_falayer=25_from1_step2/checkpoint-921/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=10/checkpoint-1760/README.md

@@ -0,0 +1,207 @@
+---
+base_model: /share/models/Qwen3-4B-Instruct-2507
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:/share/models/Qwen3-4B-Instruct-2507
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.17.1

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=10/checkpoint-1760/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=10/checkpoint-1760/tokenizer_config.json

@@ -0,0 +1,239 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 1010000,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=100/checkpoint-1760/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=100_falayer=25_from1_step2_fadec/checkpoint-921/adapter_config.json

@@ -0,0 +1,38 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "/share/models/Qwen3-4B-Instruct-2507",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": true
+}

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=100_falayer=25_from1_step2_fadec/checkpoint-921/chat_template.jinja

@@ -0,0 +1,61 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=100_falayer=25_from1_step2_fadec/checkpoint-921/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=100_falayer=25_from1_step2_fadec/checkpoint-921/tokenizer_config.json

@@ -0,0 +1,239 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 1010000,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

Qwen3-4B-Instruct-2507-sft-fusang-swa=2k_sink=100_falayer=25_from1_step2_fadec/checkpoint-921/trainer_state.json

@@ -0,0 +1,1322 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 921,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.005432568246638598,
+      "grad_norm": 1.3015767335891724,
+      "learning_rate": 8.510638297872341e-06,
+      "loss": 0.5996,
+      "step": 5
+    },
+    {
+      "epoch": 0.010865136493277196,
+      "grad_norm": 0.4534055292606354,
+      "learning_rate": 1.9148936170212766e-05,
+      "loss": 0.6183,
+      "step": 10
+    },
+    {
+      "epoch": 0.016297704739915795,
+      "grad_norm": 0.35130345821380615,
+      "learning_rate": 2.9787234042553192e-05,
+      "loss": 0.5788,
+      "step": 15
+    },
+    {
+      "epoch": 0.021730272986554392,
+      "grad_norm": 0.33144572377204895,
+      "learning_rate": 4.0425531914893614e-05,
+      "loss": 0.5739,
+      "step": 20
+    },
+    {
+      "epoch": 0.027162841233192993,
+      "grad_norm": 0.3466748893260956,
+      "learning_rate": 5.1063829787234044e-05,
+      "loss": 0.5793,
+      "step": 25
+    },
+    {
+      "epoch": 0.03259540947983159,
+      "grad_norm": 0.3162538707256317,
+      "learning_rate": 6.170212765957447e-05,
+      "loss": 0.6077,
+      "step": 30
+    },
+    {
+      "epoch": 0.03802797772647019,
+      "grad_norm": 0.28290727734565735,
+      "learning_rate": 7.23404255319149e-05,
+      "loss": 0.5945,
+      "step": 35
+    },
+    {
+      "epoch": 0.043460545973108784,
+      "grad_norm": 0.29814088344573975,
+      "learning_rate": 8.297872340425533e-05,
+      "loss": 0.5644,
+      "step": 40
+    },
+    {
+      "epoch": 0.04889311421974739,
+      "grad_norm": 0.3688049912452698,
+      "learning_rate": 9.361702127659576e-05,
+      "loss": 0.5674,
+      "step": 45
+    },
+    {
+      "epoch": 0.054325682466385986,
+      "grad_norm": 0.25144508481025696,
+      "learning_rate": 9.977116704805493e-05,
+      "loss": 0.5698,
+      "step": 50
+    },
+    {
+      "epoch": 0.05975825071302458,
+      "grad_norm": 0.306525856256485,
+      "learning_rate": 9.919908466819223e-05,
+      "loss": 0.5734,
+      "step": 55
+    },
+    {
+      "epoch": 0.06519081895966318,
+      "grad_norm": 0.24143728613853455,
+      "learning_rate": 9.862700228832952e-05,
+      "loss": 0.6082,
+      "step": 60
+    },
+    {
+      "epoch": 0.07062338720630178,
+      "grad_norm": 0.25603583455085754,
+      "learning_rate": 9.805491990846683e-05,
+      "loss": 0.5743,
+      "step": 65
+    },
+    {
+      "epoch": 0.07605595545294037,
+      "grad_norm": 0.3284883499145508,
+      "learning_rate": 9.748283752860413e-05,
+      "loss": 0.5715,
+      "step": 70
+    },
+    {
+      "epoch": 0.08148852369957897,
+      "grad_norm": 0.2864092290401459,
+      "learning_rate": 9.691075514874142e-05,
+      "loss": 0.5864,
+      "step": 75
+    },
+    {
+      "epoch": 0.08692109194621757,
+      "grad_norm": 0.24696052074432373,
+      "learning_rate": 9.633867276887872e-05,
+      "loss": 0.6029,
+      "step": 80
+    },
+    {
+      "epoch": 0.09235366019285618,
+      "grad_norm": 0.2639234960079193,
+      "learning_rate": 9.576659038901601e-05,
+      "loss": 0.5862,
+      "step": 85
+    },
+    {
+      "epoch": 0.09778622843949478,
+      "grad_norm": 0.247847318649292,
+      "learning_rate": 9.519450800915332e-05,
+      "loss": 0.574,
+      "step": 90
+    },
+    {
+      "epoch": 0.10321879668613337,
+      "grad_norm": 0.36300021409988403,
+      "learning_rate": 9.462242562929062e-05,
+      "loss": 0.5561,
+      "step": 95
+    },
+    {
+      "epoch": 0.10865136493277197,
+      "grad_norm": 0.2706812620162964,
+      "learning_rate": 9.405034324942791e-05,
+      "loss": 0.5661,
+      "step": 100
+    },
+    {
+      "epoch": 0.11408393317941057,
+      "grad_norm": 0.2586255669593811,
+      "learning_rate": 9.347826086956522e-05,
+      "loss": 0.5926,
+      "step": 105
+    },
+    {
+      "epoch": 0.11951650142604917,
+      "grad_norm": 0.2426815778017044,
+      "learning_rate": 9.290617848970252e-05,
+      "loss": 0.6078,
+      "step": 110
+    },
+    {
+      "epoch": 0.12494906967268776,
+      "grad_norm": 0.2470799833536148,
+      "learning_rate": 9.233409610983981e-05,
+      "loss": 0.576,
+      "step": 115
+    },
+    {
+      "epoch": 0.13038163791932636,
+      "grad_norm": 0.27425751090049744,
+      "learning_rate": 9.176201372997712e-05,
+      "loss": 0.5438,
+      "step": 120
+    },
+    {
+      "epoch": 0.13581420616596496,
+      "grad_norm": 0.26391464471817017,
+      "learning_rate": 9.118993135011442e-05,
+      "loss": 0.5629,
+      "step": 125
+    },
+    {
+      "epoch": 0.14124677441260355,
+      "grad_norm": 0.21842467784881592,
+      "learning_rate": 9.061784897025171e-05,
+      "loss": 0.5674,
+      "step": 130
+    },
+    {
+      "epoch": 0.14667934265924215,
+      "grad_norm": 0.24267511069774628,
+      "learning_rate": 9.004576659038902e-05,
+      "loss": 0.565,
+      "step": 135
+    },
+    {
+      "epoch": 0.15211191090588075,
+      "grad_norm": 0.2140224426984787,
+      "learning_rate": 8.947368421052632e-05,
+      "loss": 0.569,
+      "step": 140
+    },
+    {
+      "epoch": 0.15754447915251935,
+      "grad_norm": 0.21704743802547455,
+      "learning_rate": 8.890160183066361e-05,
+      "loss": 0.577,
+      "step": 145
+    },
+    {
+      "epoch": 0.16297704739915794,
+      "grad_norm": 0.26025763154029846,
+      "learning_rate": 8.832951945080092e-05,
+      "loss": 0.5639,
+      "step": 150
+    },
+    {
+      "epoch": 0.16840961564579654,
+      "grad_norm": 0.250322163105011,
+      "learning_rate": 8.775743707093822e-05,
+      "loss": 0.5774,
+      "step": 155
+    },
+    {
+      "epoch": 0.17384218389243514,
+      "grad_norm": 0.2071513682603836,
+      "learning_rate": 8.718535469107551e-05,
+      "loss": 0.5862,
+      "step": 160
+    },
+    {
+      "epoch": 0.17927475213907373,
+      "grad_norm": 0.23947393894195557,
+      "learning_rate": 8.661327231121282e-05,
+      "loss": 0.5873,
+      "step": 165
+    },
+    {
+      "epoch": 0.18470732038571236,
+      "grad_norm": 0.402304083108902,
+      "learning_rate": 8.604118993135012e-05,
+      "loss": 0.57,
+      "step": 170
+    },
+    {
+      "epoch": 0.19013988863235096,
+      "grad_norm": 0.22468920052051544,
+      "learning_rate": 8.546910755148741e-05,
+      "loss": 0.5615,
+      "step": 175
+    },
+    {
+      "epoch": 0.19557245687898955,
+      "grad_norm": 0.2339026927947998,
+      "learning_rate": 8.489702517162472e-05,
+      "loss": 0.5932,
+      "step": 180
+    },
+    {
+      "epoch": 0.20100502512562815,
+      "grad_norm": 0.2048051357269287,
+      "learning_rate": 8.432494279176202e-05,
+      "loss": 0.585,
+      "step": 185
+    },
+    {
+      "epoch": 0.20643759337226675,
+      "grad_norm": 0.21809840202331543,
+      "learning_rate": 8.375286041189931e-05,
+      "loss": 0.5817,
+      "step": 190
+    },
+    {
+      "epoch": 0.21187016161890534,
+      "grad_norm": 0.21819952130317688,
+      "learning_rate": 8.318077803203662e-05,
+      "loss": 0.5918,
+      "step": 195
+    },
+    {
+      "epoch": 0.21730272986554394,
+      "grad_norm": 0.2216586023569107,
+      "learning_rate": 8.260869565217392e-05,
+      "loss": 0.5666,
+      "step": 200
+    },
+    {
+      "epoch": 0.22273529811218254,
+      "grad_norm": 0.24591778218746185,
+      "learning_rate": 8.203661327231121e-05,
+      "loss": 0.6162,
+      "step": 205
+    },
+    {
+      "epoch": 0.22816786635882114,
+      "grad_norm": 0.2052900642156601,
+      "learning_rate": 8.146453089244852e-05,
+      "loss": 0.5862,
+      "step": 210
+    },
+    {
+      "epoch": 0.23360043460545973,
+      "grad_norm": 0.2240307331085205,
+      "learning_rate": 8.089244851258582e-05,
+      "loss": 0.5755,
+      "step": 215
+    },
+    {
+      "epoch": 0.23903300285209833,
+      "grad_norm": 0.2725607752799988,
+      "learning_rate": 8.032036613272312e-05,
+      "loss": 0.5511,
+      "step": 220
+    },
+    {
+      "epoch": 0.24446557109873693,
+      "grad_norm": 0.22214853763580322,
+      "learning_rate": 7.974828375286042e-05,
+      "loss": 0.5819,
+      "step": 225
+    },
+    {
+      "epoch": 0.24989813934537553,
+      "grad_norm": 0.23538362979888916,
+      "learning_rate": 7.917620137299772e-05,
+      "loss": 0.5759,
+      "step": 230
+    },
+    {
+      "epoch": 0.25533070759201415,
+      "grad_norm": 0.2038874477148056,
+      "learning_rate": 7.860411899313502e-05,
+      "loss": 0.5989,
+      "step": 235
+    },
+    {
+      "epoch": 0.2607632758386527,
+      "grad_norm": 0.20052260160446167,
+      "learning_rate": 7.803203661327232e-05,
+      "loss": 0.5763,
+      "step": 240
+    },
+    {
+      "epoch": 0.26619584408529134,
+      "grad_norm": 0.24630379676818848,
+      "learning_rate": 7.745995423340962e-05,
+      "loss": 0.573,
+      "step": 245
+    },
+    {
+      "epoch": 0.2716284123319299,
+      "grad_norm": 0.22534747421741486,
+      "learning_rate": 7.688787185354692e-05,
+      "loss": 0.5482,
+      "step": 250
+    },
+    {
+      "epoch": 0.27706098057856854,
+      "grad_norm": 0.21359695494174957,
+      "learning_rate": 7.631578947368422e-05,
+      "loss": 0.5826,
+      "step": 255
+    },
+    {
+      "epoch": 0.2824935488252071,
+      "grad_norm": 0.2673405110836029,
+      "learning_rate": 7.574370709382152e-05,
+      "loss": 0.6025,
+      "step": 260
+    },
+    {
+      "epoch": 0.28792611707184573,
+      "grad_norm": 0.21792642772197723,
+      "learning_rate": 7.517162471395882e-05,
+      "loss": 0.5595,
+      "step": 265
+    },
+    {
+      "epoch": 0.2933586853184843,
+      "grad_norm": 0.2374526411294937,
+      "learning_rate": 7.459954233409611e-05,
+      "loss": 0.5682,
+      "step": 270
+    },
+    {
+      "epoch": 0.2987912535651229,
+      "grad_norm": 0.20959119498729706,
+      "learning_rate": 7.402745995423342e-05,
+      "loss": 0.5561,
+      "step": 275
+    },
+    {
+      "epoch": 0.3042238218117615,
+      "grad_norm": 0.22747580707073212,
+      "learning_rate": 7.345537757437072e-05,
+      "loss": 0.5455,
+      "step": 280
+    },
+    {
+      "epoch": 0.3096563900584001,
+      "grad_norm": 0.22923846542835236,
+      "learning_rate": 7.288329519450801e-05,
+      "loss": 0.5669,
+      "step": 285
+    },
+    {
+      "epoch": 0.3150889583050387,
+      "grad_norm": 0.23467932641506195,
+      "learning_rate": 7.231121281464531e-05,
+      "loss": 0.5674,
+      "step": 290
+    },
+    {
+      "epoch": 0.3205215265516773,
+      "grad_norm": 0.24083422124385834,
+      "learning_rate": 7.17391304347826e-05,
+      "loss": 0.5652,
+      "step": 295
+    },
+    {
+      "epoch": 0.3259540947983159,
+      "grad_norm": 0.2186257690191269,
+      "learning_rate": 7.116704805491991e-05,
+      "loss": 0.5774,
+      "step": 300
+    },
+    {
+      "epoch": 0.3313866630449545,
+      "grad_norm": 0.22179163992404938,
+      "learning_rate": 7.059496567505721e-05,
+      "loss": 0.558,
+      "step": 305
+    },
+    {
+      "epoch": 0.3368192312915931,
+      "grad_norm": 0.37958696484565735,
+      "learning_rate": 7.00228832951945e-05,
+      "loss": 0.5844,
+      "step": 310
+    },
+    {
+      "epoch": 0.3422517995382317,
+      "grad_norm": 0.2163124531507492,
+      "learning_rate": 6.945080091533181e-05,
+      "loss": 0.5832,
+      "step": 315
+    },
+    {
+      "epoch": 0.3476843677848703,
+      "grad_norm": 0.22177410125732422,
+      "learning_rate": 6.887871853546911e-05,
+      "loss": 0.5828,
+      "step": 320
+    },
+    {
+      "epoch": 0.3531169360315089,
+      "grad_norm": 0.2047460824251175,
+      "learning_rate": 6.83066361556064e-05,
+      "loss": 0.5527,
+      "step": 325
+    },
+    {
+      "epoch": 0.35854950427814747,
+      "grad_norm": 0.22761861979961395,
+      "learning_rate": 6.773455377574371e-05,
+      "loss": 0.5769,
+      "step": 330
+    },
+    {
+      "epoch": 0.3639820725247861,
+      "grad_norm": 0.2984655201435089,
+      "learning_rate": 6.716247139588101e-05,
+      "loss": 0.5655,
+      "step": 335
+    },
+    {
+      "epoch": 0.3694146407714247,
+      "grad_norm": 0.2038339525461197,
+      "learning_rate": 6.65903890160183e-05,
+      "loss": 0.572,
+      "step": 340
+    },
+    {
+      "epoch": 0.3748472090180633,
+      "grad_norm": 0.19332076609134674,
+      "learning_rate": 6.601830663615561e-05,
+      "loss": 0.5915,
+      "step": 345
+    },
+    {
+      "epoch": 0.3802797772647019,
+      "grad_norm": 0.2224508821964264,
+      "learning_rate": 6.544622425629291e-05,
+      "loss": 0.5746,
+      "step": 350
+    },
+    {
+      "epoch": 0.3857123455113405,
+      "grad_norm": 0.20780149102210999,
+      "learning_rate": 6.48741418764302e-05,
+      "loss": 0.5887,
+      "step": 355
+    },
+    {
+      "epoch": 0.3911449137579791,
+      "grad_norm": 0.6229786276817322,
+      "learning_rate": 6.430205949656751e-05,
+      "loss": 0.5847,
+      "step": 360
+    },
+    {
+      "epoch": 0.3965774820046177,
+      "grad_norm": 0.19432781636714935,
+      "learning_rate": 6.372997711670481e-05,
+      "loss": 0.5845,
+      "step": 365
+    },
+    {
+      "epoch": 0.4020100502512563,
+      "grad_norm": 0.26033878326416016,
+      "learning_rate": 6.31578947368421e-05,
+      "loss": 0.5398,
+      "step": 370
+    },
+    {
+      "epoch": 0.40744261849789487,
+      "grad_norm": 0.1879839450120926,
+      "learning_rate": 6.258581235697941e-05,
+      "loss": 0.5576,
+      "step": 375
+    },
+    {
+      "epoch": 0.4128751867445335,
+      "grad_norm": 0.1986129730939865,
+      "learning_rate": 6.201372997711671e-05,
+      "loss": 0.567,
+      "step": 380
+    },
+    {
+      "epoch": 0.41830775499117206,
+      "grad_norm": 0.23794332146644592,
+      "learning_rate": 6.1441647597254e-05,
+      "loss": 0.5786,
+      "step": 385
+    },
+    {
+      "epoch": 0.4237403232378107,
+      "grad_norm": 0.17640449106693268,
+      "learning_rate": 6.086956521739131e-05,
+      "loss": 0.5489,
+      "step": 390
+    },
+    {
+      "epoch": 0.42917289148444926,
+      "grad_norm": 0.19154198467731476,
+      "learning_rate": 6.029748283752861e-05,
+      "loss": 0.5565,
+      "step": 395
+    },
+    {
+      "epoch": 0.4346054597310879,
+      "grad_norm": 0.21621261537075043,
+      "learning_rate": 5.9725400457665904e-05,
+      "loss": 0.5452,
+      "step": 400
+    },
+    {
+      "epoch": 0.44003802797772645,
+      "grad_norm": 0.20776431262493134,
+      "learning_rate": 5.915331807780321e-05,
+      "loss": 0.5642,
+      "step": 405
+    },
+    {
+      "epoch": 0.4454705962243651,
+      "grad_norm": 0.21817751228809357,
+      "learning_rate": 5.858123569794051e-05,
+      "loss": 0.5927,
+      "step": 410
+    },
+    {
+      "epoch": 0.45090316447100365,
+      "grad_norm": 0.22786065936088562,
+      "learning_rate": 5.8009153318077804e-05,
+      "loss": 0.5689,
+      "step": 415
+    },
+    {
+      "epoch": 0.4563357327176423,
+      "grad_norm": 0.21688605844974518,
+      "learning_rate": 5.743707093821511e-05,
+      "loss": 0.5644,
+      "step": 420
+    },
+    {
+      "epoch": 0.46176830096428084,
+      "grad_norm": 0.21741735935211182,
+      "learning_rate": 5.686498855835241e-05,
+      "loss": 0.5559,
+      "step": 425
+    },
+    {
+      "epoch": 0.46720086921091947,
+      "grad_norm": 0.26951777935028076,
+      "learning_rate": 5.6292906178489704e-05,
+      "loss": 0.5748,
+      "step": 430
+    },
+    {
+      "epoch": 0.47263343745755804,
+      "grad_norm": 0.21070964634418488,
+      "learning_rate": 5.5720823798627006e-05,
+      "loss": 0.5717,
+      "step": 435
+    },
+    {
+      "epoch": 0.47806600570419666,
+      "grad_norm": 0.23886334896087646,
+      "learning_rate": 5.51487414187643e-05,
+      "loss": 0.584,
+      "step": 440
+    },
+    {
+      "epoch": 0.48349857395083523,
+      "grad_norm": 0.21350622177124023,
+      "learning_rate": 5.4576659038901604e-05,
+      "loss": 0.5778,
+      "step": 445
+    },
+    {
+      "epoch": 0.48893114219747386,
+      "grad_norm": 0.1953973025083542,
+      "learning_rate": 5.4004576659038907e-05,
+      "loss": 0.5749,
+      "step": 450
+    },
+    {
+      "epoch": 0.4943637104441125,
+      "grad_norm": 0.2156413346529007,
+      "learning_rate": 5.34324942791762e-05,
+      "loss": 0.55,
+      "step": 455
+    },
+    {
+      "epoch": 0.49979627869075105,
+      "grad_norm": 0.22743190824985504,
+      "learning_rate": 5.28604118993135e-05,
+      "loss": 0.5991,
+      "step": 460
+    },
+    {
+      "epoch": 0.5052288469373897,
+      "grad_norm": 0.20552119612693787,
+      "learning_rate": 5.228832951945081e-05,
+      "loss": 0.5871,
+      "step": 465
+    },
+    {
+      "epoch": 0.5106614151840283,
+      "grad_norm": 0.29377639293670654,
+      "learning_rate": 5.17162471395881e-05,
+      "loss": 0.555,
+      "step": 470
+    },
+    {
+      "epoch": 0.5160939834306668,
+      "grad_norm": 0.1877746433019638,
+      "learning_rate": 5.11441647597254e-05,
+      "loss": 0.5846,
+      "step": 475
+    },
+    {
+      "epoch": 0.5215265516773054,
+      "grad_norm": 0.23544102907180786,
+      "learning_rate": 5.057208237986271e-05,
+      "loss": 0.5806,
+      "step": 480
+    },
+    {
+      "epoch": 0.5269591199239441,
+      "grad_norm": 0.2117416262626648,
+      "learning_rate": 5e-05,
+      "loss": 0.5842,
+      "step": 485
+    },
+    {
+      "epoch": 0.5323916881705827,
+      "grad_norm": 0.21826615929603577,
+      "learning_rate": 4.9427917620137305e-05,
+      "loss": 0.5613,
+      "step": 490
+    },
+    {
+      "epoch": 0.5378242564172212,
+      "grad_norm": 0.20202744007110596,
+      "learning_rate": 4.88558352402746e-05,
+      "loss": 0.5699,
+      "step": 495
+    },
+    {
+      "epoch": 0.5432568246638598,
+      "grad_norm": 0.21085906028747559,
+      "learning_rate": 4.82837528604119e-05,
+      "loss": 0.5666,
+      "step": 500
+    },
+    {
+      "epoch": 0.5486893929104985,
+      "grad_norm": 0.1997220665216446,
+      "learning_rate": 4.7711670480549205e-05,
+      "loss": 0.5855,
+      "step": 505
+    },
+    {
+      "epoch": 0.5541219611571371,
+      "grad_norm": 0.19225606322288513,
+      "learning_rate": 4.71395881006865e-05,
+      "loss": 0.6002,
+      "step": 510
+    },
+    {
+      "epoch": 0.5595545294037756,
+      "grad_norm": 0.2273896485567093,
+      "learning_rate": 4.65675057208238e-05,
+      "loss": 0.582,
+      "step": 515
+    },
+    {
+      "epoch": 0.5649870976504142,
+      "grad_norm": 0.2283266931772232,
+      "learning_rate": 4.5995423340961105e-05,
+      "loss": 0.5724,
+      "step": 520
+    },
+    {
+      "epoch": 0.5704196658970528,
+      "grad_norm": 0.203383207321167,
+      "learning_rate": 4.54233409610984e-05,
+      "loss": 0.5323,
+      "step": 525
+    },
+    {
+      "epoch": 0.5758522341436915,
+      "grad_norm": 0.21528829634189606,
+      "learning_rate": 4.48512585812357e-05,
+      "loss": 0.5758,
+      "step": 530
+    },
+    {
+      "epoch": 0.58128480239033,
+      "grad_norm": 0.21587879955768585,
+      "learning_rate": 4.4279176201373e-05,
+      "loss": 0.5772,
+      "step": 535
+    },
+    {
+      "epoch": 0.5867173706369686,
+      "grad_norm": 0.223943829536438,
+      "learning_rate": 4.37070938215103e-05,
+      "loss": 0.5503,
+      "step": 540
+    },
+    {
+      "epoch": 0.5921499388836072,
+      "grad_norm": 0.25736063718795776,
+      "learning_rate": 4.3135011441647596e-05,
+      "loss": 0.5818,
+      "step": 545
+    },
+    {
+      "epoch": 0.5975825071302459,
+      "grad_norm": 0.18794198334217072,
+      "learning_rate": 4.25629290617849e-05,
+      "loss": 0.5648,
+      "step": 550
+    },
+    {
+      "epoch": 0.6030150753768844,
+      "grad_norm": 0.20583099126815796,
+      "learning_rate": 4.1990846681922194e-05,
+      "loss": 0.5799,
+      "step": 555
+    },
+    {
+      "epoch": 0.608447643623523,
+      "grad_norm": 0.19779925048351288,
+      "learning_rate": 4.1418764302059497e-05,
+      "loss": 0.5782,
+      "step": 560
+    },
+    {
+      "epoch": 0.6138802118701616,
+      "grad_norm": 0.20509599149227142,
+      "learning_rate": 4.08466819221968e-05,
+      "loss": 0.5729,
+      "step": 565
+    },
+    {
+      "epoch": 0.6193127801168002,
+      "grad_norm": 0.2053207904100418,
+      "learning_rate": 4.0274599542334094e-05,
+      "loss": 0.5672,
+      "step": 570
+    },
+    {
+      "epoch": 0.6247453483634389,
+      "grad_norm": 0.17894507944583893,
+      "learning_rate": 3.97025171624714e-05,
+      "loss": 0.5459,
+      "step": 575
+    },
+    {
+      "epoch": 0.6301779166100774,
+      "grad_norm": 0.2054516226053238,
+      "learning_rate": 3.91304347826087e-05,
+      "loss": 0.5941,
+      "step": 580
+    },
+    {
+      "epoch": 0.635610484856716,
+      "grad_norm": 0.2139163464307785,
+      "learning_rate": 3.8558352402745995e-05,
+      "loss": 0.5897,
+      "step": 585
+    },
+    {
+      "epoch": 0.6410430531033546,
+      "grad_norm": 0.18362848460674286,
+      "learning_rate": 3.79862700228833e-05,
+      "loss": 0.5755,
+      "step": 590
+    },
+    {
+      "epoch": 0.6464756213499933,
+      "grad_norm": 0.25813937187194824,
+      "learning_rate": 3.74141876430206e-05,
+      "loss": 0.5657,
+      "step": 595
+    },
+    {
+      "epoch": 0.6519081895966318,
+      "grad_norm": 0.19707731902599335,
+      "learning_rate": 3.6842105263157895e-05,
+      "loss": 0.5798,
+      "step": 600
+    },
+    {
+      "epoch": 0.6573407578432704,
+      "grad_norm": 0.21467731893062592,
+      "learning_rate": 3.62700228832952e-05,
+      "loss": 0.5607,
+      "step": 605
+    },
+    {
+      "epoch": 0.662773326089909,
+      "grad_norm": 0.18817900121212006,
+      "learning_rate": 3.56979405034325e-05,
+      "loss": 0.6053,
+      "step": 610
+    },
+    {
+      "epoch": 0.6682058943365476,
+      "grad_norm": 0.1965055763721466,
+      "learning_rate": 3.5125858123569795e-05,
+      "loss": 0.5272,
+      "step": 615
+    },
+    {
+      "epoch": 0.6736384625831862,
+      "grad_norm": 0.23772867023944855,
+      "learning_rate": 3.45537757437071e-05,
+      "loss": 0.558,
+      "step": 620
+    },
+    {
+      "epoch": 0.6790710308298248,
+      "grad_norm": 0.19465915858745575,
+      "learning_rate": 3.39816933638444e-05,
+      "loss": 0.5764,
+      "step": 625
+    },
+    {
+      "epoch": 0.6845035990764634,
+      "grad_norm": 0.18953974545001984,
+      "learning_rate": 3.3409610983981695e-05,
+      "loss": 0.5839,
+      "step": 630
+    },
+    {
+      "epoch": 0.689936167323102,
+      "grad_norm": 0.3488953113555908,
+      "learning_rate": 3.2837528604119e-05,
+      "loss": 0.5677,
+      "step": 635
+    },
+    {
+      "epoch": 0.6953687355697405,
+      "grad_norm": 0.20315735042095184,
+      "learning_rate": 3.226544622425629e-05,
+      "loss": 0.5473,
+      "step": 640
+    },
+    {
+      "epoch": 0.7008013038163792,
+      "grad_norm": 0.1986822485923767,
+      "learning_rate": 3.1693363844393595e-05,
+      "loss": 0.5698,
+      "step": 645
+    },
+    {
+      "epoch": 0.7062338720630178,
+      "grad_norm": 0.1991867870092392,
+      "learning_rate": 3.112128146453089e-05,
+      "loss": 0.5637,
+      "step": 650
+    },
+    {
+      "epoch": 0.7116664403096564,
+      "grad_norm": 0.2192126363515854,
+      "learning_rate": 3.054919908466819e-05,
+      "loss": 0.5789,
+      "step": 655
+    },
+    {
+      "epoch": 0.7170990085562949,
+      "grad_norm": 0.20494726300239563,
+      "learning_rate": 2.9977116704805492e-05,
+      "loss": 0.5804,
+      "step": 660
+    },
+    {
+      "epoch": 0.7225315768029336,
+      "grad_norm": 0.28420111536979675,
+      "learning_rate": 2.9405034324942794e-05,
+      "loss": 0.5633,
+      "step": 665
+    },
+    {
+      "epoch": 0.7279641450495722,
+      "grad_norm": 0.19433985650539398,
+      "learning_rate": 2.8832951945080093e-05,
+      "loss": 0.5698,
+      "step": 670
+    },
+    {
+      "epoch": 0.7333967132962108,
+      "grad_norm": 0.19712510704994202,
+      "learning_rate": 2.826086956521739e-05,
+      "loss": 0.557,
+      "step": 675
+    },
+    {
+      "epoch": 0.7388292815428494,
+      "grad_norm": 0.2100038081407547,
+      "learning_rate": 2.768878718535469e-05,
+      "loss": 0.5807,
+      "step": 680
+    },
+    {
+      "epoch": 0.744261849789488,
+      "grad_norm": 0.29004234075546265,
+      "learning_rate": 2.7116704805491993e-05,
+      "loss": 0.5462,
+      "step": 685
+    },
+    {
+      "epoch": 0.7496944180361266,
+      "grad_norm": 0.18994742631912231,
+      "learning_rate": 2.654462242562929e-05,
+      "loss": 0.5937,
+      "step": 690
+    },
+    {
+      "epoch": 0.7551269862827652,
+      "grad_norm": 0.19477994740009308,
+      "learning_rate": 2.597254004576659e-05,
+      "loss": 0.5785,
+      "step": 695
+    },
+    {
+      "epoch": 0.7605595545294038,
+      "grad_norm": 0.20193684101104736,
+      "learning_rate": 2.5400457665903894e-05,
+      "loss": 0.5836,
+      "step": 700
+    },
+    {
+      "epoch": 0.7659921227760423,
+      "grad_norm": 0.22384333610534668,
+      "learning_rate": 2.4828375286041193e-05,
+      "loss": 0.5765,
+      "step": 705
+    },
+    {
+      "epoch": 0.771424691022681,
+      "grad_norm": 0.17794522643089294,
+      "learning_rate": 2.425629290617849e-05,
+      "loss": 0.5839,
+      "step": 710
+    },
+    {
+      "epoch": 0.7768572592693196,
+      "grad_norm": 0.21384797990322113,
+      "learning_rate": 2.368421052631579e-05,
+      "loss": 0.6081,
+      "step": 715
+    },
+    {
+      "epoch": 0.7822898275159582,
+      "grad_norm": 0.1895979642868042,
+      "learning_rate": 2.311212814645309e-05,
+      "loss": 0.5817,
+      "step": 720
+    },
+    {
+      "epoch": 0.7877223957625967,
+      "grad_norm": 0.1993294209241867,
+      "learning_rate": 2.2540045766590388e-05,
+      "loss": 0.567,
+      "step": 725
+    },
+    {
+      "epoch": 0.7931549640092354,
+      "grad_norm": 0.1895848959684372,
+      "learning_rate": 2.1967963386727687e-05,
+      "loss": 0.5772,
+      "step": 730
+    },
+    {
+      "epoch": 0.798587532255874,
+      "grad_norm": 0.2276710420846939,
+      "learning_rate": 2.139588100686499e-05,
+      "loss": 0.549,
+      "step": 735
+    },
+    {
+      "epoch": 0.8040201005025126,
+      "grad_norm": 0.20059236884117126,
+      "learning_rate": 2.082379862700229e-05,
+      "loss": 0.5411,
+      "step": 740
+    },
+    {
+      "epoch": 0.8094526687491511,
+      "grad_norm": 0.2584899067878723,
+      "learning_rate": 2.0251716247139587e-05,
+      "loss": 0.5706,
+      "step": 745
+    },
+    {
+      "epoch": 0.8148852369957897,
+      "grad_norm": 0.19888870418071747,
+      "learning_rate": 1.967963386727689e-05,
+      "loss": 0.5652,
+      "step": 750
+    },
+    {
+      "epoch": 0.8203178052424284,
+      "grad_norm": 0.23057852685451508,
+      "learning_rate": 1.910755148741419e-05,
+      "loss": 0.5925,
+      "step": 755
+    },
+    {
+      "epoch": 0.825750373489067,
+      "grad_norm": 0.23049262166023254,
+      "learning_rate": 1.853546910755149e-05,
+      "loss": 0.5722,
+      "step": 760
+    },
+    {
+      "epoch": 0.8311829417357055,
+      "grad_norm": 0.20783977210521698,
+      "learning_rate": 1.796338672768879e-05,
+      "loss": 0.572,
+      "step": 765
+    },
+    {
+      "epoch": 0.8366155099823441,
+      "grad_norm": 0.2562456727027893,
+      "learning_rate": 1.739130434782609e-05,
+      "loss": 0.5454,
+      "step": 770
+    },
+    {
+      "epoch": 0.8420480782289828,
+      "grad_norm": 0.21120162308216095,
+      "learning_rate": 1.6819221967963388e-05,
+      "loss": 0.5904,
+      "step": 775
+    },
+    {
+      "epoch": 0.8474806464756214,
+      "grad_norm": 0.21850056946277618,
+      "learning_rate": 1.6247139588100687e-05,
+      "loss": 0.5993,
+      "step": 780
+    },
+    {
+      "epoch": 0.8529132147222599,
+      "grad_norm": 0.2106572985649109,
+      "learning_rate": 1.5675057208237986e-05,
+      "loss": 0.606,
+      "step": 785
+    },
+    {
+      "epoch": 0.8583457829688985,
+      "grad_norm": 0.1892135888338089,
+      "learning_rate": 1.5102974828375288e-05,
+      "loss": 0.5748,
+      "step": 790
+    },
+    {
+      "epoch": 0.8637783512155371,
+      "grad_norm": 0.2126905769109726,
+      "learning_rate": 1.4530892448512587e-05,
+      "loss": 0.5704,
+      "step": 795
+    },
+    {
+      "epoch": 0.8692109194621758,
+      "grad_norm": 0.22039473056793213,
+      "learning_rate": 1.3958810068649886e-05,
+      "loss": 0.5759,
+      "step": 800
+    },
+    {
+      "epoch": 0.8746434877088144,
+      "grad_norm": 0.21076731383800507,
+      "learning_rate": 1.3386727688787188e-05,
+      "loss": 0.5937,
+      "step": 805
+    },
+    {
+      "epoch": 0.8800760559554529,
+      "grad_norm": 0.17734327912330627,
+      "learning_rate": 1.2814645308924485e-05,
+      "loss": 0.5809,
+      "step": 810
+    },
+    {
+      "epoch": 0.8855086242020915,
+      "grad_norm": 0.19097428023815155,
+      "learning_rate": 1.2242562929061784e-05,
+      "loss": 0.5598,
+      "step": 815
+    },
+    {
+      "epoch": 0.8909411924487302,
+      "grad_norm": 0.23482643067836761,
+      "learning_rate": 1.1670480549199085e-05,
+      "loss": 0.5682,
+      "step": 820
+    },
+    {
+      "epoch": 0.8963737606953688,
+      "grad_norm": 0.19752998650074005,
+      "learning_rate": 1.1098398169336386e-05,
+      "loss": 0.5689,
+      "step": 825
+    },
+    {
+      "epoch": 0.9018063289420073,
+      "grad_norm": 0.23479461669921875,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.5859,
+      "step": 830
+    },
+    {
+      "epoch": 0.9072388971886459,
+      "grad_norm": 0.21521364152431488,
+      "learning_rate": 9.954233409610985e-06,
+      "loss": 0.6074,
+      "step": 835
+    },
+    {
+      "epoch": 0.9126714654352845,
+      "grad_norm": 0.2059374749660492,
+      "learning_rate": 9.382151029748284e-06,
+      "loss": 0.583,
+      "step": 840
+    },
+    {
+      "epoch": 0.9181040336819232,
+      "grad_norm": 0.20290476083755493,
+      "learning_rate": 8.810068649885583e-06,
+      "loss": 0.5651,
+      "step": 845
+    },
+    {
+      "epoch": 0.9235366019285617,
+      "grad_norm": 0.18844471871852875,
+      "learning_rate": 8.237986270022884e-06,
+      "loss": 0.556,
+      "step": 850
+    },
+    {
+      "epoch": 0.9289691701752003,
+      "grad_norm": 0.18698588013648987,
+      "learning_rate": 7.665903890160184e-06,
+      "loss": 0.5899,
+      "step": 855
+    },
+    {
+      "epoch": 0.9344017384218389,
+      "grad_norm": 0.21699076890945435,
+      "learning_rate": 7.093821510297482e-06,
+      "loss": 0.6056,
+      "step": 860
+    },
+    {
+      "epoch": 0.9398343066684776,
+      "grad_norm": 0.19663618505001068,
+      "learning_rate": 6.521739130434783e-06,
+      "loss": 0.5898,
+      "step": 865
+    },
+    {
+      "epoch": 0.9452668749151161,
+      "grad_norm": 0.2180209904909134,
+      "learning_rate": 5.949656750572083e-06,
+      "loss": 0.557,
+      "step": 870
+    },
+    {
+      "epoch": 0.9506994431617547,
+      "grad_norm": 0.22569262981414795,
+      "learning_rate": 5.3775743707093824e-06,
+      "loss": 0.5516,
+      "step": 875
+    },
+    {
+      "epoch": 0.9561320114083933,
+      "grad_norm": 0.2216615080833435,
+      "learning_rate": 4.805491990846682e-06,
+      "loss": 0.6001,
+      "step": 880
+    },
+    {
+      "epoch": 0.961564579655032,
+      "grad_norm": 0.1988694965839386,
+      "learning_rate": 4.233409610983982e-06,
+      "loss": 0.5834,
+      "step": 885
+    },
+    {
+      "epoch": 0.9669971479016705,
+      "grad_norm": 0.19006194174289703,
+      "learning_rate": 3.6613272311212818e-06,
+      "loss": 0.5679,
+      "step": 890
+    },
+    {
+      "epoch": 0.9724297161483091,
+      "grad_norm": 0.23912011086940765,
+      "learning_rate": 3.0892448512585815e-06,
+      "loss": 0.5479,
+      "step": 895
+    },
+    {
+      "epoch": 0.9778622843949477,
+      "grad_norm": 0.18458198010921478,
+      "learning_rate": 2.517162471395881e-06,
+      "loss": 0.5643,
+      "step": 900
+    },
+    {
+      "epoch": 0.9832948526415863,
+      "grad_norm": 0.18322934210300446,
+      "learning_rate": 1.9450800915331807e-06,
+      "loss": 0.5439,
+      "step": 905
+    },
+    {
+      "epoch": 0.988727420888225,
+      "grad_norm": 0.24298790097236633,
+      "learning_rate": 1.3729977116704805e-06,
+      "loss": 0.5892,
+      "step": 910
+    },
+    {
+      "epoch": 0.9941599891348635,
+      "grad_norm": 0.20500127971172333,
+      "learning_rate": 8.009153318077803e-07,
+      "loss": 0.578,
+      "step": 915
+    },
+    {
+      "epoch": 0.9995925573815021,
+      "grad_norm": 0.2129577100276947,
+      "learning_rate": 2.288329519450801e-07,
+      "loss": 0.5471,
+      "step": 920
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 921,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.5239517919298716e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

Qwen3-4B-Instruct-2507-sft-fusang/checkpoint-1760/trainer_state.json

@@ -0,0 +1,2498 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 1760,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002841716396703609,
+      "grad_norm": 0.9752756357192993,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.6068,
+      "step": 5
+    },
+    {
+      "epoch": 0.005683432793407218,
+      "grad_norm": 0.5571355223655701,
+      "learning_rate": 1.0227272727272729e-05,
+      "loss": 0.5936,
+      "step": 10
+    },
+    {
+      "epoch": 0.008525149190110827,
+      "grad_norm": 0.3266292214393616,
+      "learning_rate": 1.590909090909091e-05,
+      "loss": 0.5728,
+      "step": 15
+    },
+    {
+      "epoch": 0.011366865586814436,
+      "grad_norm": 0.2658981680870056,
+      "learning_rate": 2.1590909090909093e-05,
+      "loss": 0.5402,
+      "step": 20
+    },
+    {
+      "epoch": 0.014208581983518044,
+      "grad_norm": 0.30417537689208984,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 0.5487,
+      "step": 25
+    },
+    {
+      "epoch": 0.017050298380221655,
+      "grad_norm": 0.26923367381095886,
+      "learning_rate": 3.295454545454545e-05,
+      "loss": 0.6084,
+      "step": 30
+    },
+    {
+      "epoch": 0.019892014776925263,
+      "grad_norm": 0.30384576320648193,
+      "learning_rate": 3.8636363636363636e-05,
+      "loss": 0.5957,
+      "step": 35
+    },
+    {
+      "epoch": 0.022733731173628872,
+      "grad_norm": 0.2878063917160034,
+      "learning_rate": 4.431818181818182e-05,
+      "loss": 0.5845,
+      "step": 40
+    },
+    {
+      "epoch": 0.02557544757033248,
+      "grad_norm": 0.25360503792762756,
+      "learning_rate": 5e-05,
+      "loss": 0.5963,
+      "step": 45
+    },
+    {
+      "epoch": 0.02841716396703609,
+      "grad_norm": 0.26214471459388733,
+      "learning_rate": 5.568181818181818e-05,
+      "loss": 0.5689,
+      "step": 50
+    },
+    {
+      "epoch": 0.0312588803637397,
+      "grad_norm": 0.3016664683818817,
+      "learning_rate": 6.136363636363636e-05,
+      "loss": 0.5885,
+      "step": 55
+    },
+    {
+      "epoch": 0.03410059676044331,
+      "grad_norm": 0.2853708267211914,
+      "learning_rate": 6.704545454545455e-05,
+      "loss": 0.5814,
+      "step": 60
+    },
+    {
+      "epoch": 0.03694231315714692,
+      "grad_norm": 0.2524615526199341,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 0.5719,
+      "step": 65
+    },
+    {
+      "epoch": 0.039784029553850526,
+      "grad_norm": 0.2268151193857193,
+      "learning_rate": 7.840909090909091e-05,
+      "loss": 0.5717,
+      "step": 70
+    },
+    {
+      "epoch": 0.042625745950554135,
+      "grad_norm": 0.2180832326412201,
+      "learning_rate": 8.40909090909091e-05,
+      "loss": 0.555,
+      "step": 75
+    },
+    {
+      "epoch": 0.045467462347257744,
+      "grad_norm": 0.23236437141895294,
+      "learning_rate": 8.977272727272728e-05,
+      "loss": 0.5594,
+      "step": 80
+    },
+    {
+      "epoch": 0.04830917874396135,
+      "grad_norm": 0.24800875782966614,
+      "learning_rate": 9.545454545454546e-05,
+      "loss": 0.5743,
+      "step": 85
+    },
+    {
+      "epoch": 0.05115089514066496,
+      "grad_norm": 0.1982068121433258,
+      "learning_rate": 9.994019138755981e-05,
+      "loss": 0.6041,
+      "step": 90
+    },
+    {
+      "epoch": 0.05399261153736857,
+      "grad_norm": 0.2268919050693512,
+      "learning_rate": 9.964114832535886e-05,
+      "loss": 0.5892,
+      "step": 95
+    },
+    {
+      "epoch": 0.05683432793407218,
+      "grad_norm": 0.22892409563064575,
+      "learning_rate": 9.93421052631579e-05,
+      "loss": 0.5772,
+      "step": 100
+    },
+    {
+      "epoch": 0.059676044330775786,
+      "grad_norm": 0.21798421442508698,
+      "learning_rate": 9.904306220095694e-05,
+      "loss": 0.5776,
+      "step": 105
+    },
+    {
+      "epoch": 0.0625177607274794,
+      "grad_norm": 0.20109155774116516,
+      "learning_rate": 9.874401913875598e-05,
+      "loss": 0.5717,
+      "step": 110
+    },
+    {
+      "epoch": 0.06535947712418301,
+      "grad_norm": 0.17350049316883087,
+      "learning_rate": 9.844497607655503e-05,
+      "loss": 0.5653,
+      "step": 115
+    },
+    {
+      "epoch": 0.06820119352088662,
+      "grad_norm": 0.20336377620697021,
+      "learning_rate": 9.814593301435408e-05,
+      "loss": 0.5998,
+      "step": 120
+    },
+    {
+      "epoch": 0.07104290991759023,
+      "grad_norm": 0.1792660355567932,
+      "learning_rate": 9.784688995215312e-05,
+      "loss": 0.5575,
+      "step": 125
+    },
+    {
+      "epoch": 0.07388462631429384,
+      "grad_norm": 0.2105855494737625,
+      "learning_rate": 9.754784688995216e-05,
+      "loss": 0.5898,
+      "step": 130
+    },
+    {
+      "epoch": 0.07672634271099744,
+      "grad_norm": 0.2220708131790161,
+      "learning_rate": 9.72488038277512e-05,
+      "loss": 0.5744,
+      "step": 135
+    },
+    {
+      "epoch": 0.07956805910770105,
+      "grad_norm": 0.20286299288272858,
+      "learning_rate": 9.694976076555024e-05,
+      "loss": 0.5712,
+      "step": 140
+    },
+    {
+      "epoch": 0.08240977550440466,
+      "grad_norm": 0.1932576298713684,
+      "learning_rate": 9.66507177033493e-05,
+      "loss": 0.5684,
+      "step": 145
+    },
+    {
+      "epoch": 0.08525149190110827,
+      "grad_norm": 0.21917824447155,
+      "learning_rate": 9.635167464114832e-05,
+      "loss": 0.5584,
+      "step": 150
+    },
+    {
+      "epoch": 0.08809320829781188,
+      "grad_norm": 0.21882839500904083,
+      "learning_rate": 9.605263157894737e-05,
+      "loss": 0.5508,
+      "step": 155
+    },
+    {
+      "epoch": 0.09093492469451549,
+      "grad_norm": 0.234402596950531,
+      "learning_rate": 9.575358851674642e-05,
+      "loss": 0.5544,
+      "step": 160
+    },
+    {
+      "epoch": 0.0937766410912191,
+      "grad_norm": 0.17900532484054565,
+      "learning_rate": 9.545454545454546e-05,
+      "loss": 0.5645,
+      "step": 165
+    },
+    {
+      "epoch": 0.0966183574879227,
+      "grad_norm": 0.1782204657793045,
+      "learning_rate": 9.51555023923445e-05,
+      "loss": 0.5739,
+      "step": 170
+    },
+    {
+      "epoch": 0.09946007388462631,
+      "grad_norm": 0.18758229911327362,
+      "learning_rate": 9.485645933014354e-05,
+      "loss": 0.5639,
+      "step": 175
+    },
+    {
+      "epoch": 0.10230179028132992,
+      "grad_norm": 0.2168463170528412,
+      "learning_rate": 9.455741626794258e-05,
+      "loss": 0.5877,
+      "step": 180
+    },
+    {
+      "epoch": 0.10514350667803353,
+      "grad_norm": 0.18619057536125183,
+      "learning_rate": 9.425837320574164e-05,
+      "loss": 0.5907,
+      "step": 185
+    },
+    {
+      "epoch": 0.10798522307473714,
+      "grad_norm": 0.19158032536506653,
+      "learning_rate": 9.395933014354067e-05,
+      "loss": 0.5591,
+      "step": 190
+    },
+    {
+      "epoch": 0.11082693947144075,
+      "grad_norm": 0.18429265916347504,
+      "learning_rate": 9.366028708133972e-05,
+      "loss": 0.5665,
+      "step": 195
+    },
+    {
+      "epoch": 0.11366865586814436,
+      "grad_norm": 0.17102861404418945,
+      "learning_rate": 9.336124401913876e-05,
+      "loss": 0.5824,
+      "step": 200
+    },
+    {
+      "epoch": 0.11651037226484796,
+      "grad_norm": 0.19570764899253845,
+      "learning_rate": 9.30622009569378e-05,
+      "loss": 0.5557,
+      "step": 205
+    },
+    {
+      "epoch": 0.11935208866155157,
+      "grad_norm": 0.18533465266227722,
+      "learning_rate": 9.276315789473686e-05,
+      "loss": 0.5746,
+      "step": 210
+    },
+    {
+      "epoch": 0.12219380505825518,
+      "grad_norm": 0.18167763948440552,
+      "learning_rate": 9.246411483253588e-05,
+      "loss": 0.5714,
+      "step": 215
+    },
+    {
+      "epoch": 0.1250355214549588,
+      "grad_norm": 0.2030726820230484,
+      "learning_rate": 9.216507177033494e-05,
+      "loss": 0.531,
+      "step": 220
+    },
+    {
+      "epoch": 0.1278772378516624,
+      "grad_norm": 0.18000352382659912,
+      "learning_rate": 9.186602870813398e-05,
+      "loss": 0.567,
+      "step": 225
+    },
+    {
+      "epoch": 0.13071895424836602,
+      "grad_norm": 0.191226065158844,
+      "learning_rate": 9.156698564593302e-05,
+      "loss": 0.535,
+      "step": 230
+    },
+    {
+      "epoch": 0.13356067064506963,
+      "grad_norm": 0.173014834523201,
+      "learning_rate": 9.126794258373206e-05,
+      "loss": 0.5825,
+      "step": 235
+    },
+    {
+      "epoch": 0.13640238704177324,
+      "grad_norm": 0.21560324728488922,
+      "learning_rate": 9.09688995215311e-05,
+      "loss": 0.5888,
+      "step": 240
+    },
+    {
+      "epoch": 0.13924410343847685,
+      "grad_norm": 0.21009966731071472,
+      "learning_rate": 9.066985645933016e-05,
+      "loss": 0.5284,
+      "step": 245
+    },
+    {
+      "epoch": 0.14208581983518045,
+      "grad_norm": 0.18440449237823486,
+      "learning_rate": 9.03708133971292e-05,
+      "loss": 0.5527,
+      "step": 250
+    },
+    {
+      "epoch": 0.14492753623188406,
+      "grad_norm": 0.2049122303724289,
+      "learning_rate": 9.007177033492822e-05,
+      "loss": 0.5805,
+      "step": 255
+    },
+    {
+      "epoch": 0.14776925262858767,
+      "grad_norm": 0.19917908310890198,
+      "learning_rate": 8.977272727272728e-05,
+      "loss": 0.5663,
+      "step": 260
+    },
+    {
+      "epoch": 0.15061096902529128,
+      "grad_norm": 0.16981367766857147,
+      "learning_rate": 8.947368421052632e-05,
+      "loss": 0.5969,
+      "step": 265
+    },
+    {
+      "epoch": 0.1534526854219949,
+      "grad_norm": 0.17967240512371063,
+      "learning_rate": 8.917464114832537e-05,
+      "loss": 0.5716,
+      "step": 270
+    },
+    {
+      "epoch": 0.1562944018186985,
+      "grad_norm": 0.19180747866630554,
+      "learning_rate": 8.88755980861244e-05,
+      "loss": 0.5788,
+      "step": 275
+    },
+    {
+      "epoch": 0.1591361182154021,
+      "grad_norm": 0.19337241351604462,
+      "learning_rate": 8.857655502392344e-05,
+      "loss": 0.5545,
+      "step": 280
+    },
+    {
+      "epoch": 0.16197783461210571,
+      "grad_norm": 0.2016487866640091,
+      "learning_rate": 8.82775119617225e-05,
+      "loss": 0.563,
+      "step": 285
+    },
+    {
+      "epoch": 0.16481955100880932,
+      "grad_norm": 0.18139953911304474,
+      "learning_rate": 8.797846889952154e-05,
+      "loss": 0.5178,
+      "step": 290
+    },
+    {
+      "epoch": 0.16766126740551293,
+      "grad_norm": 0.16124670207500458,
+      "learning_rate": 8.767942583732058e-05,
+      "loss": 0.5548,
+      "step": 295
+    },
+    {
+      "epoch": 0.17050298380221654,
+      "grad_norm": 0.19367216527462006,
+      "learning_rate": 8.738038277511962e-05,
+      "loss": 0.5615,
+      "step": 300
+    },
+    {
+      "epoch": 0.17334470019892015,
+      "grad_norm": 0.16001980006694794,
+      "learning_rate": 8.708133971291866e-05,
+      "loss": 0.578,
+      "step": 305
+    },
+    {
+      "epoch": 0.17618641659562376,
+      "grad_norm": 0.15890493988990784,
+      "learning_rate": 8.678229665071771e-05,
+      "loss": 0.5617,
+      "step": 310
+    },
+    {
+      "epoch": 0.17902813299232737,
+      "grad_norm": 0.1561274379491806,
+      "learning_rate": 8.648325358851675e-05,
+      "loss": 0.5917,
+      "step": 315
+    },
+    {
+      "epoch": 0.18186984938903097,
+      "grad_norm": 0.18406479060649872,
+      "learning_rate": 8.61842105263158e-05,
+      "loss": 0.5701,
+      "step": 320
+    },
+    {
+      "epoch": 0.18471156578573458,
+      "grad_norm": 0.15967822074890137,
+      "learning_rate": 8.588516746411484e-05,
+      "loss": 0.5571,
+      "step": 325
+    },
+    {
+      "epoch": 0.1875532821824382,
+      "grad_norm": 0.16989850997924805,
+      "learning_rate": 8.558612440191388e-05,
+      "loss": 0.5861,
+      "step": 330
+    },
+    {
+      "epoch": 0.1903949985791418,
+      "grad_norm": 0.17215800285339355,
+      "learning_rate": 8.528708133971293e-05,
+      "loss": 0.5985,
+      "step": 335
+    },
+    {
+      "epoch": 0.1932367149758454,
+      "grad_norm": 0.18377338349819183,
+      "learning_rate": 8.498803827751196e-05,
+      "loss": 0.5799,
+      "step": 340
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 0.18066288530826569,
+      "learning_rate": 8.468899521531101e-05,
+      "loss": 0.5835,
+      "step": 345
+    },
+    {
+      "epoch": 0.19892014776925263,
+      "grad_norm": 0.1578647494316101,
+      "learning_rate": 8.438995215311005e-05,
+      "loss": 0.5275,
+      "step": 350
+    },
+    {
+      "epoch": 0.20176186416595623,
+      "grad_norm": 0.17480728030204773,
+      "learning_rate": 8.40909090909091e-05,
+      "loss": 0.5473,
+      "step": 355
+    },
+    {
+      "epoch": 0.20460358056265984,
+      "grad_norm": 0.16667082905769348,
+      "learning_rate": 8.379186602870814e-05,
+      "loss": 0.5708,
+      "step": 360
+    },
+    {
+      "epoch": 0.20744529695936345,
+      "grad_norm": 0.15754733979701996,
+      "learning_rate": 8.349282296650718e-05,
+      "loss": 0.5781,
+      "step": 365
+    },
+    {
+      "epoch": 0.21028701335606706,
+      "grad_norm": 0.18018724024295807,
+      "learning_rate": 8.319377990430622e-05,
+      "loss": 0.5705,
+      "step": 370
+    },
+    {
+      "epoch": 0.21312872975277067,
+      "grad_norm": 0.17777396738529205,
+      "learning_rate": 8.289473684210527e-05,
+      "loss": 0.567,
+      "step": 375
+    },
+    {
+      "epoch": 0.21597044614947428,
+      "grad_norm": 0.19776058197021484,
+      "learning_rate": 8.259569377990431e-05,
+      "loss": 0.5894,
+      "step": 380
+    },
+    {
+      "epoch": 0.21881216254617789,
+      "grad_norm": 0.1884015053510666,
+      "learning_rate": 8.229665071770335e-05,
+      "loss": 0.5754,
+      "step": 385
+    },
+    {
+      "epoch": 0.2216538789428815,
+      "grad_norm": 0.1602102369070053,
+      "learning_rate": 8.19976076555024e-05,
+      "loss": 0.5486,
+      "step": 390
+    },
+    {
+      "epoch": 0.2244955953395851,
+      "grad_norm": 0.2285519540309906,
+      "learning_rate": 8.169856459330143e-05,
+      "loss": 0.5416,
+      "step": 395
+    },
+    {
+      "epoch": 0.2273373117362887,
+      "grad_norm": 0.17223553359508514,
+      "learning_rate": 8.139952153110049e-05,
+      "loss": 0.5731,
+      "step": 400
+    },
+    {
+      "epoch": 0.23017902813299232,
+      "grad_norm": 0.14779961109161377,
+      "learning_rate": 8.110047846889952e-05,
+      "loss": 0.5824,
+      "step": 405
+    },
+    {
+      "epoch": 0.23302074452969593,
+      "grad_norm": 0.1669321060180664,
+      "learning_rate": 8.080143540669857e-05,
+      "loss": 0.5872,
+      "step": 410
+    },
+    {
+      "epoch": 0.23586246092639954,
+      "grad_norm": 0.166211798787117,
+      "learning_rate": 8.050239234449761e-05,
+      "loss": 0.5956,
+      "step": 415
+    },
+    {
+      "epoch": 0.23870417732310314,
+      "grad_norm": 0.2019493579864502,
+      "learning_rate": 8.020334928229665e-05,
+      "loss": 0.5583,
+      "step": 420
+    },
+    {
+      "epoch": 0.24154589371980675,
+      "grad_norm": 0.15369437634944916,
+      "learning_rate": 7.99043062200957e-05,
+      "loss": 0.5655,
+      "step": 425
+    },
+    {
+      "epoch": 0.24438761011651036,
+      "grad_norm": 0.15199318528175354,
+      "learning_rate": 7.960526315789473e-05,
+      "loss": 0.5926,
+      "step": 430
+    },
+    {
+      "epoch": 0.24722932651321397,
+      "grad_norm": 0.17065800726413727,
+      "learning_rate": 7.930622009569379e-05,
+      "loss": 0.5705,
+      "step": 435
+    },
+    {
+      "epoch": 0.2500710429099176,
+      "grad_norm": 0.16088539361953735,
+      "learning_rate": 7.900717703349283e-05,
+      "loss": 0.5828,
+      "step": 440
+    },
+    {
+      "epoch": 0.2529127593066212,
+      "grad_norm": 0.17311115562915802,
+      "learning_rate": 7.870813397129187e-05,
+      "loss": 0.5547,
+      "step": 445
+    },
+    {
+      "epoch": 0.2557544757033248,
+      "grad_norm": 0.17743873596191406,
+      "learning_rate": 7.840909090909091e-05,
+      "loss": 0.5652,
+      "step": 450
+    },
+    {
+      "epoch": 0.25859619210002843,
+      "grad_norm": 0.1869676560163498,
+      "learning_rate": 7.811004784688995e-05,
+      "loss": 0.5748,
+      "step": 455
+    },
+    {
+      "epoch": 0.26143790849673204,
+      "grad_norm": 0.19634956121444702,
+      "learning_rate": 7.7811004784689e-05,
+      "loss": 0.5824,
+      "step": 460
+    },
+    {
+      "epoch": 0.26427962489343565,
+      "grad_norm": 0.17620642483234406,
+      "learning_rate": 7.751196172248805e-05,
+      "loss": 0.5739,
+      "step": 465
+    },
+    {
+      "epoch": 0.26712134129013926,
+      "grad_norm": 0.18111678957939148,
+      "learning_rate": 7.721291866028707e-05,
+      "loss": 0.5917,
+      "step": 470
+    },
+    {
+      "epoch": 0.26996305768684287,
+      "grad_norm": 0.17866134643554688,
+      "learning_rate": 7.691387559808613e-05,
+      "loss": 0.5409,
+      "step": 475
+    },
+    {
+      "epoch": 0.2728047740835465,
+      "grad_norm": 0.17009033262729645,
+      "learning_rate": 7.661483253588517e-05,
+      "loss": 0.5758,
+      "step": 480
+    },
+    {
+      "epoch": 0.2756464904802501,
+      "grad_norm": 0.15558762848377228,
+      "learning_rate": 7.631578947368422e-05,
+      "loss": 0.5642,
+      "step": 485
+    },
+    {
+      "epoch": 0.2784882068769537,
+      "grad_norm": 0.14788499474525452,
+      "learning_rate": 7.601674641148325e-05,
+      "loss": 0.5673,
+      "step": 490
+    },
+    {
+      "epoch": 0.2813299232736573,
+      "grad_norm": 0.16207771003246307,
+      "learning_rate": 7.571770334928229e-05,
+      "loss": 0.5725,
+      "step": 495
+    },
+    {
+      "epoch": 0.2841716396703609,
+      "grad_norm": 0.16308216750621796,
+      "learning_rate": 7.541866028708135e-05,
+      "loss": 0.5719,
+      "step": 500
+    },
+    {
+      "epoch": 0.2870133560670645,
+      "grad_norm": 0.14916519820690155,
+      "learning_rate": 7.511961722488039e-05,
+      "loss": 0.6092,
+      "step": 505
+    },
+    {
+      "epoch": 0.2898550724637681,
+      "grad_norm": 0.15949136018753052,
+      "learning_rate": 7.482057416267943e-05,
+      "loss": 0.5912,
+      "step": 510
+    },
+    {
+      "epoch": 0.29269678886047173,
+      "grad_norm": 0.16426579654216766,
+      "learning_rate": 7.452153110047847e-05,
+      "loss": 0.5834,
+      "step": 515
+    },
+    {
+      "epoch": 0.29553850525717534,
+      "grad_norm": 0.15682825446128845,
+      "learning_rate": 7.422248803827751e-05,
+      "loss": 0.5572,
+      "step": 520
+    },
+    {
+      "epoch": 0.29838022165387895,
+      "grad_norm": 0.1494492143392563,
+      "learning_rate": 7.392344497607656e-05,
+      "loss": 0.5618,
+      "step": 525
+    },
+    {
+      "epoch": 0.30122193805058256,
+      "grad_norm": 0.1563103348016739,
+      "learning_rate": 7.36244019138756e-05,
+      "loss": 0.582,
+      "step": 530
+    },
+    {
+      "epoch": 0.30406365444728617,
+      "grad_norm": 0.16356562077999115,
+      "learning_rate": 7.332535885167465e-05,
+      "loss": 0.5805,
+      "step": 535
+    },
+    {
+      "epoch": 0.3069053708439898,
+      "grad_norm": 0.17900019884109497,
+      "learning_rate": 7.302631578947369e-05,
+      "loss": 0.5506,
+      "step": 540
+    },
+    {
+      "epoch": 0.3097470872406934,
+      "grad_norm": 0.1820659339427948,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 0.5616,
+      "step": 545
+    },
+    {
+      "epoch": 0.312588803637397,
+      "grad_norm": 0.1745116263628006,
+      "learning_rate": 7.242822966507178e-05,
+      "loss": 0.5537,
+      "step": 550
+    },
+    {
+      "epoch": 0.3154305200341006,
+      "grad_norm": 0.18802793323993683,
+      "learning_rate": 7.212918660287081e-05,
+      "loss": 0.5985,
+      "step": 555
+    },
+    {
+      "epoch": 0.3182722364308042,
+      "grad_norm": 0.18087241053581238,
+      "learning_rate": 7.183014354066986e-05,
+      "loss": 0.5638,
+      "step": 560
+    },
+    {
+      "epoch": 0.3211139528275078,
+      "grad_norm": 0.15812823176383972,
+      "learning_rate": 7.15311004784689e-05,
+      "loss": 0.56,
+      "step": 565
+    },
+    {
+      "epoch": 0.32395566922421143,
+      "grad_norm": 0.227083221077919,
+      "learning_rate": 7.123205741626795e-05,
+      "loss": 0.5504,
+      "step": 570
+    },
+    {
+      "epoch": 0.32679738562091504,
+      "grad_norm": 0.16168774664402008,
+      "learning_rate": 7.093301435406699e-05,
+      "loss": 0.5497,
+      "step": 575
+    },
+    {
+      "epoch": 0.32963910201761865,
+      "grad_norm": 0.19533157348632812,
+      "learning_rate": 7.063397129186603e-05,
+      "loss": 0.5762,
+      "step": 580
+    },
+    {
+      "epoch": 0.33248081841432225,
+      "grad_norm": 0.1547001451253891,
+      "learning_rate": 7.033492822966508e-05,
+      "loss": 0.5617,
+      "step": 585
+    },
+    {
+      "epoch": 0.33532253481102586,
+      "grad_norm": 0.1422310620546341,
+      "learning_rate": 7.003588516746412e-05,
+      "loss": 0.5683,
+      "step": 590
+    },
+    {
+      "epoch": 0.33816425120772947,
+      "grad_norm": 0.1877267211675644,
+      "learning_rate": 6.973684210526315e-05,
+      "loss": 0.5568,
+      "step": 595
+    },
+    {
+      "epoch": 0.3410059676044331,
+      "grad_norm": 0.17235694825649261,
+      "learning_rate": 6.94377990430622e-05,
+      "loss": 0.5585,
+      "step": 600
+    },
+    {
+      "epoch": 0.3438476840011367,
+      "grad_norm": 0.15059125423431396,
+      "learning_rate": 6.913875598086125e-05,
+      "loss": 0.5884,
+      "step": 605
+    },
+    {
+      "epoch": 0.3466894003978403,
+      "grad_norm": 0.24481505155563354,
+      "learning_rate": 6.883971291866029e-05,
+      "loss": 0.5789,
+      "step": 610
+    },
+    {
+      "epoch": 0.3495311167945439,
+      "grad_norm": 0.20350155234336853,
+      "learning_rate": 6.854066985645934e-05,
+      "loss": 0.5779,
+      "step": 615
+    },
+    {
+      "epoch": 0.3523728331912475,
+      "grad_norm": 0.1397661566734314,
+      "learning_rate": 6.824162679425837e-05,
+      "loss": 0.6028,
+      "step": 620
+    },
+    {
+      "epoch": 0.3552145495879511,
+      "grad_norm": 0.15289326012134552,
+      "learning_rate": 6.794258373205742e-05,
+      "loss": 0.5846,
+      "step": 625
+    },
+    {
+      "epoch": 0.35805626598465473,
+      "grad_norm": 0.18911781907081604,
+      "learning_rate": 6.764354066985646e-05,
+      "loss": 0.5734,
+      "step": 630
+    },
+    {
+      "epoch": 0.36089798238135834,
+      "grad_norm": 0.17620323598384857,
+      "learning_rate": 6.73444976076555e-05,
+      "loss": 0.5798,
+      "step": 635
+    },
+    {
+      "epoch": 0.36373969877806195,
+      "grad_norm": 0.16716086864471436,
+      "learning_rate": 6.704545454545455e-05,
+      "loss": 0.573,
+      "step": 640
+    },
+    {
+      "epoch": 0.36658141517476556,
+      "grad_norm": 0.15906579792499542,
+      "learning_rate": 6.674641148325359e-05,
+      "loss": 0.5881,
+      "step": 645
+    },
+    {
+      "epoch": 0.36942313157146917,
+      "grad_norm": 0.16022855043411255,
+      "learning_rate": 6.644736842105264e-05,
+      "loss": 0.5792,
+      "step": 650
+    },
+    {
+      "epoch": 0.3722648479681728,
+      "grad_norm": 0.15457934141159058,
+      "learning_rate": 6.614832535885168e-05,
+      "loss": 0.5695,
+      "step": 655
+    },
+    {
+      "epoch": 0.3751065643648764,
+      "grad_norm": 0.19283390045166016,
+      "learning_rate": 6.584928229665072e-05,
+      "loss": 0.5883,
+      "step": 660
+    },
+    {
+      "epoch": 0.37794828076158,
+      "grad_norm": 0.1575131118297577,
+      "learning_rate": 6.555023923444976e-05,
+      "loss": 0.5746,
+      "step": 665
+    },
+    {
+      "epoch": 0.3807899971582836,
+      "grad_norm": 0.1719864159822464,
+      "learning_rate": 6.52511961722488e-05,
+      "loss": 0.5707,
+      "step": 670
+    },
+    {
+      "epoch": 0.3836317135549872,
+      "grad_norm": 0.1621527373790741,
+      "learning_rate": 6.495215311004786e-05,
+      "loss": 0.5738,
+      "step": 675
+    },
+    {
+      "epoch": 0.3864734299516908,
+      "grad_norm": 0.16876845061779022,
+      "learning_rate": 6.465311004784689e-05,
+      "loss": 0.5559,
+      "step": 680
+    },
+    {
+      "epoch": 0.3893151463483944,
+      "grad_norm": 0.16403600573539734,
+      "learning_rate": 6.435406698564594e-05,
+      "loss": 0.6129,
+      "step": 685
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 0.16663706302642822,
+      "learning_rate": 6.405502392344498e-05,
+      "loss": 0.5915,
+      "step": 690
+    },
+    {
+      "epoch": 0.39499857914180164,
+      "grad_norm": 0.1531498283147812,
+      "learning_rate": 6.375598086124402e-05,
+      "loss": 0.584,
+      "step": 695
+    },
+    {
+      "epoch": 0.39784029553850525,
+      "grad_norm": 0.1686207801103592,
+      "learning_rate": 6.345693779904308e-05,
+      "loss": 0.5637,
+      "step": 700
+    },
+    {
+      "epoch": 0.40068201193520886,
+      "grad_norm": 0.17847242951393127,
+      "learning_rate": 6.31578947368421e-05,
+      "loss": 0.579,
+      "step": 705
+    },
+    {
+      "epoch": 0.40352372833191247,
+      "grad_norm": 0.1523343026638031,
+      "learning_rate": 6.285885167464114e-05,
+      "loss": 0.5938,
+      "step": 710
+    },
+    {
+      "epoch": 0.4063654447286161,
+      "grad_norm": 0.15405000746250153,
+      "learning_rate": 6.25598086124402e-05,
+      "loss": 0.5758,
+      "step": 715
+    },
+    {
+      "epoch": 0.4092071611253197,
+      "grad_norm": 0.1683824509382248,
+      "learning_rate": 6.226076555023924e-05,
+      "loss": 0.5722,
+      "step": 720
+    },
+    {
+      "epoch": 0.4120488775220233,
+      "grad_norm": 0.18010036647319794,
+      "learning_rate": 6.196172248803828e-05,
+      "loss": 0.5362,
+      "step": 725
+    },
+    {
+      "epoch": 0.4148905939187269,
+      "grad_norm": 0.19708412885665894,
+      "learning_rate": 6.166267942583732e-05,
+      "loss": 0.5757,
+      "step": 730
+    },
+    {
+      "epoch": 0.4177323103154305,
+      "grad_norm": 0.15475505590438843,
+      "learning_rate": 6.136363636363636e-05,
+      "loss": 0.6017,
+      "step": 735
+    },
+    {
+      "epoch": 0.4205740267121341,
+      "grad_norm": 0.1665363609790802,
+      "learning_rate": 6.106459330143542e-05,
+      "loss": 0.58,
+      "step": 740
+    },
+    {
+      "epoch": 0.4234157431088377,
+      "grad_norm": 0.18012359738349915,
+      "learning_rate": 6.076555023923445e-05,
+      "loss": 0.5596,
+      "step": 745
+    },
+    {
+      "epoch": 0.42625745950554134,
+      "grad_norm": 0.15219123661518097,
+      "learning_rate": 6.046650717703349e-05,
+      "loss": 0.5698,
+      "step": 750
+    },
+    {
+      "epoch": 0.42909917590224494,
+      "grad_norm": 0.1898249089717865,
+      "learning_rate": 6.016746411483254e-05,
+      "loss": 0.5801,
+      "step": 755
+    },
+    {
+      "epoch": 0.43194089229894855,
+      "grad_norm": 0.17774541676044464,
+      "learning_rate": 5.9868421052631587e-05,
+      "loss": 0.5662,
+      "step": 760
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 0.16042256355285645,
+      "learning_rate": 5.956937799043063e-05,
+      "loss": 0.5582,
+      "step": 765
+    },
+    {
+      "epoch": 0.43762432509235577,
+      "grad_norm": 0.1782207190990448,
+      "learning_rate": 5.927033492822967e-05,
+      "loss": 0.5645,
+      "step": 770
+    },
+    {
+      "epoch": 0.4404660414890594,
+      "grad_norm": 0.15645284950733185,
+      "learning_rate": 5.897129186602871e-05,
+      "loss": 0.5733,
+      "step": 775
+    },
+    {
+      "epoch": 0.443307757885763,
+      "grad_norm": 0.1498394012451172,
+      "learning_rate": 5.867224880382776e-05,
+      "loss": 0.5894,
+      "step": 780
+    },
+    {
+      "epoch": 0.4461494742824666,
+      "grad_norm": 0.14148397743701935,
+      "learning_rate": 5.8373205741626804e-05,
+      "loss": 0.5592,
+      "step": 785
+    },
+    {
+      "epoch": 0.4489911906791702,
+      "grad_norm": 0.1406758725643158,
+      "learning_rate": 5.807416267942584e-05,
+      "loss": 0.565,
+      "step": 790
+    },
+    {
+      "epoch": 0.4518329070758738,
+      "grad_norm": 0.17311552166938782,
+      "learning_rate": 5.777511961722488e-05,
+      "loss": 0.5548,
+      "step": 795
+    },
+    {
+      "epoch": 0.4546746234725774,
+      "grad_norm": 0.16679054498672485,
+      "learning_rate": 5.747607655502393e-05,
+      "loss": 0.5522,
+      "step": 800
+    },
+    {
+      "epoch": 0.45751633986928103,
+      "grad_norm": 0.1412728875875473,
+      "learning_rate": 5.7177033492822975e-05,
+      "loss": 0.5506,
+      "step": 805
+    },
+    {
+      "epoch": 0.46035805626598464,
+      "grad_norm": 0.1498885601758957,
+      "learning_rate": 5.687799043062201e-05,
+      "loss": 0.5833,
+      "step": 810
+    },
+    {
+      "epoch": 0.46319977266268825,
+      "grad_norm": 0.1563435047864914,
+      "learning_rate": 5.6578947368421056e-05,
+      "loss": 0.5926,
+      "step": 815
+    },
+    {
+      "epoch": 0.46604148905939186,
+      "grad_norm": 0.1565445065498352,
+      "learning_rate": 5.62799043062201e-05,
+      "loss": 0.582,
+      "step": 820
+    },
+    {
+      "epoch": 0.46888320545609546,
+      "grad_norm": 0.14581437408924103,
+      "learning_rate": 5.5980861244019145e-05,
+      "loss": 0.5936,
+      "step": 825
+    },
+    {
+      "epoch": 0.4717249218527991,
+      "grad_norm": 0.14694763720035553,
+      "learning_rate": 5.568181818181818e-05,
+      "loss": 0.5672,
+      "step": 830
+    },
+    {
+      "epoch": 0.4745666382495027,
+      "grad_norm": 0.18627385795116425,
+      "learning_rate": 5.5382775119617226e-05,
+      "loss": 0.5826,
+      "step": 835
+    },
+    {
+      "epoch": 0.4774083546462063,
+      "grad_norm": 0.1670178472995758,
+      "learning_rate": 5.5083732057416274e-05,
+      "loss": 0.5488,
+      "step": 840
+    },
+    {
+      "epoch": 0.4802500710429099,
+      "grad_norm": 0.21247391402721405,
+      "learning_rate": 5.4784688995215315e-05,
+      "loss": 0.5664,
+      "step": 845
+    },
+    {
+      "epoch": 0.4830917874396135,
+      "grad_norm": 0.16471049189567566,
+      "learning_rate": 5.448564593301436e-05,
+      "loss": 0.56,
+      "step": 850
+    },
+    {
+      "epoch": 0.4859335038363171,
+      "grad_norm": 0.15371420979499817,
+      "learning_rate": 5.4186602870813397e-05,
+      "loss": 0.5633,
+      "step": 855
+    },
+    {
+      "epoch": 0.4887752202330207,
+      "grad_norm": 0.14671985805034637,
+      "learning_rate": 5.3887559808612444e-05,
+      "loss": 0.6158,
+      "step": 860
+    },
+    {
+      "epoch": 0.49161693662972433,
+      "grad_norm": 0.16204605996608734,
+      "learning_rate": 5.3588516746411485e-05,
+      "loss": 0.5805,
+      "step": 865
+    },
+    {
+      "epoch": 0.49445865302642794,
+      "grad_norm": 0.15839259326457977,
+      "learning_rate": 5.328947368421053e-05,
+      "loss": 0.5521,
+      "step": 870
+    },
+    {
+      "epoch": 0.49730036942313155,
+      "grad_norm": 0.17325200140476227,
+      "learning_rate": 5.299043062200957e-05,
+      "loss": 0.5709,
+      "step": 875
+    },
+    {
+      "epoch": 0.5001420858198352,
+      "grad_norm": 0.157701313495636,
+      "learning_rate": 5.2691387559808614e-05,
+      "loss": 0.5732,
+      "step": 880
+    },
+    {
+      "epoch": 0.5029838022165388,
+      "grad_norm": 0.1631327122449875,
+      "learning_rate": 5.239234449760766e-05,
+      "loss": 0.5595,
+      "step": 885
+    },
+    {
+      "epoch": 0.5058255186132424,
+      "grad_norm": 0.150161013007164,
+      "learning_rate": 5.20933014354067e-05,
+      "loss": 0.5655,
+      "step": 890
+    },
+    {
+      "epoch": 0.508667235009946,
+      "grad_norm": 0.16893287003040314,
+      "learning_rate": 5.179425837320574e-05,
+      "loss": 0.588,
+      "step": 895
+    },
+    {
+      "epoch": 0.5115089514066496,
+      "grad_norm": 0.15415838360786438,
+      "learning_rate": 5.1495215311004785e-05,
+      "loss": 0.5725,
+      "step": 900
+    },
+    {
+      "epoch": 0.5143506678033533,
+      "grad_norm": 0.18115073442459106,
+      "learning_rate": 5.119617224880383e-05,
+      "loss": 0.5887,
+      "step": 905
+    },
+    {
+      "epoch": 0.5171923842000569,
+      "grad_norm": 0.16705133020877838,
+      "learning_rate": 5.089712918660288e-05,
+      "loss": 0.5963,
+      "step": 910
+    },
+    {
+      "epoch": 0.5200341005967605,
+      "grad_norm": 0.17432160675525665,
+      "learning_rate": 5.0598086124401914e-05,
+      "loss": 0.5516,
+      "step": 915
+    },
+    {
+      "epoch": 0.5228758169934641,
+      "grad_norm": 0.1740056574344635,
+      "learning_rate": 5.0299043062200955e-05,
+      "loss": 0.5684,
+      "step": 920
+    },
+    {
+      "epoch": 0.5257175333901677,
+      "grad_norm": 0.14044822752475739,
+      "learning_rate": 5e-05,
+      "loss": 0.5794,
+      "step": 925
+    },
+    {
+      "epoch": 0.5285592497868713,
+      "grad_norm": 0.1570068746805191,
+      "learning_rate": 4.970095693779905e-05,
+      "loss": 0.5871,
+      "step": 930
+    },
+    {
+      "epoch": 0.5314009661835749,
+      "grad_norm": 0.15085619688034058,
+      "learning_rate": 4.940191387559809e-05,
+      "loss": 0.5596,
+      "step": 935
+    },
+    {
+      "epoch": 0.5342426825802785,
+      "grad_norm": 0.1418176144361496,
+      "learning_rate": 4.910287081339713e-05,
+      "loss": 0.5617,
+      "step": 940
+    },
+    {
+      "epoch": 0.5370843989769821,
+      "grad_norm": 0.1582961082458496,
+      "learning_rate": 4.880382775119617e-05,
+      "loss": 0.5582,
+      "step": 945
+    },
+    {
+      "epoch": 0.5399261153736857,
+      "grad_norm": 0.16853678226470947,
+      "learning_rate": 4.850478468899522e-05,
+      "loss": 0.5724,
+      "step": 950
+    },
+    {
+      "epoch": 0.5427678317703893,
+      "grad_norm": 0.17164859175682068,
+      "learning_rate": 4.820574162679426e-05,
+      "loss": 0.5869,
+      "step": 955
+    },
+    {
+      "epoch": 0.545609548167093,
+      "grad_norm": 0.15529300272464752,
+      "learning_rate": 4.790669856459331e-05,
+      "loss": 0.5825,
+      "step": 960
+    },
+    {
+      "epoch": 0.5484512645637966,
+      "grad_norm": 0.16565600037574768,
+      "learning_rate": 4.760765550239234e-05,
+      "loss": 0.551,
+      "step": 965
+    },
+    {
+      "epoch": 0.5512929809605002,
+      "grad_norm": 0.24184668064117432,
+      "learning_rate": 4.730861244019139e-05,
+      "loss": 0.5573,
+      "step": 970
+    },
+    {
+      "epoch": 0.5541346973572038,
+      "grad_norm": 0.15095993876457214,
+      "learning_rate": 4.700956937799043e-05,
+      "loss": 0.5584,
+      "step": 975
+    },
+    {
+      "epoch": 0.5569764137539074,
+      "grad_norm": 0.16745509207248688,
+      "learning_rate": 4.671052631578948e-05,
+      "loss": 0.582,
+      "step": 980
+    },
+    {
+      "epoch": 0.559818130150611,
+      "grad_norm": 0.1616193652153015,
+      "learning_rate": 4.641148325358852e-05,
+      "loss": 0.5837,
+      "step": 985
+    },
+    {
+      "epoch": 0.5626598465473146,
+      "grad_norm": 0.1706905961036682,
+      "learning_rate": 4.611244019138756e-05,
+      "loss": 0.5943,
+      "step": 990
+    },
+    {
+      "epoch": 0.5655015629440182,
+      "grad_norm": 0.16318464279174805,
+      "learning_rate": 4.58133971291866e-05,
+      "loss": 0.556,
+      "step": 995
+    },
+    {
+      "epoch": 0.5683432793407218,
+      "grad_norm": 0.12837469577789307,
+      "learning_rate": 4.551435406698565e-05,
+      "loss": 0.5527,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5711849957374254,
+      "grad_norm": 0.1470261663198471,
+      "learning_rate": 4.521531100478469e-05,
+      "loss": 0.5893,
+      "step": 1005
+    },
+    {
+      "epoch": 0.574026712134129,
+      "grad_norm": 0.4100290834903717,
+      "learning_rate": 4.491626794258373e-05,
+      "loss": 0.6028,
+      "step": 1010
+    },
+    {
+      "epoch": 0.5768684285308326,
+      "grad_norm": 0.15563902258872986,
+      "learning_rate": 4.461722488038278e-05,
+      "loss": 0.5607,
+      "step": 1015
+    },
+    {
+      "epoch": 0.5797101449275363,
+      "grad_norm": 0.15360063314437866,
+      "learning_rate": 4.431818181818182e-05,
+      "loss": 0.5833,
+      "step": 1020
+    },
+    {
+      "epoch": 0.5825518613242399,
+      "grad_norm": 0.14984561502933502,
+      "learning_rate": 4.401913875598087e-05,
+      "loss": 0.5429,
+      "step": 1025
+    },
+    {
+      "epoch": 0.5853935777209435,
+      "grad_norm": 0.15414854884147644,
+      "learning_rate": 4.372009569377991e-05,
+      "loss": 0.5868,
+      "step": 1030
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 0.13921906054019928,
+      "learning_rate": 4.342105263157895e-05,
+      "loss": 0.5879,
+      "step": 1035
+    },
+    {
+      "epoch": 0.5910770105143507,
+      "grad_norm": 0.21716231107711792,
+      "learning_rate": 4.312200956937799e-05,
+      "loss": 0.5636,
+      "step": 1040
+    },
+    {
+      "epoch": 0.5939187269110543,
+      "grad_norm": 0.15942206978797913,
+      "learning_rate": 4.282296650717704e-05,
+      "loss": 0.5603,
+      "step": 1045
+    },
+    {
+      "epoch": 0.5967604433077579,
+      "grad_norm": 0.14409542083740234,
+      "learning_rate": 4.252392344497608e-05,
+      "loss": 0.5535,
+      "step": 1050
+    },
+    {
+      "epoch": 0.5996021597044615,
+      "grad_norm": 0.15243200957775116,
+      "learning_rate": 4.2224880382775126e-05,
+      "loss": 0.5695,
+      "step": 1055
+    },
+    {
+      "epoch": 0.6024438761011651,
+      "grad_norm": 0.16177250444889069,
+      "learning_rate": 4.192583732057416e-05,
+      "loss": 0.5949,
+      "step": 1060
+    },
+    {
+      "epoch": 0.6052855924978687,
+      "grad_norm": 0.14140868186950684,
+      "learning_rate": 4.162679425837321e-05,
+      "loss": 0.5875,
+      "step": 1065
+    },
+    {
+      "epoch": 0.6081273088945723,
+      "grad_norm": 0.17228946089744568,
+      "learning_rate": 4.132775119617225e-05,
+      "loss": 0.5824,
+      "step": 1070
+    },
+    {
+      "epoch": 0.610969025291276,
+      "grad_norm": 0.14932122826576233,
+      "learning_rate": 4.1028708133971296e-05,
+      "loss": 0.582,
+      "step": 1075
+    },
+    {
+      "epoch": 0.6138107416879796,
+      "grad_norm": 0.1791832447052002,
+      "learning_rate": 4.0729665071770337e-05,
+      "loss": 0.5507,
+      "step": 1080
+    },
+    {
+      "epoch": 0.6166524580846832,
+      "grad_norm": 0.15530230104923248,
+      "learning_rate": 4.043062200956938e-05,
+      "loss": 0.5851,
+      "step": 1085
+    },
+    {
+      "epoch": 0.6194941744813868,
+      "grad_norm": 0.15048903226852417,
+      "learning_rate": 4.0131578947368425e-05,
+      "loss": 0.5629,
+      "step": 1090
+    },
+    {
+      "epoch": 0.6223358908780904,
+      "grad_norm": 0.14732453227043152,
+      "learning_rate": 3.9832535885167466e-05,
+      "loss": 0.5752,
+      "step": 1095
+    },
+    {
+      "epoch": 0.625177607274794,
+      "grad_norm": 0.15778231620788574,
+      "learning_rate": 3.9533492822966514e-05,
+      "loss": 0.5607,
+      "step": 1100
+    },
+    {
+      "epoch": 0.6280193236714976,
+      "grad_norm": 0.16504403948783875,
+      "learning_rate": 3.9234449760765554e-05,
+      "loss": 0.5887,
+      "step": 1105
+    },
+    {
+      "epoch": 0.6308610400682012,
+      "grad_norm": 0.15236054360866547,
+      "learning_rate": 3.8935406698564595e-05,
+      "loss": 0.5575,
+      "step": 1110
+    },
+    {
+      "epoch": 0.6337027564649048,
+      "grad_norm": 0.15717913210391998,
+      "learning_rate": 3.8636363636363636e-05,
+      "loss": 0.537,
+      "step": 1115
+    },
+    {
+      "epoch": 0.6365444728616084,
+      "grad_norm": 0.16705450415611267,
+      "learning_rate": 3.8337320574162684e-05,
+      "loss": 0.5769,
+      "step": 1120
+    },
+    {
+      "epoch": 0.639386189258312,
+      "grad_norm": 0.15541672706604004,
+      "learning_rate": 3.8038277511961725e-05,
+      "loss": 0.5634,
+      "step": 1125
+    },
+    {
+      "epoch": 0.6422279056550156,
+      "grad_norm": 0.23576730489730835,
+      "learning_rate": 3.7739234449760765e-05,
+      "loss": 0.583,
+      "step": 1130
+    },
+    {
+      "epoch": 0.6450696220517192,
+      "grad_norm": 0.14925654232501984,
+      "learning_rate": 3.7440191387559806e-05,
+      "loss": 0.5774,
+      "step": 1135
+    },
+    {
+      "epoch": 0.6479113384484229,
+      "grad_norm": 0.133391872048378,
+      "learning_rate": 3.7141148325358854e-05,
+      "loss": 0.5981,
+      "step": 1140
+    },
+    {
+      "epoch": 0.6507530548451265,
+      "grad_norm": 0.1577860563993454,
+      "learning_rate": 3.6842105263157895e-05,
+      "loss": 0.5844,
+      "step": 1145
+    },
+    {
+      "epoch": 0.6535947712418301,
+      "grad_norm": 0.15949822962284088,
+      "learning_rate": 3.654306220095694e-05,
+      "loss": 0.5646,
+      "step": 1150
+    },
+    {
+      "epoch": 0.6564364876385337,
+      "grad_norm": 0.15498408675193787,
+      "learning_rate": 3.624401913875598e-05,
+      "loss": 0.5679,
+      "step": 1155
+    },
+    {
+      "epoch": 0.6592782040352373,
+      "grad_norm": 0.18732386827468872,
+      "learning_rate": 3.5944976076555024e-05,
+      "loss": 0.5714,
+      "step": 1160
+    },
+    {
+      "epoch": 0.6621199204319409,
+      "grad_norm": 0.1633375883102417,
+      "learning_rate": 3.5645933014354065e-05,
+      "loss": 0.5654,
+      "step": 1165
+    },
+    {
+      "epoch": 0.6649616368286445,
+      "grad_norm": 0.15558752417564392,
+      "learning_rate": 3.534688995215311e-05,
+      "loss": 0.5445,
+      "step": 1170
+    },
+    {
+      "epoch": 0.6678033532253481,
+      "grad_norm": 0.14220796525478363,
+      "learning_rate": 3.504784688995216e-05,
+      "loss": 0.5756,
+      "step": 1175
+    },
+    {
+      "epoch": 0.6706450696220517,
+      "grad_norm": 0.1663515418767929,
+      "learning_rate": 3.4748803827751194e-05,
+      "loss": 0.5892,
+      "step": 1180
+    },
+    {
+      "epoch": 0.6734867860187553,
+      "grad_norm": 0.1588052362203598,
+      "learning_rate": 3.444976076555024e-05,
+      "loss": 0.5954,
+      "step": 1185
+    },
+    {
+      "epoch": 0.6763285024154589,
+      "grad_norm": 0.1659833937883377,
+      "learning_rate": 3.415071770334928e-05,
+      "loss": 0.5849,
+      "step": 1190
+    },
+    {
+      "epoch": 0.6791702188121626,
+      "grad_norm": 0.1444244086742401,
+      "learning_rate": 3.385167464114833e-05,
+      "loss": 0.5807,
+      "step": 1195
+    },
+    {
+      "epoch": 0.6820119352088662,
+      "grad_norm": 0.14998075366020203,
+      "learning_rate": 3.355263157894737e-05,
+      "loss": 0.5676,
+      "step": 1200
+    },
+    {
+      "epoch": 0.6848536516055698,
+      "grad_norm": 0.14586575329303741,
+      "learning_rate": 3.325358851674641e-05,
+      "loss": 0.5833,
+      "step": 1205
+    },
+    {
+      "epoch": 0.6876953680022734,
+      "grad_norm": 0.16619591414928436,
+      "learning_rate": 3.295454545454545e-05,
+      "loss": 0.5738,
+      "step": 1210
+    },
+    {
+      "epoch": 0.690537084398977,
+      "grad_norm": 0.15603290498256683,
+      "learning_rate": 3.26555023923445e-05,
+      "loss": 0.549,
+      "step": 1215
+    },
+    {
+      "epoch": 0.6933788007956806,
+      "grad_norm": 0.16562019288539886,
+      "learning_rate": 3.235645933014354e-05,
+      "loss": 0.5681,
+      "step": 1220
+    },
+    {
+      "epoch": 0.6962205171923842,
+      "grad_norm": 0.1772131323814392,
+      "learning_rate": 3.205741626794259e-05,
+      "loss": 0.5649,
+      "step": 1225
+    },
+    {
+      "epoch": 0.6990622335890878,
+      "grad_norm": 0.21865883469581604,
+      "learning_rate": 3.175837320574162e-05,
+      "loss": 0.5487,
+      "step": 1230
+    },
+    {
+      "epoch": 0.7019039499857914,
+      "grad_norm": 0.1668052226305008,
+      "learning_rate": 3.145933014354067e-05,
+      "loss": 0.5594,
+      "step": 1235
+    },
+    {
+      "epoch": 0.704745666382495,
+      "grad_norm": 0.17883329093456268,
+      "learning_rate": 3.116028708133971e-05,
+      "loss": 0.5677,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7075873827791986,
+      "grad_norm": 0.17108604311943054,
+      "learning_rate": 3.086124401913876e-05,
+      "loss": 0.5732,
+      "step": 1245
+    },
+    {
+      "epoch": 0.7104290991759022,
+      "grad_norm": 0.13776730000972748,
+      "learning_rate": 3.05622009569378e-05,
+      "loss": 0.5526,
+      "step": 1250
+    },
+    {
+      "epoch": 0.7132708155726059,
+      "grad_norm": 0.15627716481685638,
+      "learning_rate": 3.0263157894736844e-05,
+      "loss": 0.5679,
+      "step": 1255
+    },
+    {
+      "epoch": 0.7161125319693095,
+      "grad_norm": 0.15834760665893555,
+      "learning_rate": 2.996411483253589e-05,
+      "loss": 0.5875,
+      "step": 1260
+    },
+    {
+      "epoch": 0.7189542483660131,
+      "grad_norm": 0.1533428430557251,
+      "learning_rate": 2.966507177033493e-05,
+      "loss": 0.572,
+      "step": 1265
+    },
+    {
+      "epoch": 0.7217959647627167,
+      "grad_norm": 0.17058540880680084,
+      "learning_rate": 2.9366028708133974e-05,
+      "loss": 0.537,
+      "step": 1270
+    },
+    {
+      "epoch": 0.7246376811594203,
+      "grad_norm": 0.15672728419303894,
+      "learning_rate": 2.9066985645933014e-05,
+      "loss": 0.5552,
+      "step": 1275
+    },
+    {
+      "epoch": 0.7274793975561239,
+      "grad_norm": 1.2518200874328613,
+      "learning_rate": 2.8767942583732062e-05,
+      "loss": 0.5934,
+      "step": 1280
+    },
+    {
+      "epoch": 0.7303211139528275,
+      "grad_norm": 0.1489401012659073,
+      "learning_rate": 2.84688995215311e-05,
+      "loss": 0.596,
+      "step": 1285
+    },
+    {
+      "epoch": 0.7331628303495311,
+      "grad_norm": 0.14582081139087677,
+      "learning_rate": 2.8169856459330147e-05,
+      "loss": 0.5869,
+      "step": 1290
+    },
+    {
+      "epoch": 0.7360045467462347,
+      "grad_norm": 0.20285780727863312,
+      "learning_rate": 2.7870813397129185e-05,
+      "loss": 0.5854,
+      "step": 1295
+    },
+    {
+      "epoch": 0.7388462631429383,
+      "grad_norm": 0.14528052508831024,
+      "learning_rate": 2.7571770334928232e-05,
+      "loss": 0.5594,
+      "step": 1300
+    },
+    {
+      "epoch": 0.7416879795396419,
+      "grad_norm": 0.1703929305076599,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 0.563,
+      "step": 1305
+    },
+    {
+      "epoch": 0.7445296959363455,
+      "grad_norm": 0.1452837884426117,
+      "learning_rate": 2.6973684210526317e-05,
+      "loss": 0.5627,
+      "step": 1310
+    },
+    {
+      "epoch": 0.7473714123330492,
+      "grad_norm": 0.14418819546699524,
+      "learning_rate": 2.6674641148325358e-05,
+      "loss": 0.5594,
+      "step": 1315
+    },
+    {
+      "epoch": 0.7502131287297528,
+      "grad_norm": 0.17991849780082703,
+      "learning_rate": 2.6375598086124402e-05,
+      "loss": 0.5791,
+      "step": 1320
+    },
+    {
+      "epoch": 0.7530548451264564,
+      "grad_norm": 0.15669088065624237,
+      "learning_rate": 2.6076555023923443e-05,
+      "loss": 0.5747,
+      "step": 1325
+    },
+    {
+      "epoch": 0.75589656152316,
+      "grad_norm": 0.1644110381603241,
+      "learning_rate": 2.5777511961722488e-05,
+      "loss": 0.5576,
+      "step": 1330
+    },
+    {
+      "epoch": 0.7587382779198636,
+      "grad_norm": 0.14174972474575043,
+      "learning_rate": 2.5478468899521535e-05,
+      "loss": 0.5909,
+      "step": 1335
+    },
+    {
+      "epoch": 0.7615799943165672,
+      "grad_norm": 0.16112005710601807,
+      "learning_rate": 2.5179425837320576e-05,
+      "loss": 0.5808,
+      "step": 1340
+    },
+    {
+      "epoch": 0.7644217107132708,
+      "grad_norm": 0.17508378624916077,
+      "learning_rate": 2.4880382775119617e-05,
+      "loss": 0.5462,
+      "step": 1345
+    },
+    {
+      "epoch": 0.7672634271099744,
+      "grad_norm": 0.14891307055950165,
+      "learning_rate": 2.458133971291866e-05,
+      "loss": 0.5756,
+      "step": 1350
+    },
+    {
+      "epoch": 0.770105143506678,
+      "grad_norm": 0.16033193469047546,
+      "learning_rate": 2.4282296650717702e-05,
+      "loss": 0.5692,
+      "step": 1355
+    },
+    {
+      "epoch": 0.7729468599033816,
+      "grad_norm": 0.16894228756427765,
+      "learning_rate": 2.3983253588516746e-05,
+      "loss": 0.5942,
+      "step": 1360
+    },
+    {
+      "epoch": 0.7757885763000852,
+      "grad_norm": 0.15012076497077942,
+      "learning_rate": 2.368421052631579e-05,
+      "loss": 0.5623,
+      "step": 1365
+    },
+    {
+      "epoch": 0.7786302926967889,
+      "grad_norm": 0.1476900577545166,
+      "learning_rate": 2.3385167464114835e-05,
+      "loss": 0.538,
+      "step": 1370
+    },
+    {
+      "epoch": 0.7814720090934925,
+      "grad_norm": 0.1336384266614914,
+      "learning_rate": 2.308612440191388e-05,
+      "loss": 0.5683,
+      "step": 1375
+    },
+    {
+      "epoch": 0.7843137254901961,
+      "grad_norm": 0.15805655717849731,
+      "learning_rate": 2.278708133971292e-05,
+      "loss": 0.5713,
+      "step": 1380
+    },
+    {
+      "epoch": 0.7871554418868997,
+      "grad_norm": 0.1553046703338623,
+      "learning_rate": 2.2488038277511964e-05,
+      "loss": 0.5401,
+      "step": 1385
+    },
+    {
+      "epoch": 0.7899971582836033,
+      "grad_norm": 0.1558494120836258,
+      "learning_rate": 2.2188995215311005e-05,
+      "loss": 0.5717,
+      "step": 1390
+    },
+    {
+      "epoch": 0.7928388746803069,
+      "grad_norm": 0.18873989582061768,
+      "learning_rate": 2.188995215311005e-05,
+      "loss": 0.5636,
+      "step": 1395
+    },
+    {
+      "epoch": 0.7956805910770105,
+      "grad_norm": 0.14990967512130737,
+      "learning_rate": 2.1590909090909093e-05,
+      "loss": 0.5581,
+      "step": 1400
+    },
+    {
+      "epoch": 0.7985223074737141,
+      "grad_norm": 0.19469797611236572,
+      "learning_rate": 2.1291866028708134e-05,
+      "loss": 0.5549,
+      "step": 1405
+    },
+    {
+      "epoch": 0.8013640238704177,
+      "grad_norm": 0.1545160710811615,
+      "learning_rate": 2.099282296650718e-05,
+      "loss": 0.5888,
+      "step": 1410
+    },
+    {
+      "epoch": 0.8042057402671213,
+      "grad_norm": 0.1678376942873001,
+      "learning_rate": 2.069377990430622e-05,
+      "loss": 0.5655,
+      "step": 1415
+    },
+    {
+      "epoch": 0.8070474566638249,
+      "grad_norm": 0.16768623888492584,
+      "learning_rate": 2.0394736842105264e-05,
+      "loss": 0.5558,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8098891730605285,
+      "grad_norm": 0.13661779463291168,
+      "learning_rate": 2.0095693779904308e-05,
+      "loss": 0.5747,
+      "step": 1425
+    },
+    {
+      "epoch": 0.8127308894572322,
+      "grad_norm": 0.1540144979953766,
+      "learning_rate": 1.979665071770335e-05,
+      "loss": 0.5679,
+      "step": 1430
+    },
+    {
+      "epoch": 0.8155726058539358,
+      "grad_norm": 0.13239075243473053,
+      "learning_rate": 1.9497607655502393e-05,
+      "loss": 0.5707,
+      "step": 1435
+    },
+    {
+      "epoch": 0.8184143222506394,
+      "grad_norm": 0.14625592529773712,
+      "learning_rate": 1.9198564593301434e-05,
+      "loss": 0.5849,
+      "step": 1440
+    },
+    {
+      "epoch": 0.821256038647343,
+      "grad_norm": 0.14197929203510284,
+      "learning_rate": 1.8899521531100478e-05,
+      "loss": 0.5631,
+      "step": 1445
+    },
+    {
+      "epoch": 0.8240977550440466,
+      "grad_norm": 0.13680891692638397,
+      "learning_rate": 1.8600478468899522e-05,
+      "loss": 0.5652,
+      "step": 1450
+    },
+    {
+      "epoch": 0.8269394714407502,
+      "grad_norm": 0.175118550658226,
+      "learning_rate": 1.8301435406698566e-05,
+      "loss": 0.5749,
+      "step": 1455
+    },
+    {
+      "epoch": 0.8297811878374538,
+      "grad_norm": 0.18835903704166412,
+      "learning_rate": 1.800239234449761e-05,
+      "loss": 0.5794,
+      "step": 1460
+    },
+    {
+      "epoch": 0.8326229042341574,
+      "grad_norm": 0.3392263352870941,
+      "learning_rate": 1.770334928229665e-05,
+      "loss": 0.5951,
+      "step": 1465
+    },
+    {
+      "epoch": 0.835464620630861,
+      "grad_norm": 0.139173686504364,
+      "learning_rate": 1.7404306220095696e-05,
+      "loss": 0.5562,
+      "step": 1470
+    },
+    {
+      "epoch": 0.8383063370275646,
+      "grad_norm": 0.24194057285785675,
+      "learning_rate": 1.7105263157894737e-05,
+      "loss": 0.5626,
+      "step": 1475
+    },
+    {
+      "epoch": 0.8411480534242682,
+      "grad_norm": 0.1710263043642044,
+      "learning_rate": 1.680622009569378e-05,
+      "loss": 0.5832,
+      "step": 1480
+    },
+    {
+      "epoch": 0.8439897698209718,
+      "grad_norm": 0.15678372979164124,
+      "learning_rate": 1.6507177033492825e-05,
+      "loss": 0.5809,
+      "step": 1485
+    },
+    {
+      "epoch": 0.8468314862176755,
+      "grad_norm": 0.15351912379264832,
+      "learning_rate": 1.6208133971291866e-05,
+      "loss": 0.5664,
+      "step": 1490
+    },
+    {
+      "epoch": 0.8496732026143791,
+      "grad_norm": 0.16644535958766937,
+      "learning_rate": 1.590909090909091e-05,
+      "loss": 0.5514,
+      "step": 1495
+    },
+    {
+      "epoch": 0.8525149190110827,
+      "grad_norm": 0.14531302452087402,
+      "learning_rate": 1.561004784688995e-05,
+      "loss": 0.5646,
+      "step": 1500
+    },
+    {
+      "epoch": 0.8553566354077863,
+      "grad_norm": 0.1624838262796402,
+      "learning_rate": 1.5311004784688995e-05,
+      "loss": 0.5483,
+      "step": 1505
+    },
+    {
+      "epoch": 0.8581983518044899,
+      "grad_norm": 0.1581767201423645,
+      "learning_rate": 1.5011961722488038e-05,
+      "loss": 0.6055,
+      "step": 1510
+    },
+    {
+      "epoch": 0.8610400682011935,
+      "grad_norm": 0.13509996235370636,
+      "learning_rate": 1.471291866028708e-05,
+      "loss": 0.5702,
+      "step": 1515
+    },
+    {
+      "epoch": 0.8638817845978971,
+      "grad_norm": 0.1682087630033493,
+      "learning_rate": 1.4413875598086125e-05,
+      "loss": 0.55,
+      "step": 1520
+    },
+    {
+      "epoch": 0.8667235009946007,
+      "grad_norm": 0.13495485484600067,
+      "learning_rate": 1.4114832535885167e-05,
+      "loss": 0.563,
+      "step": 1525
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 0.15126639604568481,
+      "learning_rate": 1.3815789473684213e-05,
+      "loss": 0.5607,
+      "step": 1530
+    },
+    {
+      "epoch": 0.8724069337880079,
+      "grad_norm": 0.1609649509191513,
+      "learning_rate": 1.3516746411483256e-05,
+      "loss": 0.5767,
+      "step": 1535
+    },
+    {
+      "epoch": 0.8752486501847115,
+      "grad_norm": 0.1570013165473938,
+      "learning_rate": 1.3217703349282298e-05,
+      "loss": 0.5838,
+      "step": 1540
+    },
+    {
+      "epoch": 0.8780903665814151,
+      "grad_norm": 0.1471746861934662,
+      "learning_rate": 1.291866028708134e-05,
+      "loss": 0.5687,
+      "step": 1545
+    },
+    {
+      "epoch": 0.8809320829781188,
+      "grad_norm": 0.13889586925506592,
+      "learning_rate": 1.2619617224880383e-05,
+      "loss": 0.572,
+      "step": 1550
+    },
+    {
+      "epoch": 0.8837737993748224,
+      "grad_norm": 0.1460907906293869,
+      "learning_rate": 1.2320574162679427e-05,
+      "loss": 0.5731,
+      "step": 1555
+    },
+    {
+      "epoch": 0.886615515771526,
+      "grad_norm": 0.14351268112659454,
+      "learning_rate": 1.202153110047847e-05,
+      "loss": 0.5438,
+      "step": 1560
+    },
+    {
+      "epoch": 0.8894572321682296,
+      "grad_norm": 0.14592865109443665,
+      "learning_rate": 1.1722488038277513e-05,
+      "loss": 0.577,
+      "step": 1565
+    },
+    {
+      "epoch": 0.8922989485649332,
+      "grad_norm": 0.16943998634815216,
+      "learning_rate": 1.1423444976076555e-05,
+      "loss": 0.5645,
+      "step": 1570
+    },
+    {
+      "epoch": 0.8951406649616368,
+      "grad_norm": 0.1501266062259674,
+      "learning_rate": 1.1124401913875598e-05,
+      "loss": 0.5524,
+      "step": 1575
+    },
+    {
+      "epoch": 0.8979823813583404,
+      "grad_norm": 0.14711496233940125,
+      "learning_rate": 1.0825358851674642e-05,
+      "loss": 0.5793,
+      "step": 1580
+    },
+    {
+      "epoch": 0.900824097755044,
+      "grad_norm": 0.15462088584899902,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.5662,
+      "step": 1585
+    },
+    {
+      "epoch": 0.9036658141517476,
+      "grad_norm": 0.14514751732349396,
+      "learning_rate": 1.0227272727272729e-05,
+      "loss": 0.5707,
+      "step": 1590
+    },
+    {
+      "epoch": 0.9065075305484512,
+      "grad_norm": 0.14459151029586792,
+      "learning_rate": 9.928229665071771e-06,
+      "loss": 0.5649,
+      "step": 1595
+    },
+    {
+      "epoch": 0.9093492469451548,
+      "grad_norm": 0.1554642617702484,
+      "learning_rate": 9.629186602870814e-06,
+      "loss": 0.5921,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9121909633418585,
+      "grad_norm": 0.14525796473026276,
+      "learning_rate": 9.330143540669856e-06,
+      "loss": 0.5847,
+      "step": 1605
+    },
+    {
+      "epoch": 0.9150326797385621,
+      "grad_norm": 0.1505240797996521,
+      "learning_rate": 9.0311004784689e-06,
+      "loss": 0.577,
+      "step": 1610
+    },
+    {
+      "epoch": 0.9178743961352657,
+      "grad_norm": 0.14378653466701508,
+      "learning_rate": 8.732057416267943e-06,
+      "loss": 0.5321,
+      "step": 1615
+    },
+    {
+      "epoch": 0.9207161125319693,
+      "grad_norm": 0.15809813141822815,
+      "learning_rate": 8.433014354066986e-06,
+      "loss": 0.5712,
+      "step": 1620
+    },
+    {
+      "epoch": 0.9235578289286729,
+      "grad_norm": 0.14394868910312653,
+      "learning_rate": 8.133971291866028e-06,
+      "loss": 0.5347,
+      "step": 1625
+    },
+    {
+      "epoch": 0.9263995453253765,
+      "grad_norm": 0.1609821915626526,
+      "learning_rate": 7.834928229665072e-06,
+      "loss": 0.5843,
+      "step": 1630
+    },
+    {
+      "epoch": 0.9292412617220801,
+      "grad_norm": 0.14901985228061676,
+      "learning_rate": 7.535885167464116e-06,
+      "loss": 0.5899,
+      "step": 1635
+    },
+    {
+      "epoch": 0.9320829781187837,
+      "grad_norm": 0.16525277495384216,
+      "learning_rate": 7.236842105263158e-06,
+      "loss": 0.5651,
+      "step": 1640
+    },
+    {
+      "epoch": 0.9349246945154873,
+      "grad_norm": 0.1574372798204422,
+      "learning_rate": 6.937799043062202e-06,
+      "loss": 0.5702,
+      "step": 1645
+    },
+    {
+      "epoch": 0.9377664109121909,
+      "grad_norm": 0.18484370410442352,
+      "learning_rate": 6.638755980861244e-06,
+      "loss": 0.5855,
+      "step": 1650
+    },
+    {
+      "epoch": 0.9406081273088945,
+      "grad_norm": 0.15279646217823029,
+      "learning_rate": 6.339712918660287e-06,
+      "loss": 0.5672,
+      "step": 1655
+    },
+    {
+      "epoch": 0.9434498437055981,
+      "grad_norm": 0.1466342955827713,
+      "learning_rate": 6.040669856459331e-06,
+      "loss": 0.5818,
+      "step": 1660
+    },
+    {
+      "epoch": 0.9462915601023018,
+      "grad_norm": 0.146029993891716,
+      "learning_rate": 5.741626794258374e-06,
+      "loss": 0.5569,
+      "step": 1665
+    },
+    {
+      "epoch": 0.9491332764990054,
+      "grad_norm": 0.16354189813137054,
+      "learning_rate": 5.442583732057416e-06,
+      "loss": 0.5768,
+      "step": 1670
+    },
+    {
+      "epoch": 0.951974992895709,
+      "grad_norm": 0.14640045166015625,
+      "learning_rate": 5.14354066985646e-06,
+      "loss": 0.5683,
+      "step": 1675
+    },
+    {
+      "epoch": 0.9548167092924126,
+      "grad_norm": 0.15549986064434052,
+      "learning_rate": 4.844497607655503e-06,
+      "loss": 0.5649,
+      "step": 1680
+    },
+    {
+      "epoch": 0.9576584256891162,
+      "grad_norm": 0.18970870971679688,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.5964,
+      "step": 1685
+    },
+    {
+      "epoch": 0.9605001420858198,
+      "grad_norm": 0.13758742809295654,
+      "learning_rate": 4.246411483253589e-06,
+      "loss": 0.5725,
+      "step": 1690
+    },
+    {
+      "epoch": 0.9633418584825234,
+      "grad_norm": 0.175694540143013,
+      "learning_rate": 3.9473684210526315e-06,
+      "loss": 0.5436,
+      "step": 1695
+    },
+    {
+      "epoch": 0.966183574879227,
+      "grad_norm": 0.1392461508512497,
+      "learning_rate": 3.6483253588516753e-06,
+      "loss": 0.5469,
+      "step": 1700
+    },
+    {
+      "epoch": 0.9690252912759306,
+      "grad_norm": 0.14097018539905548,
+      "learning_rate": 3.349282296650718e-06,
+      "loss": 0.562,
+      "step": 1705
+    },
+    {
+      "epoch": 0.9718670076726342,
+      "grad_norm": 0.1417640596628189,
+      "learning_rate": 3.0502392344497608e-06,
+      "loss": 0.574,
+      "step": 1710
+    },
+    {
+      "epoch": 0.9747087240693378,
+      "grad_norm": 0.149598628282547,
+      "learning_rate": 2.751196172248804e-06,
+      "loss": 0.5619,
+      "step": 1715
+    },
+    {
+      "epoch": 0.9775504404660414,
+      "grad_norm": 0.14410585165023804,
+      "learning_rate": 2.452153110047847e-06,
+      "loss": 0.5669,
+      "step": 1720
+    },
+    {
+      "epoch": 0.9803921568627451,
+      "grad_norm": 0.15505129098892212,
+      "learning_rate": 2.15311004784689e-06,
+      "loss": 0.5573,
+      "step": 1725
+    },
+    {
+      "epoch": 0.9832338732594487,
+      "grad_norm": 0.13935989141464233,
+      "learning_rate": 1.854066985645933e-06,
+      "loss": 0.5529,
+      "step": 1730
+    },
+    {
+      "epoch": 0.9860755896561523,
+      "grad_norm": 0.15468458831310272,
+      "learning_rate": 1.555023923444976e-06,
+      "loss": 0.5788,
+      "step": 1735
+    },
+    {
+      "epoch": 0.9889173060528559,
+      "grad_norm": 0.15632416307926178,
+      "learning_rate": 1.2559808612440192e-06,
+      "loss": 0.5673,
+      "step": 1740
+    },
+    {
+      "epoch": 0.9917590224495595,
+      "grad_norm": 0.19154873490333557,
+      "learning_rate": 9.569377990430622e-07,
+      "loss": 0.5613,
+      "step": 1745
+    },
+    {
+      "epoch": 0.9946007388462631,
+      "grad_norm": 0.13292330503463745,
+      "learning_rate": 6.578947368421053e-07,
+      "loss": 0.5693,
+      "step": 1750
+    },
+    {
+      "epoch": 0.9974424552429667,
+      "grad_norm": 0.16198578476905823,
+      "learning_rate": 3.5885167464114835e-07,
+      "loss": 0.5678,
+      "step": 1755
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2121179699897766,
+      "learning_rate": 5.980861244019139e-08,
+      "loss": 0.5747,
+      "step": 1760
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1760,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.910811156402248e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

Qwen3-4B-Thinking-2507-sft-fusang-swa=2k_sink=0_fadec/checkpoint-2778/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

Qwen3-4B-Thinking-2507-sft-fusang-swa=2k_sink=0_fadec/checkpoint-2778/chat_template.jinja

@@ -0,0 +1,86 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n<think>\n' }}
+{%- endif %}

Qwen3-4B-Thinking-2507-sft-fusang-swa=2k_sink=0_fadec/checkpoint-2778/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}