Model save

README.md

@@ -2,15 +2,11 @@
 license: apache-2.0
 base_model: BioMistral/BioMistral-7B
 tags:
-- alignment-handbook
-- trl
-- sft
-- generated_from_trainer
 - trl
 - sft
 - generated_from_trainer
 datasets:
-- HuggingFaceH4/deita-10k-v0-sft
+- generator
 model-index:
 - name: biomistral-7b-wo-kqa_golden-iter-sft-step1
   results: []
@@ -21,9 +17,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # biomistral-7b-wo-kqa_golden-iter-sft-step1
 
-This model is a fine-tuned version of [BioMistral/BioMistral-7B](https://huggingface.co/BioMistral/BioMistral-7B) on the HuggingFaceH4/deita-10k-v0-sft dataset.
+This model is a fine-tuned version of [BioMistral/BioMistral-7B](https://huggingface.co/BioMistral/BioMistral-7B) on the generator dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.4481
+- Loss: 1.4295
 
 ## Model description
 
@@ -60,9 +56,9 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 0.7793        | 0.92  | 9    | 1.4149          |
-| 0.53          | 1.95  | 19   | 1.4281          |
-| 0.4017        | 2.77  | 27   | 1.4481          |
+| 0.8334        | 0.92  | 9    | 1.4131          |
+| 0.6303        | 1.95  | 19   | 1.4060          |
+| 0.4647        | 2.77  | 27   | 1.4295          |
 
 
 ### Framework versions

all_results.json

@@ -1,13 +1,8 @@
 {
     "epoch": 2.77,
-    "eval_loss": 1.4481348991394043,
-    "eval_runtime": 56.5827,
-    "eval_samples": 4044,
-    "eval_samples_per_second": 6.08,
-    "eval_steps_per_second": 0.389,
-    "train_loss": 0.5429546391522443,
-    "train_runtime": 799.7979,
+    "train_loss": 0.6124309632513258,
+    "train_runtime": 676.0028,
     "train_samples": 4750,
-    "train_samples_per_second": 2.296,
-    "train_steps_per_second": 0.034
+    "train_samples_per_second": 2.72,
+    "train_steps_per_second": 0.04
 }

config.json

@@ -21,6 +21,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.39.0.dev0",
-  "use_cache": true,
+  "use_cache": false,
   "vocab_size": 32000
 }

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c18da0a1a3f09bc87f97cfe1c0446f523a7c2f9ae81ce7ae332e3db5ee777e4e
+oid sha256:da75fcbb729830fc5bffe2a3dde01be33068b3c5d1138b6f9e6f17fd94b3dff8
 size 4943162336

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f3a4bd00937ce42817336113fd5335cfb14d61cd88d88f473416b1aa7b27a802
+oid sha256:71e80be36b549dbfc42cbfe4fc54e3b51d0f03a4dc48a7f47646010492d2f314
 size 4999819336

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:54d96cc79ddfa2f761ea508c6b6d3554e83caff9716885c615560f64626cade6
+oid sha256:867985b25124dc090645cae581460d25989b30e7b2f668fc2769f6e68ded1b79
 size 4540516344

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 2.77,
-    "train_loss": 0.5429546391522443,
-    "train_runtime": 799.7979,
+    "train_loss": 0.6124309632513258,
+    "train_runtime": 676.0028,
     "train_samples": 4750,
-    "train_samples_per_second": 2.296,
-    "train_steps_per_second": 0.034
+    "train_samples_per_second": 2.72,
+    "train_steps_per_second": 0.04
 }

trainer_state.json

@@ -10,78 +10,78 @@
   "log_history": [
     {
       "epoch": 0.1,
-      "grad_norm": 12.305452972312954,
+      "grad_norm": 24.191700961175112,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 0.6151,
+      "loss": 0.5581,
       "step": 1
     },
     {
       "epoch": 0.51,
-      "grad_norm": 39.79159513758007,
+      "grad_norm": 12.576350745768869,
       "learning_rate": 1.9659258262890683e-05,
-      "loss": 0.7793,
+      "loss": 0.8334,
       "step": 5
     },
     {
       "epoch": 0.92,
-      "eval_loss": 1.4149478673934937,
-      "eval_runtime": 41.8351,
-      "eval_samples_per_second": 8.223,
-      "eval_steps_per_second": 0.526,
+      "eval_loss": 1.4131343364715576,
+      "eval_runtime": 38.4185,
+      "eval_samples_per_second": 8.954,
+      "eval_steps_per_second": 0.573,
       "step": 9
     },
     {
       "epoch": 1.03,
-      "grad_norm": 3.3298268761353786,
+      "grad_norm": 17.837211332799143,
       "learning_rate": 1.608761429008721e-05,
-      "loss": 0.7329,
+      "loss": 0.738,
       "step": 10
     },
     {
       "epoch": 1.54,
-      "grad_norm": 2.6517437110683546,
+      "grad_norm": 3.797877129136344,
       "learning_rate": 1e-05,
-      "loss": 0.53,
+      "loss": 0.6303,
       "step": 15
     },
     {
       "epoch": 1.95,
-      "eval_loss": 1.4280900955200195,
-      "eval_runtime": 42.1671,
-      "eval_samples_per_second": 8.158,
-      "eval_steps_per_second": 0.522,
+      "eval_loss": 1.4060215950012207,
+      "eval_runtime": 39.241,
+      "eval_samples_per_second": 8.766,
+      "eval_steps_per_second": 0.561,
       "step": 19
     },
     {
       "epoch": 2.05,
-      "grad_norm": 3.365901304289387,
+      "grad_norm": 7.485764754081964,
       "learning_rate": 3.912385709912794e-06,
-      "loss": 0.4284,
+      "loss": 0.5348,
       "step": 20
     },
     {
       "epoch": 2.56,
-      "grad_norm": 4.7684471498019105,
+      "grad_norm": 3.0173543034517323,
       "learning_rate": 3.4074173710931804e-07,
-      "loss": 0.4017,
+      "loss": 0.4647,
       "step": 25
     },
     {
       "epoch": 2.77,
-      "eval_loss": 1.4481348991394043,
-      "eval_runtime": 42.1215,
-      "eval_samples_per_second": 8.167,
-      "eval_steps_per_second": 0.522,
+      "eval_loss": 1.4294962882995605,
+      "eval_runtime": 38.6474,
+      "eval_samples_per_second": 8.901,
+      "eval_steps_per_second": 0.569,
       "step": 27
     },
     {
       "epoch": 2.77,
       "step": 27,
       "total_flos": 5600905789440.0,
-      "train_loss": 0.5429546391522443,
-      "train_runtime": 799.7979,
-      "train_samples_per_second": 2.296,
-      "train_steps_per_second": 0.034
+      "train_loss": 0.6124309632513258,
+      "train_runtime": 676.0028,
+      "train_samples_per_second": 2.72,
+      "train_steps_per_second": 0.04
     }
   ],
   "logging_steps": 5,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bd2ec34fa4d25f81a1a420f99434940129f0d4a5a125492f1dadb1493bd5e669
-size 6264
+oid sha256:9605dab41a6f0bcc6542eda4f5de7efa4b0c431b8dd3c26e5d4c64263bed48ba
+size 6200