End of training

README.md

@@ -66,7 +66,7 @@ lora_model_dir: null
 lora_r: 8
 lora_target_linear: true
 lr_scheduler: cosine
-max_steps: 1
+max_steps: 50
 micro_batch_size: 2
 mlflow_experiment_name: /tmp/b9a70c6c53b91cc1_train_data.json
 model_type: AutoModelForCausalLM
@@ -93,7 +93,7 @@ wandb_name: b2642ded-964c-4561-a3e2-e9fdd938c7f8
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: b2642ded-964c-4561-a3e2-e9fdd938c7f8
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -104,6 +104,8 @@ xformers_attention: null
 # b2642ded-964c-4561-a3e2-e9fdd938c7f8
 
 This model is a fine-tuned version of [openlm-research/open_llama_3b](https://huggingface.co/openlm-research/open_llama_3b) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 1.1437
 
 ## Model description
 
@@ -130,14 +132,19 @@ The following hyperparameters were used during training:
 - total_train_batch_size: 8
 - optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
 - lr_scheduler_type: cosine
-- lr_scheduler_warmup_steps: 2
-- training_steps: 1
+- lr_scheduler_warmup_steps: 10
+- training_steps: 50
 
 ### Training results
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
 | No log        | 0.0000 | 1    | 1.1904          |
+| 1.2073        | 0.0002 | 10   | 1.1881          |
+| 1.226         | 0.0003 | 20   | 1.1721          |
+| 0.9136        | 0.0005 | 30   | 1.1538          |
+| 1.1429        | 0.0006 | 40   | 1.1452          |
+| 0.9416        | 0.0008 | 50   | 1.1437          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c016110b5d10dd487358dccff4426849b5920791a1b0a5a33f007c40adece202
+oid sha256:f058d690560036fdfbf2512d674dd0007b54cab595d0d01c0ca5a422dfec8a77
 size 50982842