End of training

README.md

@@ -64,7 +64,7 @@ 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/b897f8023b3b08e7_train_data.json
 model_type: AutoModelForCausalLM
@@ -89,7 +89,7 @@ wandb_name: 88195d73-80bc-492e-98e7-652dbc88caca
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 88195d73-80bc-492e-98e7-652dbc88caca
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -100,6 +100,8 @@ xformers_attention: null
 # a7b1257d-bef1-e991-0e8a-b810068584ec
 
 This model is a fine-tuned version of [furiosa-ai/mlperf-gpt-j-6b](https://huggingface.co/furiosa-ai/mlperf-gpt-j-6b) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 1.9041
 
 ## Model description
 
@@ -126,14 +128,19 @@ 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        | 0.0003 | 1    | 2.7202          |
+| 10.8068       | 0.0032 | 10   | 2.5562          |
+| 8.3174        | 0.0064 | 20   | 2.0805          |
+| 7.674         | 0.0096 | 30   | 1.9492          |
+| 8.0616        | 0.0129 | 40   | 1.9112          |
+| 7.3475        | 0.0161 | 50   | 1.9041          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6c8b4596c21376165242e2469958d38a7488a4ea5b86cbbf82dd822d354caed2
+oid sha256:3a7423cdf0648a5fbf027f5b684e12500927ffb6ddb6fcef69f9e35f9e832100
 size 66180362