End of training

README.md

@@ -65,7 +65,7 @@ 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/cf29151880ad1480_train_data.json
 model_type: AutoModelForCausalLM
@@ -92,7 +92,7 @@ wandb_name: d10b2049-61ce-4cc5-abbb-07f7e605260e
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: d10b2049-61ce-4cc5-abbb-07f7e605260e
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -103,6 +103,8 @@ xformers_attention: null
 # 4ebd7bc1-1d10-87dc-95fa-71a0d9543354
 
 This model is a fine-tuned version of [JackFram/llama-160m](https://huggingface.co/JackFram/llama-160m) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 2.7482
 
 ## Model description
 
@@ -129,14 +131,19 @@ 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    | 2.7642          |
+| 2.8559        | 0.0002 | 10   | 2.7627          |
+| 2.7007        | 0.0005 | 20   | 2.7559          |
+| 2.2824        | 0.0007 | 30   | 2.7507          |
+| 2.6814        | 0.0009 | 40   | 2.7486          |
+| 2.6367        | 0.0012 | 50   | 2.7482          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ba3fc75d79b76f8cb88ef91d4d5fedfcb0542ea0b7a80c10a83d00df1cf45527
+oid sha256:de8e70cf9f33bf437954405952b6c36a2c299e9e032f7bdc689735bf28251d59
 size 6843306