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/21ccb12743019c0f_train_data.json
 model_type: AutoModelForCausalLM
@@ -90,7 +90,7 @@ wandb_name: fabd9608-29c3-43da-bc76-bba7268e1ba1
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: fabd9608-29c3-43da-bc76-bba7268e1ba1
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -101,6 +101,8 @@ xformers_attention: null
 # ebe2c396-1285-41be-807a-0d8f0f01209b
 
 This model is a fine-tuned version of [NousResearch/Hermes-2-Pro-Llama-3-8B](https://huggingface.co/NousResearch/Hermes-2-Pro-Llama-3-8B) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.6257
 
 ## Model description
 
@@ -127,14 +129,19 @@ 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        | 0.0025 | 1    | 3.9906          |
+| 3.5027        | 0.0254 | 10   | 2.7543          |
+| 1.2498        | 0.0508 | 20   | 0.9245          |
+| 0.5926        | 0.0762 | 30   | 0.7531          |
+| 0.7188        | 0.1017 | 40   | 0.6442          |
+| 0.5736        | 0.1271 | 50   | 0.6257          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e2cbbc6f31a2256bedfcf89d12d0b3c08defed674a158a1c9777c91e85b534f0
+oid sha256:1ae0c5f633c8c190e22cf17f13c11d6b45a2fe3ef135f73f8bc23236ebce1222
 size 84047370