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/128d74f7f30eb851_train_data.json
 model_type: AutoModelForCausalLM
@@ -91,7 +91,7 @@ wandb_name: e12ffd4b-267d-4361-acd1-56e79665ba15
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: e12ffd4b-267d-4361-acd1-56e79665ba15
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -102,6 +102,8 @@ xformers_attention: null
 # 4ac1f2c3-731d-04e1-5ae1-3c6b7cb82beb
 
 This model is a fine-tuned version of [NousResearch/Hermes-3-Llama-3.1-8B](https://huggingface.co/NousResearch/Hermes-3-Llama-3.1-8B) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.2424
 
 ## Model description
 
@@ -128,14 +130,19 @@ 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    | 0.3730          |
+| 0.4345        | 0.0004 | 10   | 0.3391          |
+| 0.2963        | 0.0009 | 20   | 0.2872          |
+| 0.3313        | 0.0013 | 30   | 0.2549          |
+| 0.2518        | 0.0018 | 40   | 0.2442          |
+| 0.268         | 0.0022 | 50   | 0.2424          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c64e829015f997ff4089b1311d8621d080395efe4e9a178292a502ab5d9ae897
+oid sha256:c24977d2fa307d11d5f9e3024b5bb7322a8188886351c52bed6bac63c3355801
 size 84047370