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/a75ea88447d62fa7_train_data.json
 model_type: AutoModelForCausalLM
@@ -91,7 +91,7 @@ wandb_name: 1e0bf128-7bb1-42b2-a572-3532305a7d37
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 1e0bf128-7bb1-42b2-a572-3532305a7d37
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -102,6 +102,8 @@ xformers_attention: null
 # 1e0bf128-7bb1-42b2-a572-3532305a7d37
 
 This model is a fine-tuned version of [NousResearch/Nous-Hermes-llama-2-7b](https://huggingface.co/NousResearch/Nous-Hermes-llama-2-7b) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 2.0455
 
 ## 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        | 0.0007 | 1    | 2.7259          |
+| No log        | 0.0007 | 1    | 2.7258          |
+| 2.6879        | 0.0068 | 10   | 2.4667          |
+| 2.1009        | 0.0137 | 20   | 2.1361          |
+| 1.9384        | 0.0205 | 30   | 2.0720          |
+| 2.1117        | 0.0274 | 40   | 2.0499          |
+| 1.8624        | 0.0342 | 50   | 2.0455          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:852e3678bd738660a2a36d5d59f746a6e5cbaa32b88b77369bbb3ff754e6c561
+oid sha256:19f39a59bb3a00134ae3b98f6b863a35180945e74097738c73dca69c90c53259
 size 604420306