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/c274d3f9f29134e6_train_data.json
 model_type: AutoModelForCausalLM
@@ -91,7 +91,7 @@ wandb_name: 600f7b65-8c53-44d5-86ec-0a32e02dbb7c
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 600f7b65-8c53-44d5-86ec-0a32e02dbb7c
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -102,6 +102,8 @@ xformers_attention: null
 # 427437c1-919c-4323-99b1-975644ec4c65
 
 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: 2.7248
 
 ## 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        | 0.0006 | 1    | 3.3249          |
+| No log        | 0.0006 | 1    | 3.3250          |
+| 3.3191        | 0.0056 | 10   | 3.2550          |
+| 3.0077        | 0.0113 | 20   | 2.8879          |
+| 2.8783        | 0.0169 | 30   | 2.7702          |
+| 2.7063        | 0.0226 | 40   | 2.7314          |
+| 2.773         | 0.0282 | 50   | 2.7248          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4bbd3048fbf9f0f7505dc002056f2cf4aa6b4c7125734661a9d58948d5284baf
+oid sha256:e93912c04b0f57cbf4d0ea74ea12cf0c2e721b7ec590dc53040d2c63ea9690f9
 size 604420306