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: 50
+max_steps: 1
 micro_batch_size: 2
 mlflow_experiment_name: /tmp/fe7846cfa624d72d_train_data.json
 model_type: AutoModelForCausalLM
@@ -93,7 +93,7 @@ wandb_name: 3943b24b-8e8c-4496-973b-5e7477a15e9e
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 3943b24b-8e8c-4496-973b-5e7477a15e9e
-warmup_steps: 10
+warmup_steps: 1
 weight_decay: 0.0
 xformers_attention: null
 
@@ -104,8 +104,6 @@ xformers_attention: null
 # 3943b24b-8e8c-4496-973b-5e7477a15e9e
 
 This model is a fine-tuned version of [NousResearch/Meta-Llama-3-8B](https://huggingface.co/NousResearch/Meta-Llama-3-8B) on the None dataset.
-It achieves the following results on the evaluation set:
-- Loss: 0.7487
 
 ## Model description
 
@@ -132,19 +130,14 @@ 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: 10
-- training_steps: 50
+- lr_scheduler_warmup_steps: 2
+- training_steps: 1
 
 ### Training results
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
 | No log        | 0.0008 | 1    | 1.5018          |
-| 1.3659        | 0.0077 | 10   | 1.1262          |
-| 0.8855        | 0.0155 | 20   | 0.8373          |
-| 0.799         | 0.0232 | 30   | 0.7749          |
-| 0.762         | 0.0309 | 40   | 0.7536          |
-| 0.7108        | 0.0387 | 50   | 0.7487          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5853405f6916ff2bda2c565c9f7ecdec331ccb1a35f45ddfffb788aa1ae2cb01
+oid sha256:675138f0a3d7a76a482679bd312d176040691b671d68d55c2d8f8074b46b17d8
 size 84047370