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/2f18748c78791ad7_train_data.json
 model_type: AutoModelForCausalLM
@@ -91,7 +91,7 @@ wandb_name: 27406a9d-a68f-46a0-b4f4-aa3e9c20a557
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 27406a9d-a68f-46a0-b4f4-aa3e9c20a557
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -102,6 +102,8 @@ xformers_attention: null
 # 3ec58508-0e15-f966-dd29-315ed4d3f36b
 
 This model is a fine-tuned version of [NousResearch/Meta-Llama-3-8B-Alternate-Tokenizer](https://huggingface.co/NousResearch/Meta-Llama-3-8B-Alternate-Tokenizer) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.2112
 
 ## 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        | 0.0019 | 1    | 0.3210          |
+| 0.3558        | 0.0191 | 10   | 0.2864          |
+| 0.2604        | 0.0381 | 20   | 0.2270          |
+| 0.3168        | 0.0572 | 30   | 0.2162          |
+| 0.2392        | 0.0763 | 40   | 0.2120          |
+| 0.1954        | 0.0954 | 50   | 0.2112          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:13656f5e168741681fd885263404b4da955dc55b0693e1c8ff99bb7e9cc65137
+oid sha256:ce3bcd1b6af818ff091fb8d62be6e0e94d7cc3a5f9c5f8c15f085220b4343e9e
 size 84047370