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/0e19c81b2d1bfc0d_train_data.json
 model_type: AutoModelForCausalLM
@@ -91,7 +91,7 @@ wandb_name: 84f4338b-1c2a-4390-947c-9ec339075319
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 84f4338b-1c2a-4390-947c-9ec339075319
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -102,6 +102,8 @@ xformers_attention: null
 # 2a51a0c5-f44c-55e9-aaa5-d4d4cc6f1c2e
 
 This model is a fine-tuned version of [Qwen/Qwen2.5-1.5B](https://huggingface.co/Qwen/Qwen2.5-1.5B) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 3.2683
 
 ## 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        | 0.0008 | 1    | 3.8918          |
+| 3.8075        | 0.0078 | 10   | 3.7259          |
+| 3.1717        | 0.0156 | 20   | 3.3650          |
+| 3.0412        | 0.0234 | 30   | 3.2844          |
+| 3.2979        | 0.0312 | 40   | 3.2724          |
+| 3.4014        | 0.0390 | 50   | 3.2683          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:686e334c5911633fd8a49acaf76e216e6a92808d5428c6a84c06cf520830141c
+oid sha256:3d1a7b2d9f888c8cc85fd1567e98462ca3d857f316c466afccc6ff6048b43fec
 size 37070634