SFT-Qwen2.5-Coder-3B_v1.1

This model is a fine-tuned version of Qwen/Qwen2.5-Coder-3B-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7507

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0001
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.PAGED_ADAMW_8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
0.9928	0.2857	20	0.9394
1.0199	0.5714	40	0.8743
0.838	0.8571	60	0.8337
0.835	1.1429	80	0.8078
0.7992	1.4286	100	0.7930
0.7439	1.7143	120	0.7782
0.696	2.0	140	0.7625
0.6318	2.2857	160	0.7599
0.6666	2.5714	180	0.7571
0.8047	2.8571	200	0.7507

Framework versions

• PEFT 0.18.0
• Transformers 4.57.1
• Pytorch 2.9.0+cu126
• Datasets 4.4.1
• Tokenizers 0.22.1