Qwen3-30B-A3B-Base-medqa-seed-42

This model is a fine-tuned version of Qwen/Qwen3-30B-A3B-Base on the medqa dataset. It achieves the following results on the evaluation set:

• Loss: 0.0266

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: 0.0002
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 64
• gradient_accumulation_steps: 8
• total_train_batch_size: 1024
• total_eval_batch_size: 128
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss
2.8757	0.2326	10	2.6217
0.7735	0.4651	20	0.6109
0.0325	0.6977	30	0.0351
0.0289	0.9302	40	0.0294
0.0215	1.1628	50	0.0276
0.021	1.3953	60	0.0267
0.0182	1.6279	70	0.0258
0.021	1.8605	80	0.0254
0.0123	2.0930	90	0.0257
0.0149	2.3256	100	0.0269
0.012	2.5581	110	0.0269
0.0102	2.7907	120	0.0268

Framework versions

• PEFT 0.15.2
• Transformers 4.52.3
• Pytorch 2.7.0+cu126
• Datasets 3.6.0
• Tokenizers 0.21.1