finalform/foamQwen3-8B

factory_qwen_results

This model is a fine-tuned version of Qwen/Qwen3-8B on the train dataset. It achieves the following results on the evaluation set:

• Loss: 0.2496
• Accuracy: 0.9448

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
• distributed_type: multi-GPU
• num_devices: 3
• gradient_accumulation_steps: 4
• total_train_batch_size: 24
• total_eval_batch_size: 6
• 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: 0.03
• num_epochs: 9.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.3591	1.0	42	0.3988	0.9035
0.2673	2.0	84	0.3260	0.9197
0.2425	3.0	126	0.2898	0.9289
0.2069	4.0	168	0.2659	0.9346
0.138	5.0	210	0.2525	0.9391
0.1251	6.0	252	0.2497	0.9431
0.1072	7.0	294	0.2475	0.9439
0.1059	8.0	336	0.2483	0.9446
0.1073	8.7904	369	0.2496	0.9448

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.7.1+cu126
• Datasets 3.6.0
• Tokenizers 0.21.1