Leon97ZJU/sft_qwen_base

qwen

This model is a fine-tuned version of /remote-home1/yli/Model/Generator/Qwen2.5/7B/base on the 2wikimultihopqa_train, the hotpotqa_train and the musique_train datasets. It achieves the following results on the evaluation set:

• Loss: 0.1620

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: 7e-06
• train_batch_size: 4
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• total_eval_batch_size: 8
• 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
0.1654	0.3994	500	0.1538
0.1531	0.7987	1000	0.1395
0.0938	1.1981	1500	0.1406
0.0948	1.5974	2000	0.1360
0.0844	1.9968	2500	0.1315
0.0418	2.3962	3000	0.1611
0.0379	2.7955	3500	0.1621

Framework versions

• Transformers 4.46.1
• Pytorch 2.4.0
• Datasets 3.1.0
• Tokenizers 0.20.3