qwen25_7b_rm_eng_5e6_2ep

This model is a fine-tuned version of Jennny/qwen25_7b_sft_eng_math on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.2623
• Accuracy: 0.71

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: 5e-06
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• total_eval_batch_size: 16
• optimizer: Use OptimizerNames.PAGED_ADAMW 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: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.5427	0.1677	10	0.6014	0.6733
0.0398	0.3354	20	1.1002	0.64
0.0221	0.5031	30	1.1894	0.64
0.0386	0.6709	40	0.8497	0.6667
0.045	0.8386	50	1.7023	0.58
0.0474	1.0	60	1.5256	0.6233
0.0054	1.1677	70	1.4504	0.6567
0.007	1.3354	80	1.6286	0.65
0.0264	1.5031	90	1.6946	0.6733
0.0656	1.6709	100	1.3171	0.67
0.044	1.8386	110	0.8320	0.7133
0.0737	2.0	120	0.6218	0.75
0.0035	2.1677	130	0.8807	0.7133
0.0002	2.3354	140	1.1659	0.71
0.0003	2.5031	150	1.2517	0.7067
0.0006	2.6709	160	1.2588	0.71
0.0016	2.8386	170	1.2623	0.71

Framework versions

• Transformers 4.49.0
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.1