qwen25_7b_rm_eng_5e5_5ep

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: 0.8016
• Accuracy: 0.8067

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-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 8
• total_train_batch_size: 256
• total_eval_batch_size: 32
• 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: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.5809	0.3347	10	0.8320	0.5767
0.8499	0.6695	20	0.6577	0.5733
0.6563	1.0335	30	0.5270	0.6833
0.5003	1.3682	40	0.4127	0.77
0.446	1.7029	50	0.4154	0.79
0.4317	2.0669	60	0.4063	0.79
0.3174	2.4017	70	0.3632	0.84
0.3063	2.7364	80	0.4297	0.7967
0.2931	3.1004	90	0.3729	0.8133
0.1145	3.4351	100	0.7719	0.8067
0.0974	3.7699	110	0.6630	0.7833
0.074	4.1339	120	0.6148	0.7967
0.0348	4.4686	130	0.7528	0.8
0.0274	4.8033	140	0.8016	0.8067

Framework versions

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