masterLan/liujx-78k

GenPRM-78k-train-5:5-decontamination

This model is a fine-tuned version of /data1/model/Qwen2.5-Math-7B-Instruct on the GenPRM-78k-train-5:5-decontamination dataset. It achieves the following results on the evaluation set:

• Loss: 0.2910

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: 4
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• total_eval_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
0.3951	0.0823	100	0.3771
0.3471	0.1647	200	0.3431
0.3295	0.2470	300	0.3266
0.3162	0.3294	400	0.3161
0.3143	0.4117	500	0.3084
0.3054	0.4940	600	0.3029
0.3031	0.5764	700	0.2985
0.2988	0.6587	800	0.2953
0.2965	0.7410	900	0.2932
0.2935	0.8234	1000	0.2918
0.2975	0.9057	1100	0.2911
0.304	0.9881	1200	0.2910

Framework versions

• Transformers 4.45.2
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.20.1