cutelemonlili/Qwen2.5-Math-1.5B_Teacher_forget_RL_data_QwQ-Preview

Teacher_forget_RL_data_QwQ-Preview

This model is a fine-tuned version of Qwen/Qwen2.5-Math-1.5B on the Teacher_forget_RL_data_QwQ-Preview dataset. It achieves the following results on the evaluation set:

• Loss: 0.4022

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: 1e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 4
• total_eval_batch_size: 4
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
0.4509	0.1650	100	0.4296
0.3338	0.3300	200	0.4039
0.3784	0.4950	300	0.3926
0.4608	0.6601	400	0.3853
0.2662	0.8251	500	0.3799
0.3289	0.9901	600	0.3764
0.1755	1.1551	700	0.3848
0.1973	1.3201	800	0.3817
0.1901	1.4851	900	0.3803
0.2886	1.6502	1000	0.3795
0.2431	1.8152	1100	0.3795
0.2318	1.9802	1200	0.3781
0.2471	2.1452	1300	0.3996
0.2475	2.3102	1400	0.4021
0.1613	2.4752	1500	0.4004
0.2302	2.6403	1600	0.4018
0.2446	2.8053	1700	0.4019
0.1505	2.9703	1800	0.4021

Framework versions

• Transformers 4.46.1
• Pytorch 2.6.0+cu124
• Datasets 3.1.0
• Tokenizers 0.20.3