qwen_orpo

This model is a fine-tuned version of /mnt/pfs-guan-ssai/nlu/data/tianxy/MODELS/qwen/Qwen1.5-7B-Chat on the orpo dataset.

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
• trainbatchsize: 1
• evalbatchsize: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 24
• gradientaccumulationsteps: 4
• totaltrainbatch_size: 96
• totalevalbatch_size: 192
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lrschedulertype: cosine
• lrschedulerwarmup_steps: 0.1
• num_epochs: 3.0

Training results

Framework versions

• Transformers 4.41.0
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.19.1