ProceduralTree/chinese_qa_model

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

chineseqamodel

This model is a fine-tuned version of bert-base-chinese on the None dataset. It achieves the following results on the evaluation set:

• —Loss: 0.5698
• —Accuracy: 0.7401

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: 2e-05
• —trainbatchsize: 4
• —evalbatchsize: 4
• —seed: 42
• —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• —lrschedulertype: linear
• —num_epochs: 2

Training results

Framework versions

• —Transformers 4.25.1
• —Pytorch 1.13.0+cu116
• —Datasets 2.8.0
• —Tokenizers 0.13.2