bert-base-chinese-finetuned-amazon_zh_20000

This model is a fine-tuned version of bert-base-chinese on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1683
• Accuracy: 0.5224
• F1: 0.5194

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
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
1.2051	1.0	2500	1.1717	0.506	0.4847
1.0035	2.0	5000	1.1683	0.5224	0.5194

Framework versions

• Transformers 4.15.0
• Pytorch 1.9.1
• Datasets 1.18.3
• Tokenizers 0.10.3