chinese-bert-wwm-finetuned-product-1

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

• eval_loss: 0.0000
• eval_runtime: 10.6737
• eval_samples_per_second: 362.572
• eval_steps_per_second: 5.715
• epoch: 11.61
• step: 18797

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: 256
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Framework versions

• Transformers 4.17.0
• Pytorch 1.6.0
• Datasets 2.0.0
• Tokenizers 0.11.6