chinese_chitchat

This model is a fine-tuned version of qgyd2021/chinese_chitchat on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.1314

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: 0.0002
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• total_eval_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10000
• num_epochs: 40.0

Training results

Training Loss	Epoch	Step	Validation Loss
1.5203	0.29	1000	2.0882
1.4243	0.58	2000	2.1525
1.3502	0.86	3000	2.1544
1.5332	1.15	4000	2.0826
1.5208	1.44	5000	2.0789
1.5521	1.73	6000	2.0613
1.5634	2.02	7000	2.1124
1.5067	2.3	8000	2.1014
1.5573	2.59	9000	2.0972
1.5949	2.88	10000	2.0907
1.5491	3.17	11000	2.1314

Framework versions

• Transformers 4.33.0
• Pytorch 2.0.0
• Datasets 2.1.0
• Tokenizers 0.13.3