QwenLM / Qwen-VL

The official repo of Qwen-VL (通义千问-VL) chat & pretrained large vision language model proposed by Alibaba Cloud.
Other
4.27k stars 327 forks source link

💡 [REQUEST] - <title>学习率不改变,有人知道吗? #397

Open xuyiming010912 opened 1 month ago

xuyiming010912 commented 1 month ago

起始日期 | Start Date

No response

实现PR | Implementation PR

No response

相关Issues | Reference Issues

No response

摘要 | Summary

1

基本示例 | Basic Example

{'loss': 1.1777, 'learning_rate': 1e-05, 'epoch': 0.79}
{'loss': 1.0297, 'learning_rate': 1e-05, 'epoch': 0.8}
{'loss': 1.0975, 'learning_rate': 1e-05, 'epoch': 0.81}
{'loss': 1.0213, 'learning_rate': 1e-05, 'epoch': 0.82}
{'loss': 0.9937, 'learning_rate': 1e-05, 'epoch': 0.83}
{'loss': 0.979, 'learning_rate': 1e-05, 'epoch': 0.84}
{'loss': 0.9382, 'learning_rate': 1e-05, 'epoch': 0.85}
{'loss': 0.9343, 'learning_rate': 1e-05, 'epoch': 0.86}
{'loss': 0.8786, 'learning_rate': 1e-05, 'epoch': 0.87}
{'loss': 0.9547, 'learning_rate': 1e-05, 'epoch': 0.88}
{'loss': 0.9132, 'learning_rate': 1e-05, 'epoch': 0.89}
{'loss': 0.8164, 'learning_rate': 1e-05, 'epoch': 0.9}
{'loss': 0.8481, 'learning_rate': 1e-05, 'epoch': 0.91}
{'loss': 0.7252, 'learning_rate': 1e-05, 'epoch': 0.92}
{'loss': 0.7812, 'learning_rate': 1e-05, 'epoch': 0.93}
{'loss': 0.7753, 'learning_rate': 1e-05, 'epoch': 0.94}
{'loss': 0.7889, 'learning_rate': 1e-05, 'epoch': 0.95}
{'loss': 0.734, 'learning_rate': 1e-05, 'epoch': 0.96}
{'loss': 0.7014, 'learning_rate': 1e-05, 'epoch': 0.97}
{'loss': 0.6886, 'learning_rate': 1e-05, 'epoch': 0.98}
{'loss': 0.6154, 'learning_rate': 1e-05, 'epoch': 0.99}
{'loss': 0.677, 'learning_rate': 1e-05, 'epoch': 1.0}

缺陷 | Drawbacks

1

未解决问题 | Unresolved questions

1