shibing624 / MedicalGPT

MedicalGPT: Training Your Own Medical GPT Model with ChatGPT Training Pipeline. 训练医疗大模型,实现了包括增量预训练(PT)、有监督微调(SFT)、RLHF、DPO、ORPO。
Apache License 2.0
2.94k stars 452 forks source link

colab乱码,求大佬解答 #299

Closed small-white-zs closed 6 months ago

small-white-zs commented 6 months ago

/ 大佬,我运行您colab的代码的时候,乱码了, ![Uploading 1704540301190.png…]() 正常是您的这样 ![Uploading 1704540342999.jpg…]()

small-white-zs commented 6 months ago

1704540301190 ![Uploading 1704540342999.jpg…]()

small-white-zs commented 6 months ago

屏幕截图 2024-01-06 193302

shibing624 commented 6 months ago

没有乱码,最后有文本。可以打开--group_by_length True 就没有padding操作。

small-white-zs commented 6 months ago

大佬好,我没表述清楚,后面的第二张截图是您正常运行时的截图(有文本的),第一张截图是我的(出错的)。 然后按照您的方法加了--group_by_length True 后 屏幕截图 2024-01-06 201657

shibing624 commented 6 months ago

https://github.com/shibing624/MedicalGPT/commit/3a7922a600cd044d698d290cb9c4e6d6c5d2d85a fixed

shibing624 commented 5 months ago

https://github.com/shibing624/MedicalGPT/commit/9c9eaf16ac9287124f70edc2edb1827b43440c2f updated.