Wayne Yuan

Results 4 issues of Wayne Yuan

### Is there an existing issue for this? - [X] I have searched the existing issues ### Current Behavior 原code的ptuning里的train.sh脚本里没有加入prefix_projection参数,这样默认是通过p-tuning训练吗? 我尝试增加参数 --prefix_projection True 但是出现训练报错 显存不够?batch缩到很小也不行。 ### Expected Behavior _No response_...

非常优秀的工作👍 目前我面临的问题是: 1、用垂直语料进行continual training,然后拿收敛的model进行fine-tuning发现效果很差,甚至比原始开源模型还差了不少?所以是否是不需要continual training到收敛才选择checkpoint? 2、训练的step和评估的指标如下图,感觉评估的精度随step一直在波动变化,没有呈现一定的规律性?请问这个可能的原因是什么呢? ![Uploading performance_metrics.png…]()

I see that sgpt-bloom-7b1-mamarco model has a vector length of 300,but If I input more than the maximum length, for example, input more than 400 Chinese characters, it seems that...

I want to customize the adapt function, change the language to Chinese this is code: ```py cache_dir = ".cache" adapt( metrics=evaluate_stk, language="Chinese", cache_dir=cache_dir, llm=llm ) ``` and this is the...

bug