syspider
Results
4
comments of
syspider
生成的时候可以使用past_key_values机制加速。博主生成的时候每一次输入都是整个序列，改进后能快不少
他复用了GPT2LMHeadModel这个类里的loss，这里面计算loss的时候已经错位了，你可以看源码
> Yes, the training data is under preparation. We will release it soon. quite excited
All the samples seem to have the same problem when the data source is "twitter_post"