afalf
afalf
您好，请问一下具体是哪个模型呢？ROM语义相关性模型的基础底座是达摩院(structbert)[https://modelscope.cn/models/damo/nlp_structbert_backbone_base_std/summary], 并不是原版的BERT，两者的词表存在一定差异。
已经定位问题，请在调用ROM模型时指定model_revision 为“v1.1.0”。
> The result I got was only 65. I don't know what was wrong. Sorry, I have already seen it. Could you please show your hyparameters for training?
> Right from this is seems like we should update the scores on the leaderboard with the new reproducible scores. Since the authors has been made aware (issue on NVIDIA...
@AlexeyVatolin Thansk a lot! We have used execution with normalized. Sorry for these mirrors in our scripts.
> I have reviewed the PR and everything looks good. @afalf you might want to resubmit the scores using the new scores given the improvements. Okay, we will update the...