ArvinZhuang / DSI-transformers

A huggingface transformers implementation of "Transformer Memory as a Differentiable Search Index"
MIT License
155 stars 14 forks source link

About the figures of hist@1 and hits@10 #7

Closed liyongqi67 closed 10 months ago

liyongqi67 commented 10 months ago

I notice the code set max_steps=1000000 (1000k). But the figures of hist@1 and hits@10 only illustrate the scores until 120k. Will it continue training until 1000k steps?

ArvinZhuang commented 10 months ago

Hi @liyongqi67 , yes the code will run for 1m steps, It will take a long time to finish, thus I only show 120k in the figures and the improvements are not very significant after this point.

liyongqi67 commented 10 months ago

Thanks for your quick reply!