search

bigscience-workshop / t-zero

Reproduce results and replicate training fo T0 (Multitask Prompted Training Enables Zero-Shot Task Generalization)
Apache License 2.0
456 stars 53 forks source link

Instructions for fine-tuning a trained t-zero #8

Closed louis030195 closed 2 years ago

louis030195 commented 2 years ago

First, thanks for the great work!

Does it make sense to request instructions for fine-tuning a trained t-zero? That would be really great. If it is outside the scope of this repo, feel free to close this issue.

If yes, I would suggest providing the computing power required, thanks!

VictorSanh commented 2 years ago

hi @louis030195! I think you will find this PR #3 interesting. it supports single task fine-tuning (including the few-shot case).

If yes, I would suggest providing the computing power required, thanks!

If you mean, documenting how much resources (in number of GPUs for instance) are needed to fine-tune on a dataset of size XXX, yes we should do that!

VictorSanh commented 2 years ago

pr #3 is merged