search

Kipok / NeMo-Skills

A pipeline to improve skills of large language models
https://kipok.github.io/NeMo-Skills/
Apache License 2.0
185 stars 41 forks source link

Optimizing SFT recipe #82

Closed Kipok closed 3 months ago

Kipok commented 3 months ago

Adding more optimal parameters for training llama3 8B model + fixing a bug with incorrect timeout

Kipok commented 3 months ago

Let me merge this one as it contains some important fixes. We can follow up with another PR if there are any problems.