Update example command lines in the module docstring of train_sft.py to remove weight_dtype=bfloat16 argument; this shouldn't be the default weight_dtype for SFT because of potential loss of gradient updates w/master weights in bfloat16. by copybara-service[bot] · Pull Request #4710 · AI-Hypercomputer/maxtext

✅ All contributors are covered under a CLA with Google

See https://cla.developers.google.com/ for more info about Google's Contributor License Agreement (CLA).

ℹ️ Googlers: Go here to view more details and manage scans for this pull request.

@copybara-service

The following contributors were found for this pull request:

✅ 5ae391c PR Opener: @copybara-service[bot]
✅ 5ae391c Author: @Google-ML-Automation <googl**********ation@google.com>

(Only the first commit for a unique contributor is listed.)