search

allenai / RL4LMs

A modular RL library to fine-tune language models to human preferences
https://rl4lms.apps.allenai.org/
Apache License 2.0
2.13k stars 191 forks source link

CPU Support Minor Bug #59

Open tedmoskovitz opened 1 year ago

tedmoskovitz commented 1 year ago

Hello, I believe I found a minor bug in IntentAccuracyDailyDialog, lines 672-3 in envs/text_generation/metric.py. The device is currently set with the following two lines:

self._device = "cuda" if torch.cuda.is_available() else "cpu"
self._device = f"cuda:{torch.cuda.device_count() - 1}"

However, I believe it should be such that the device is f"cuda:{torch.cuda.device_count() - 1}" if on GPU and "cpu" otherwise (as currently it can never be set to CPU, and instead tries to find "cuda:-1"). Thanks so much for putting this library together!