Skip to content

add multi-gpu support #394

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
mike-ferguson merged 1 commit into from
Mar 16, 2026
Merged

add multi-gpu support #394

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
24 changes: 21 additions & 3 deletions brainscore_language/model_helpers/huggingface.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -46,16 +46,34 @@ def __init__(
self.model_id = model_id
self.use_localizer = use_localizer
self.region_layer_mapping = region_layer_mapping
self.basemodel = (model if model is not None else AutoModelForCausalLM.from_pretrained(self.model_id))
if torch.backends.mps.is_available():

multi_gpu = (model is None
and torch.cuda.is_available()
and torch.cuda.device_count() > 1)

if model is not None:
self.basemodel = model
elif multi_gpu:
self.basemodel = AutoModelForCausalLM.from_pretrained(
self.model_id, low_cpu_mem_usage=True, device_map='auto')
else:
self.basemodel = AutoModelForCausalLM.from_pretrained(
self.model_id, low_cpu_mem_usage=True)

if multi_gpu:
self.device = 'cuda'
self._logger.info(f"Using device_map='auto' across {torch.cuda.device_count()} GPUs")
print(f"Using device_map='auto' across {torch.cuda.device_count()} GPUs")
elif torch.backends.mps.is_available():
self.device = 'mps'
self.basemodel.to(self.device)
elif torch.cuda.is_available():
self.device = 'cuda'
self.basemodel.to(self.device)
else:
self.device = 'cpu'
self._logger.info(f"Using device: {self.device}")
print(f"Using device: {self.device}")
self.basemodel.to(self.device)
self.tokenizer = tokenizer if tokenizer is not None else AutoTokenizer.from_pretrained(self.model_id,
truncation_side='left')
self.current_tokens = None # keep track of current tokens
Expand Down
Loading