Skip to content

Commit

Permalink
Revert batch_size to feasible number
Browse files Browse the repository at this point in the history
  • Loading branch information
Peter McAughan committed Jul 10, 2023
1 parent a55e6b3 commit 447efb0
Showing 1 changed file with 2 additions and 1 deletion.
3 changes: 2 additions & 1 deletion torchbenchmark/models/hf_Falcon_7b/__init__.py
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,8 @@
class Model(HuggingFaceModel):
task = NLP.LANGUAGE_MODELING
# Published training batch size is 2304: see https://huggingface.co/tiiuae/falcon-7b/blob/main/README.md
DEFAULT_TRAIN_BSIZE = 2304
# Setting to default value of 4 for feasibility
DEFAULT_TRAIN_BSIZE = 4
DEFAULT_EVAL_BSIZE = 1

def __init__(self, test, device, jit=False, batch_size=None, extra_args=[]):
Expand Down

0 comments on commit 447efb0

Please sign in to comment.