Skip to content

fix: use element_size #140

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
kmehant merged 1 commit into from
Apr 23, 2025
Merged

fix: use element_size #140

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
6 changes: 3 additions & 3 deletions plugins/accelerated-peft/src/fms_acceleration_peft/gptqmodel/models/base.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,7 +33,6 @@
PreTrainedModel,
)
from transformers.modeling_utils import (
dtype_byte_size,
is_local_dist_rank_0,
no_init_weights,
)
Expand Down Expand Up @@ -829,8 +828,9 @@ def shard_checkpoint(
block_id = storage_id_to_block[storage_id]
sharded_state_dicts[block_id][key] = weight
continue

weight_size = weight.numel() * dtype_byte_size(weight.dtype)
# dtype_byte_size is no more supported in transformers
# due to its inaccuracies - https://github.com/huggingface/transformers/pull/37144
weight_size = weight.numel() * weight.element_size()
# If this weight is going to tip up over the maximal size, we split, but only if we have put at least one
# weight in the current shard.
if (
Expand Down