Skip to content

Fix bnb regression due to empty state dict #36663

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
LysandreJik merged 2 commits into from
Mar 12, 2025
Merged

Fix bnb regression due to empty state dict #36663

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
10d904c
fix
SunMarc Mar 12, 2025
416f09c
Merge remote-tracking branch 'upstream/main' into fix-state-dict-bnb
SunMarc Mar 12, 2025
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
13 changes: 10 additions & 3 deletions src/transformers/modeling_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -833,7 +833,7 @@ def _load_state_dict_into_meta_model(
bin_state_dict = None
if shard_file.endswith(".safetensors"):
file_pointer = safe_open(shard_file, framework="pt", device=tensor_device)
else:
elif shard_file.endswith(".bin"):
map_location = "cpu"
if (
device_map is not None
Expand All @@ -848,6 +848,13 @@ def _load_state_dict_into_meta_model(

is_quantized = hf_quantizer is not None

# get full state dict
if is_quantized:
if shard_file.endswith(".safetensors"):
full_state_dict = load_state_dict(shard_file, map_location="cpu")
elif shard_file.endswith(".bin"):
full_state_dict = bin_state_dict

for serialized_param_name, empty_param in state_dict.items():
# serialized_param_name is the raw, serialized name
# fixed_param_name is the model's equivalent
Expand Down Expand Up @@ -912,7 +919,7 @@ def _load_state_dict_into_meta_model(
model,
param,
fixed_param_name,
state_dict,
full_state_dict,
param_device=param_device,
device_map=device_map,
)
Expand All @@ -928,7 +935,7 @@ def _load_state_dict_into_meta_model(
)
else:
hf_quantizer.create_quantized_param(
model, param, fixed_param_name, param_device, state_dict, unexpected_keys
model, param, fixed_param_name, param_device, full_state_dict, unexpected_keys
)
# For quantized modules with FSDP/DeepSpeed Stage 3, we need to quantize the parameter on the GPU
# and then cast it to CPU to avoid excessive memory usage on each GPU
Expand Down