Skip to content

Remove the save of gptq config #1993

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
merged 4 commits into from
Sep 3, 2024
Merged

Remove the save of gptq config #1993

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
baa9b5c
support xpu lw forward
Kaihui-intel Aug 20, 2024
8e04e40
Merge branch 'master' of https://github.com/intel/neural-compressor
Kaihui-intel Sep 3, 2024
3f3b239
update gptq config
Kaihui-intel Sep 3, 2024
d969d8c
revert code
Kaihui-intel Sep 3, 2024
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
10 changes: 2 additions & 8 deletions neural_compressor/torch/algorithms/weight_only/gptq.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -930,12 +930,7 @@ def tmp(_, inp, out):

logger.info("Quantization done")
# self.model.config.use_cache = self.use_cache

# obtain model (all weight only quantization API function should return)
for k, v in gptq_config.items():
for m, n in v.items():
gptq_config[k][m] = n.tolist()
return self.model, gptq_config
return self.model


class GPTQ:
Expand Down Expand Up @@ -1379,9 +1374,8 @@ def convert(self, model, *args, **kwargs):
self.gptq_quantizer.model = model
self.gptq_quantizer.remove_prepare_for_calibration()

q_model, gptq_config = self.gptq_quantizer.execute_quantization()
q_model = self.gptq_quantizer.execute_quantization()
if not self.gptq_quantizer.use_layer_wise:
q_model = q_model.to(self.model_device)
q_model.gptq_config = gptq_config
logger.info("GPTQ quantizing done.")
return q_model
5 changes: 0 additions & 5 deletions neural_compressor/torch/algorithms/weight_only/save_load.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -53,11 +53,6 @@ def save(model, output_dir="./saved_results"):
# saving process
save_config_mapping(model.qconfig, qconfig_file_path)

if hasattr(model, "gptq_config") and model.gptq_config:
gptq_config_path = os.path.join(os.path.abspath(os.path.expanduser(output_dir)), "gptq_config.json")
with open(gptq_config_path, "w") as f:
json.dump(model.gptq_config, f, indent=4)

# MethodType 'save' not in state_dict
del model.save
torch.save(model.state_dict(), qmodel_weight_file_path)
Expand Down