@codex Check this PR for bugs

@Qubitium Please check error during save, can be related to this PR:
main branch hash: 3da0344

INFO  Format: Converting GPTQ v2 to v1                                                                                     
Traceback (most recent call last):
  File "/home/ubuntu/Documents/Quantize/quantize-glm4.5-air-gptqmodel-clean.py", line 63, in <module>
    model.save(OUTPUT_DIR)
    ~~~~~~~~~~^^^^^^^^^^^^
  File "/home/ubuntu/venvs/gptqmodelt/lib/python3.13t/site-packages/gptqmodel/models/base.py", line 832, in save
    self.save_quantized(
    ~~~~~~~~~~~~~~~~~~~^
        save_dir=save_dir,
        ^^^^^^^^^^^^^^^^^^
    ...<2 lines>...
        meta_quantizer=meta_quantizer,
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
        eora_path=eora_path)
        ^^^^^^^^^^^^^^^^^^^^
  File "/home/ubuntu/venvs/gptqmodelt/lib/python3.13t/site-packages/gptqmodel/models/writer.py", line 231, in save_quantized
    model = convert_gptq_v2_to_v1_format(
        model, quantize_config=quantize_config, qlinear_kernel=self.qlinear_kernel
    )
  File "/home/ubuntu/venvs/gptqmodelt/lib/python3.13t/site-packages/gptqmodel/utils/model.py", line 669, in convert_gptq_v2_to_v1_format
    convert_gptq_v2_to_v1_format_module(module=submodule, quantize_config=quantize_config)
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ubuntu/venvs/gptqmodelt/lib/python3.13t/site-packages/gptqmodel/utils/model.py", line 635, in convert_gptq_v2_to_v1_format_module
    module.qzeros.data[:, range(0, module.qzeros.data.shape[1], 3)] -= (
    ~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: Inplace update to inference tensor outside InferenceMode is not allowed.You can make a clone to get a normal tensor before doing inplace update.See https://github.com/pytorch/rfcs/pull/17 for more details.

@Qubitium Please check error during save, can be related to this PR: main branch hash: 3da0344

INFO  Format: Converting GPTQ v2 to v1                                                                                     
Traceback (most recent call last):
  File "/home/ubuntu/Documents/Quantize/quantize-glm4.5-air-gptqmodel-clean.py", line 63, in <module>
    model.save(OUTPUT_DIR)
    ~~~~~~~~~~^^^^^^^^^^^^
  File "/home/ubuntu/venvs/gptqmodelt/lib/python3.13t/site-packages/gptqmodel/models/base.py", line 832, in save
    self.save_quantized(
    ~~~~~~~~~~~~~~~~~~~^
        save_dir=save_dir,
        ^^^^^^^^^^^^^^^^^^
    ...<2 lines>...
        meta_quantizer=meta_quantizer,
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
        eora_path=eora_path)
        ^^^^^^^^^^^^^^^^^^^^
  File "/home/ubuntu/venvs/gptqmodelt/lib/python3.13t/site-packages/gptqmodel/models/writer.py", line 231, in save_quantized
    model = convert_gptq_v2_to_v1_format(
        model, quantize_config=quantize_config, qlinear_kernel=self.qlinear_kernel
    )
  File "/home/ubuntu/venvs/gptqmodelt/lib/python3.13t/site-packages/gptqmodel/utils/model.py", line 669, in convert_gptq_v2_to_v1_format
    convert_gptq_v2_to_v1_format_module(module=submodule, quantize_config=quantize_config)
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ubuntu/venvs/gptqmodelt/lib/python3.13t/site-packages/gptqmodel/utils/model.py", line 635, in convert_gptq_v2_to_v1_format_module
    module.qzeros.data[:, range(0, module.qzeros.data.shape[1], 3)] -= (
    ~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: Inplace update to inference tensor outside InferenceMode is not allowed.You can make a clone to get a normal tensor before doing inplace update.See https://github.com/pytorch/rfcs/pull/17 for more details.

Ok. this look to be an easy fix. Looks unrelated to the PR but new code caused different timing so you are hitting more/different thread state bugs.