Bug when quantize GLM-MoE model

[mxjmtxrm]

Describe the bug

When I tried to quantize GLM-MoE model GLM-4.5 using example/llm_ptq/hf_ptq.py, I met the following error:

Traceback (most recent call last):
  File "/TensorRT-Model-Optimizer/examples/llm_ptq/hf_ptq.py", line 772, in <module>
    main(args)
  File "/TensorRT-Model-Optimizer/examples/llm_ptq/hf_ptq.py", line 625, in main
    export_hf_checkpoint(
  File "/TensorRT-Model-Optimizer/modelopt/torch/export/unified_export_hf.py", line 545, in export_hf_checkpoint
    raise e
  File "/TensorRT-Model-Optimizer/modelopt/torch/export/unified_export_hf.py", line 511, in export_hf_checkpoint
    post_state_dict, hf_quant_config = _export_hf_checkpoint(model, dtype)
                                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/TensorRT-Model-Optimizer/modelopt/torch/export/unified_export_hf.py", line 461, in _export_hf_checkpoint
    _export_quantized_weight(sub_module, dtype)
  File "/TensorRT-Model-Optimizer/modelopt/torch/export/unified_export_hf.py", line 266, in _export_quantized_weight
    quantizer_attrs.weight_scale, get_weight_scaling_factor(sub_module, weight_name)
                                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/TensorRT-Model-Optimizer/modelopt/torch/export/quant_utils.py", line 275, in get_weight_scaling_factor
    NVFP4QTensor.get_weights_scaling_factor_2_from_quantizer(weight_quantizer).to(
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/TensorRT-Model-Optimizer/modelopt/torch/quantization/qtensor/nvfp4_tensor.py", line 59, in get_weights_scaling_factor_2_from_quantizer
    assert hasattr(weight_quantizer, "_amax"), "Weight quantizer does not have attribute amax"
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: Weight quantizer does not have attribute amax

Steps/Code to reproduce bug

MODEL_PATH=xxx
 SAVE_PATH=xxx
 QFORMAT=nvfp4
 CALIB_SIZE=64
 CALIB_BATCH_SIZE=1
 EXPORT_FORMAT=hf
 python hf_ptq.py \
    --pyt_ckpt_path=$MODEL_PATH \
    --export_path=$SAVE_PATH \
    --qformat=$QFORMAT \
    --calib_size=$CALIB_SIZE \
    --batch_size=$CALIB_BATCH_SIZE \
    --export_fmt=$EXPORT_FORMAT \
    --kv_cache_qformat=none \
    --trust_remote_code

System information

• ModelOpt 0.37.0.dev30