Llama-3.2 SmoothQuant convert checkpoint error

[lyffly]

System Info

x86_64, ubuntu, A100

Who can help?

No response

Information

• The official example scripts
• My own modified scripts

Tasks

• An officially supported task in the examples folder (such as GLUE/SQuAD, ...)
• My own task or dataset (give details below)

Reproduction

python3 convert_checkpoint.py --model_dir ./Llama-3.2-3B-Instruct/  \
                    --output_dir tllm_checkpoint_1gpu_sq \
                    --dtype float16 \
                    --smoothquant 0.5

Expected behavior

no error

actual behavior

  File "/usr/local/lib/python3.10/dist-packages/tensorrt_llm/models/llama/model.py", line 529, in quantize
    convert.quantize(hf_model_dir,
  File "/usr/local/lib/python3.10/dist-packages/tensorrt_llm/models/llama/convert.py", line 1146, in quantize
    weights = load_weights_from_hf_model(
  File "/usr/local/lib/python3.10/dist-packages/tensorrt_llm/models/llama/convert.py", line 1059, in load_weights_from_hf_model
    lm_head_weights = get_weight(model_params, param_name_map["lm_head"], dtype)
  File "/usr/local/lib/python3.10/dist-packages/tensorrt_llm/models/llama/convert.py", line 211, in get_weight
    if named_params[prefix + '.weight'].dtype != dtype:
KeyError: 'lm_head.weight'

additional notes

using pdb , shows model_params ,many value is none, for example

'model.layers.0.mlp.gate_proj.weight': None, 'model.layers.0.mlp.up_proj.weight': None, 'model.layers.0.mlp.down_proj.weight': None, 'model.layers.0.input_layernorm.weight': None, 'model.layers.0.post_attention_layernorm.weight': None, 'model.layers.1.self_attn.q_proj.weight': None, 'model.layers.1.self_attn.k_proj.weight': None, 'model.layers.1.self_attn.v_proj.weight': None, 'model.layers.1.self_attn.o_proj.weight': None, 'model.layers.1.mlp.gate_proj.weight': None, 'model.layers.1.mlp.up_proj.weight': None, 'model.layers.1.mlp.down_proj.weight': None, 'model.layers.1.input_layernorm.weight': None, 'model.layers.1.post_attention_layernorm.weight': None, 'model.layers.2.self_attn.q_proj.weight': None, 'model.layers.2.self_attn.k_proj.weight': None, 'model.layers.2.self_attn.v_proj.weight': None, 'model.layers.2.self_attn.o_proj.weight': None, 'model.layers.2.mlp.gate_proj.weight': None,