Tlntin/Qwen-TensorRT-LLM

Qwen-VL build.py: error: unrecognized arguments: --use_rmsnorm_plugin --use_lookup_plugin float16 --max_prompt_embedding_table_size 2048

Closed this issue · 1 comments

According to the readme, when run build.py in the docker as the following command:
python3 build.py --hf_model_dir=./Qwen-VL-Chat --dtype float16 --max_batch_size 4 --remove_input_padding --use_gpt_attention_plugin float16 --use_gemm_plugin float16 --enable_context_fmha --use_rmsnorm_plugin --log_level error --use_lookup_plugin float16 --max_prompt_embedding_table_size 2048 --output_dir=trt_engines/Qwen-VL-7B-fp16

build.py: error: unrecognized arguments: --use_rmsnorm_plugin --use_lookup_plugin float16 --max_prompt_embedding_table_size 2048

build.py has no above agrs params actually.

Do you use examples/qwen-vl/build.py?