usage: ct2-transformers-converter [-h] --model MODEL [--activation_scales ACTIVATION_SCALES] [--copy_files COPY_FILES [COPY_FILES ...]] [--revision REVISION] [--low_cpu_mem_usage] [--trust_remote_code] --output_dir OUTPUT_DIR [--vocab_mapping VOCAB_MAPPING] [--quantization {int8,int8_float32,int8_float16,int8_bfloat16,int16,float16,bfloat16,float32}] [--force] options: -h, --help show this help message and exit --model MODEL Name of the pretrained model to download, or path to a directory containing the pretrained model. (default: None) --activation_scales ACTIVATION_SCALES Path to the pre-computed activation scales. Models may use them to rescale some weights to smooth the intermediate activations and improve the quantization accuracy. See https://github.com/mit-han-lab/smoothquant. (default: None) --copy_files COPY_FILES [COPY_FILES ...] List of filenames to copy from the Hugging Face model to the converted model directory. (default: None) --revision REVISION Revision of the model to download from the Hugging Face Hub. (default: None) --low_cpu_mem_usage Enable the flag low_cpu_mem_usage when loading the model with from_pretrained. (default: False) --trust_remote_code Allow converting models using custom code. (default: False) --output_dir OUTPUT_DIR Output model directory. (default: None) --vocab_mapping VOCAB_MAPPING Vocabulary mapping file (optional). (default: None) --quantization {int8,int8_float32,int8_float16,int8_bfloat16,int16,float16,bfloat16,float32} Weight quantization type. (default: None) --force Force conversion even if the output directory already exists. (default: False)