- 模型切分
python tools/model_partition.py --config_file tasks/medusa_llama/config/vicuna_7b_config.json- 运行pipeline inference
python pipeline_inference.py --world 4 --rank xxx --config_file tasks/medusa_llama/config/vicuna_7b_config.json --load_in_8bit| Name | Name | Last commit date | ||
|---|---|---|---|---|
python tools/model_partition.py --config_file tasks/medusa_llama/config/vicuna_7b_config.jsonpython pipeline_inference.py --world 4 --rank xxx --config_file tasks/medusa_llama/config/vicuna_7b_config.json --load_in_8bit