Yanbo Li
Yanbo Li
``` import copy from transformers import AutoModelForCausalLM, AutoTokenizer,AutoConfig from modify_llama import convert_kvcache_llama_heavy_recent import torch import time model_path = "/dataset/crosspipe/OriginModel/Llama-2-7b-chat-hf/" prompt_text = 'In a small, bustling cafe nestled in the heart...
thanks for your excellent job! So my question is how to set the appropriate coefficients according to different models? the coefficients in FLUX is [here](https://github.com/ali-vilab/TeaCache/blob/7c10efc4702c6b619f47805f7abe4a7a08085aa0/TeaCache4FLUX/teacache_flux.py#L113C17-L113C115), while the [coefficients](https://github.com/ali-vilab/TeaCache/blob/7c10efc4702c6b619f47805f7abe4a7a08085aa0/TeaCache4HunyuanVideo/teacache_sample_video.py#L102) in HunyuanVideo...