from opencompass.models import HuggingFaceCausalLM | |
models = [ | |
# LLaMA 30B | |
dict( | |
type=HuggingFaceCausalLM, | |
abbr='llama-30b-hf', | |
path="huggyllama/llama-30b", | |
tokenizer_path='huggyllama/llama-30b', | |
tokenizer_kwargs=dict(padding_side='left', | |
truncation_side='left', | |
use_fast=False, | |
), | |
max_out_len=100, | |
max_seq_len=2048, | |
batch_size=8, | |
model_kwargs=dict(device_map='auto'), | |
batch_padding=False, # if false, inference with for-loop without batch padding | |
run_cfg=dict(num_gpus=4, num_procs=1), | |
) | |
] | |