OpenCompass/opencompass/configs/models/qwen/lmdeploy_qwen2_7b_instruct.py

16 lines
486 B
Python

from opencompass.models import TurboMindModelwithChatTemplate
models = [
dict(
type=TurboMindModelwithChatTemplate,
abbr='qwen2-7b-instruct-turbomind',
path='Qwen/Qwen2-7B-Instruct',
engine_config=dict(session_len=16384, max_batch_size=16, tp=1),
gen_config=dict(top_k=1, temperature=1e-6, top_p=0.9, max_new_tokens=4096),
max_seq_len=16384,
max_out_len=4096,
batch_size=16,
run_cfg=dict(num_gpus=1),
)
]