- support deepseek quant - add w8a8_dynamic quant see #391 Signed-off-by: MengqingCao <cmq0113@163.com> Co-authored-by: zzzzwwjj <1183291235@qq.com>
18 lines
564 B
Python
18 lines
564 B
Python
from vllm import ModelRegistry
|
|
|
|
|
|
def register_model():
|
|
from .qwen2_vl import CustomQwen2VLForConditionalGeneration # noqa: F401
|
|
|
|
ModelRegistry.register_model(
|
|
"Qwen2VLForConditionalGeneration",
|
|
"vllm_ascend.models.qwen2_vl:CustomQwen2VLForConditionalGeneration")
|
|
|
|
ModelRegistry.register_model(
|
|
"DeepseekV2ForCausalLM",
|
|
"vllm_ascend.models.deepseek_v2:CustomDeepseekV2ForCausalLM")
|
|
|
|
ModelRegistry.register_model(
|
|
"DeepseekV3ForCausalLM",
|
|
"vllm_ascend.models.deepseek_v2:CustomDeepseekV3ForCausalLM")
|