[FlashInfer] Upgrade to 0.2.0 (#11194)

Signed-off-by: Bowen Wang <abmfy@icloud.com>
Signed-off-by: youkaichao <youkaichao@gmail.com>
Co-authored-by: youkaichao <youkaichao@gmail.com>
This commit is contained in:
Bowen Wang
2025-01-27 18:19:24 +00:00
committed by GitHub
co-authored by youkaichao
parent 3f1fc7425a
commit 2bc3fbba0c
10 changed files with 256 additions and 74 deletions
@@ -288,7 +288,8 @@ class TensorizerAgent:
model_args.torch_dtype = self.tensorizer_config.dtype
assert self.tensorizer_config.model_class is not None
# TODO: Do we need to consider old-style model class?
with no_init_or_tensor(), set_current_vllm_config(self.vllm_config):
with no_init_or_tensor(), set_current_vllm_config(self.vllm_config,
check_compile=True):
return self.tensorizer_config.model_class(
vllm_config=self.vllm_config, )