[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
2025-01-28 02:19:24 +08:00
committed by GitHub
parent 3f1fc7425a
commit 2bc3fbba0c
10 changed files with 256 additions and 74 deletions

View File

@@ -61,9 +61,10 @@ def test_models(
if backend == "FLASHINFER" and current_platform.is_rocm():
pytest.skip("Flashinfer does not support ROCm/HIP.")
if backend == "XFORMERS" and model == "google/gemma-2-2b-it":
if backend in ("XFORMERS",
"FLASHINFER") and model == "google/gemma-2-2b-it":
pytest.skip(
"XFORMERS does not support gemma2 with full context length.")
f"{backend} does not support gemma2 with full context length.")
os.environ["VLLM_ATTENTION_BACKEND"] = backend