biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[BugFix][Spec Decode] Fix out-of-range index triggered by eagle3; re-enable test for LlamaForCausalLMEagle3 (#24392)

Browse Source 
Signed-off-by: wwl2755 <wangwenlong2755@gmail.com>
This commit is contained in:
Wenlong Wang
2025-09-09 21:24:23 -07:00
committed by GitHub
parent 309d7aa401
commit 53b42f4102
7 changed files with 58 additions and 41 deletions
Download Patch File Download Diff File Expand all files Collapse all files

55
tests/v1/e2e/test_spec_decode.py
View File

@@ -125,37 +125,30 @@ def test_ngram_correctness(
cleanup_dist_env_and_memory()
@pytest.mark.parametrize(
["model_setup", "mm_enabled"],
[
# TODO: Re-enable this once tests/models/test_initialization.py is fixed, see PR #22333 #22611 # noqa: E501
# (("eagle3", "Qwen/Qwen3-8B", "AngelSlim/Qwen3-8B_eagle3", 1), False),
(("eagle", "meta-llama/Llama-3.1-8B-Instruct",
"yuhuili/EAGLE-LLaMA3.1-Instruct-8B", 1), False),
(("eagle3", "meta-llama/Llama-3.1-8B-Instruct",
"yuhuili/EAGLE3-LLaMA3.1-Instruct-8B", 1), False),
pytest.param(
("eagle", "meta-llama/Llama-4-Scout-17B-16E-Instruct",
"morgendave/EAGLE-Llama-4-Scout-17B-16E-Instruct", 4),
False,
marks=pytest.mark.skip(reason="Skipping due to CI OOM issues")),
pytest.param(
("eagle", "meta-llama/Llama-4-Scout-17B-16E-Instruct",
"morgendave/EAGLE-Llama-4-Scout-17B-16E-Instruct", 4),
True,
marks=pytest.mark.skip(reason="Skipping due to CI OOM issues")),
(("eagle", "eagle618/deepseek-v3-random",
"eagle618/eagle-deepseek-v3-random", 1), False),
],
ids=[
# TODO: Re-enable this once tests/models/test_initialization.py is fixed, see PR #22333 #22611 # noqa: E501
# "qwen3_eagle3",
"llama3_eagle",
"llama3_eagle3",
"llama4_eagle",
"llama4_eagle_mm",
"deepseek_eagle"
])
@pytest.mark.parametrize(["model_setup", "mm_enabled"], [
(("eagle3", "Qwen/Qwen3-8B", "AngelSlim/Qwen3-8B_eagle3", 1), False),
(("eagle", "meta-llama/Llama-3.1-8B-Instruct",
"yuhuili/EAGLE-LLaMA3.1-Instruct-8B", 1), False),
(("eagle3", "meta-llama/Llama-3.1-8B-Instruct",
"yuhuili/EAGLE3-LLaMA3.1-Instruct-8B", 1), False),
pytest.param(
("eagle", "meta-llama/Llama-4-Scout-17B-16E-Instruct",
"morgendave/EAGLE-Llama-4-Scout-17B-16E-Instruct", 4),
False,
marks=pytest.mark.skip(reason="Skipping due to CI OOM issues")),
pytest.param(
("eagle", "meta-llama/Llama-4-Scout-17B-16E-Instruct",
"morgendave/EAGLE-Llama-4-Scout-17B-16E-Instruct", 4),
True,
marks=pytest.mark.skip(reason="Skipping due to CI OOM issues")),
(("eagle", "eagle618/deepseek-v3-random",
"eagle618/eagle-deepseek-v3-random", 1), False),
],
ids=[
"qwen3_eagle3", "llama3_eagle", "llama3_eagle3",
"llama4_eagle", "llama4_eagle_mm",
"deepseek_eagle"
])
@pytest.mark.parametrize("attn_backend",
get_attn_backend_list_based_on_platform())
def test_eagle_correctness(