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

[Bugfix][SpecDecode] apply sampling parameters to target probabilities for consistency in rejection sampling. (#10198)

Browse Source 
Signed-off-by: jeongin601 <0200angela@gmail.com>
Signed-off-by: jeong_in.bae <jeong_in.bae@navercorp.com>
This commit is contained in:
jeongin601
2024-11-27 14:07:30 +09:00
committed by GitHub
parent 0a4d968500
commit 1bf905ddaa
3 changed files with 10 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
tests/spec_decode/e2e/test_mlp_correctness.py
View File

@@ -203,7 +203,7 @@ def test_mlp_e2e_acceptance_rate(vllm_runner, common_llm_kwargs,
@pytest.mark.parametrize("test_llm_kwargs", [{"seed": 5}])
@pytest.mark.parametrize("output_len", [64])
@pytest.mark.parametrize("batch_size", [1, 32])
@pytest.mark.parametrize("temperature", [0.1, 1.0])
@pytest.mark.parametrize("temperature", [1.0])
@pytest.mark.parametrize("seed", [1])
def test_mlp_e2e_seeded_correctness(vllm_runner, common_llm_kwargs,
per_test_common_llm_kwargs,