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

[CORE] Quantized lm-head Framework (#4442)

Browse Source 
Co-authored-by: Robert Shaw <rshaw@neuralmagic.com>
Co-authored-by: ZX <zx@lbx.dev>
This commit is contained in:
Qubitium-ModelCloud
2024-07-03 06:25:17 +08:00
committed by GitHub
parent 7c008c51a9
commit ee93f4f92a
48 changed files with 268 additions and 121 deletions
Download Patch File Download Diff File Expand all files Collapse all files

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

@@ -34,7 +34,7 @@ SPEC_MODEL = "ibm-granite/granite-3b-code-instruct-accelerator"
MAX_SPEC_TOKENS = 5
# precision
PRECISION = "float16"
PRECISION = "float32"
@pytest.mark.parametrize(