[Bugfix] Align vLLM YaRN with Transformers and stop re-scaling max_model_len (#56446)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Harry Mellor
2026-09-11 13:01:38 +00:00
committed by GitHub
co-authored by Claude Opus 5
parent a5b714eee4
commit c191787a68
11 changed files with 133 additions and 134 deletions
@@ -70,10 +70,10 @@ def yarn_linear_ramp_mask(
return ramp_func
def yarn_get_mscale(scale: float = 1) -> float:
def yarn_get_mscale(scale: float = 1, mscale: float = 1) -> float:
if scale <= 1:
return 1.0
return 0.1 * math.log(scale) + 1.0
return 0.1 * mscale * math.log(scale) + 1.0
def _flashinfer_rotary_embedding(