[Bugfix] Treat generation_config max_tokens as default not ceiling (#34063)

Signed-off-by: almogtavor <almogtavor@gmail.com>
This commit is contained in:
Almog Tavor
2026-02-16 07:58:24 -08:00
committed by GitHub
parent a3205beffb
commit 72d5951d02
7 changed files with 157 additions and 17 deletions
+12 -6
View File
@@ -177,17 +177,23 @@ def get_max_tokens(
max_tokens: int | None,
input_length: int,
default_sampling_params: dict,
override_max_tokens: int | None = None,
) -> int:
default_max_tokens = max_model_len - input_length
max_output_tokens = current_platform.get_max_output_tokens(input_length)
model_max_tokens = max_model_len - input_length
platform_max_tokens = current_platform.get_max_output_tokens(input_length)
fallback_max_tokens = (
max_tokens
if max_tokens is not None
else default_sampling_params.get("max_tokens")
)
return min(
val
for val in (
default_max_tokens,
max_tokens,
max_output_tokens,
default_sampling_params.get("max_tokens"),
model_max_tokens,
fallback_max_tokens,
override_max_tokens,
platform_max_tokens,
)
if val is not None
)