Skip to content

GitLab

Unverified Commit a32c8669 authored by Nick Hill's avatar Nick Hill Committed by GitHub
Browse files

[V1][Minor] Remove obsolete FIXME comment (#14304) 


Signed-off-by: default avatarNick Hill <nhill@redhat.com>
parent ca2ca8de
Hide whitespace changes
Inline Side-by-side
Showing with 0 additions and 5 deletions
vllm/v1/worker/gpu_input_batch.py
View file @ a32c8669
...@@ -298,11 +298,6 @@ class InputBatch: ...@@ -298,11 +298,6 @@ class InputBatch:
if sampling_params.logit_bias is not None: if sampling_params.logit_bias is not None:
self.logit_bias[req_index] = sampling_params.logit_bias self.logit_bias[req_index] = sampling_params.logit_bias
# FIXME: this implementation is incorrect. We create this mask
# then apply -inf to these specific tokens, which means we never
# select the allowed tokens! We cannot do the reverse, since
# this will impact the requests that do not have allowed_token_ids.
# This feature is currently disabled on V1 (we reject in Processor).
if sampling_params.allowed_token_ids: if sampling_params.allowed_token_ids:
self.has_allowed_token_ids.add(req_id) self.has_allowed_token_ids.add(req_id)
if self.allowed_token_ids_mask_cpu_tensor is None: if self.allowed_token_ids_mask_cpu_tensor is None:
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment