A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 13, 14:30 UTC
Bump arctic-inference requirement (#28174)
Co-authored-by: Aurick Qiao <aurick.qiao@snowflake.com> Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
Aurick Qiao committed
Nov 7, 2025 at 18:31 UTC
781f5ebf524b36d64b732aec4e7d7df0c6be8c29
3 files changed
+3
-3
requirements/test.in
+1
-1
@@ -48,7 +48,7 @@ buildkite-test-collector==0.1.9
48
genai_perf==0.0.8
49
tritonclient==2.51.0
50
51
-arctic-inference == 0.1.0 # Required for suffix decoding test
51
+arctic-inference == 0.1.1 # Required for suffix decoding test
52
numba == 0.61.2 # Required for N-gram speculative decoding
53
numpy
54
runai-model-streamer[s3,gcs]==0.15.0
requirements/test.txt
+1
-1
@@ -40,7 +40,7 @@ anyio==4.6.2.post1
40
# via
41
# httpx
42
# starlette
43
-arctic-inference==0.1.0
43
+arctic-inference==0.1.1
44
# via -r requirements/test.in
45
argcomplete==3.5.1
46
# via datamodel-code-generator
vllm/config/speculative.py
+1
-1
@@ -469,7 +469,7 @@ class SpeculativeConfig:
469
if not has_arctic_inference():
470
raise ImportError(
471
"Arctic Inference is required for suffix decoding. "
472
- "Install via `pip install arctic-inference==0.1.0`."
472
+ "Install via `pip install arctic-inference==0.1.1`."
473
)
474
if self.num_speculative_tokens is None:
475
# Suffix decoding decides the actual number of speculative tokens