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 8, 22:45 UTC
Apply #33621 to main (#33758)
Signed-off-by: Zachary Aristei <zaristei@nvidia.com> Co-authored-by: zaristei2 <zaristei2@gmail.com> Co-authored-by: Zachary Aristei <zaristei@nvidia.com>
Cyrus Leung committed
Feb 4, 2026 at 21:35 UTC
32a02c7ca29180f70c1c8c73d0f57445231b17b5
3 files changed
+3
-3
requirements/common.txt
+1
-1
@@ -11,7 +11,7 @@ transformers >= 4.56.0, < 5
11
tokenizers >= 0.21.1 # Required for fast incremental detokenization.
12
protobuf # Required by LlamaTokenizer, gRPC.
13
fastapi[standard] >= 0.115.0 # Required by FastAPI's form models in the OpenAI API server's audio transcriptions endpoint.
14
-aiohttp
14
+aiohttp >= 3.13.3
15
openai >= 1.99.1 # For Responses API with reasoning content
16
pydantic >= 2.12.0
17
prometheus_client >= 0.18.0
requirements/rocm-test.txt
+1
-1
@@ -14,7 +14,7 @@ pytest-shard==0.1.2
14
# Async/HTTP dependencies
15
anyio==4.6.2.post1
16
# via httpx, starlette
17
-aiohttp==3.13.0
17
+aiohttp==3.13.3
18
# via gpt-oss
19
httpx==0.27.2
20
# HTTP testing
requirements/test.txt
+1
-1
@@ -12,7 +12,7 @@ affine==2.4.0
12
# via rasterio
13
aiohappyeyeballs==2.6.1
14
# via aiohttp
15
-aiohttp==3.13.0
15
+aiohttp==3.13.3
16
# via
17
# aiohttp-cors
18
# datasets