20231088/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Restrict prometheus_client >= 0.18.0 to prevent errors when importing pkgs (#3070)

Browse Source
This commit is contained in:
Allen.Dou 2024-02-28 13:38:26 +08:00 committed by GitHub
parent a8683102cc
commit e46fa5d52e
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
requirements-neuron.txt
View File

@ -6,4 +6,4 @@ neuronx-cc
fastapi
uvicorn[standard]
pydantic >= 2.0 # Required for OpenAI server.
prometheus_client
prometheus_client >= 0.18.0

2
requirements-rocm.txt
View File

@ -10,4 +10,4 @@ transformers >= 4.38.0 # Required for Gemma.
fastapi
uvicorn[standard]
pydantic >= 2.0 # Required for OpenAI server.
prometheus_client
prometheus_client >= 0.18.0

2
requirements.txt
View File

@ -9,7 +9,7 @@ xformers == 0.0.23.post1 # Required for CUDA 12.1.
fastapi
uvicorn[standard]
pydantic >= 2.0 # Required for OpenAI server.
prometheus_client
prometheus_client >= 0.18.0
pynvml == 11.5.0
triton >= 2.1.0
cupy-cuda12x == 12.1.0 # Required for CUDA graphs. CUDA 11.8 users should install cupy-cuda11x instead.