@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / e46fa5d52e
Fork 0

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 17, 04:15 UTC

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

Allen.Dou committed Feb 28, 2024 at 13:38 UTC e46fa5d52e02ee48d5fdd12b35e39993008b4bd6
3 files changed +3 -3
requirements-neuron.txt
+1 -1
@@ -6,4 +6,4 @@ neuronx-cc
6 fastapi
7 uvicorn[standard]
8 pydantic >= 2.0 # Required for OpenAI server.
9 -prometheus_client
9 +prometheus_client >= 0.18.0
requirements-rocm.txt
+1 -1
@@ -10,4 +10,4 @@ transformers >= 4.38.0 # Required for Gemma.
10 fastapi
11 uvicorn[standard]
12 pydantic >= 2.0 # Required for OpenAI server.
13 -prometheus_client
13 +prometheus_client >= 0.18.0
requirements.txt
+1 -1
@@ -9,7 +9,7 @@ xformers == 0.0.23.post1 # Required for CUDA 12.1.
9 fastapi
10 uvicorn[standard]
11 pydantic >= 2.0 # Required for OpenAI server.
12 -prometheus_client
12 +prometheus_client >= 0.18.0
13 pynvml == 11.5.0
14 triton >= 2.1.0
15 cupy-cuda12x == 12.1.0 # Required for CUDA graphs. CUDA 11.8 users should install cupy-cuda11x instead.