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.