2023-05-08 07:30:43 +08:00
|
|
|
ninja # For faster builds.
|
|
|
|
|
psutil
|
2024-01-29 06:27:22 +08:00
|
|
|
ray >= 2.9
|
2023-05-08 07:30:43 +08:00
|
|
|
sentencepiece # Required for LLaMA tokenizer.
|
|
|
|
|
numpy
|
2023-12-17 17:46:54 +08:00
|
|
|
torch == 2.1.2
|
2024-02-22 01:38:03 +08:00
|
|
|
transformers >= 4.38.0 # Required for Gemma.
|
2023-12-17 18:28:02 +08:00
|
|
|
xformers == 0.0.23.post1 # Required for CUDA 12.1.
|
2023-05-08 07:30:43 +08:00
|
|
|
fastapi
|
2023-09-28 08:41:36 +08:00
|
|
|
uvicorn[standard]
|
2024-01-22 08:05:56 +08:00
|
|
|
pydantic >= 2.0 # Required for OpenAI server.
|
2024-02-28 13:38:26 +08:00
|
|
|
prometheus_client >= 0.18.0
|
2024-01-28 04:46:35 +08:00
|
|
|
pynvml == 11.5.0
|
2024-02-05 15:07:36 +08:00
|
|
|
triton >= 2.1.0
|
2024-02-15 02:17:57 +08:00
|
|
|
cupy-cuda12x == 12.1.0 # Required for CUDA graphs. CUDA 11.8 users should install cupy-cuda11x instead.
|
