leaderboard / deployment /docker-compose-0.yaml
Jae-Won Chung
Switch to MPT
9d117bb
raw
history blame
2.27 kB
services:
MPT-7B:
container_name: worker0
image: mlenergy/tgi:latest
command: ["--model-id", "mosaicml/mpt-7b-chat", "--num-shard", "1", "--otlp-endpoint", "http://jaeger:4317"]
shm_size: 1g
networks:
- leaderboard
volumes:
- /data/leaderboard/tgi-data:/data
deploy:
restart_policy:
condition: any
resources:
reservations:
devices:
- driver: nvidia
device_ids: ["0"]
capabilities: [gpu]
Llama2-7B:
container_name: worker1
image: mlenergy/tgi:latest
command: ["--model-id", "/weights/metaai/Llama-2-7b-chat-hf", "--num-shard", "1", "--otlp-endpoint", "http://jaeger:4317"]
shm_size: 1g
networks:
- leaderboard
volumes:
- /data/leaderboard/tgi-data:/data
- /data/leaderboard/weights:/weights
deploy:
restart_policy:
condition: any
resources:
reservations:
devices:
- driver: nvidia
device_ids: ["1"]
capabilities: [gpu]
FastChat-T5-3B:
container_name: worker2
image: mlenergy/tgi:latest
command: ["--model-id", "lmsys/fastchat-t5-3b-v1.0", "--num-shard", "1", "--otlp-endpoint", "http://jaeger:4317"]
environment:
PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION: python
shm_size: 1g
networks:
- leaderboard
volumes:
- /data/leaderboard/tgi-data:/data
deploy:
restart_policy:
condition: any
resources:
reservations:
devices:
- driver: nvidia
device_ids: ["2"]
capabilities: [gpu]
Llama2-13B:
container_name: worker3
image: mlenergy/tgi:latest
command: ["--model-id", "/weights/metaai/Llama-2-13b-chat-hf", "--num-shard", "1", "--otlp-endpoint", "http://jaeger:4317"]
shm_size: 1g
networks:
- leaderboard
volumes:
- /data/leaderboard/tgi-data:/data
- /data/leaderboard/weights:/weights
deploy:
restart_policy:
condition: any
resources:
reservations:
devices:
- driver: nvidia
device_ids: ["3"]
capabilities: [gpu]
networks:
leaderboard:
name: leaderboard
external: true