first add

This commit is contained in:
2024-08-28 15:18:15 +08:00
commit 873429d4e6
57 changed files with 4892 additions and 0 deletions

View File

@@ -0,0 +1,37 @@
version: '3.8'
services:
ubuntu-finetune:
image: hotwa/deepspeed:pt23
ports:
- 3228:22
environment:
- NVIDIA_VISIBLE_DEVICES=all
- NVIDIA_DRIVER_CAPABILITIES=compute,utility
- TMPDIR=/var/tmp
deploy:
replicas: 1
resources:
reservations:
generic_resources:
- discrete_resource_spec:
kind: "NVIDIA-GPU"
value: 1
placement:
constraints:
- node.labels.gpu == true
cap_add:
- IPC_LOCK
networks:
default:
driver: overlay
# 为节点添加标签:
# docker node ls
# docker node update --label-add gpu=true node1
# docker stack deploy -c docker-compose.yml rdma_stack