Skip to content

Commit 1466879

Browse files
committed
feat: add omni app
1 parent 648d5a0 commit 1466879

45 files changed

Lines changed: 2389 additions & 44 deletions

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

.github/workflows/build-push-images.yml

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -19,6 +19,7 @@ jobs:
1919
- chat
2020
- image-analysis
2121
- flux-image-gen
22+
- omni
2223
permissions:
2324
contents: read
2425
id-token: write # needed for signing the images with GitHub OIDC Token
Lines changed: 33 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,33 @@
1+
# Patterns to ignore when building packages.
2+
# This supports shell glob matching, relative path matching, and
3+
# negation (prefixed with !). Only one pattern per line.
4+
.DS_Store
5+
# Common VCS dirs
6+
.git/
7+
.gitignore
8+
.bzr/
9+
.bzrignore
10+
.hg/
11+
.hgignore
12+
.svn/
13+
# Common backup files
14+
*.swp
15+
*.bak
16+
*.tmp
17+
*.orig
18+
*~
19+
# Various IDEs
20+
.project
21+
.idea/
22+
*.tmproj
23+
.vscode/
24+
25+
# Others
26+
README.md
27+
Dockerfile
28+
*kubeconfig.y[a]ml
29+
venv/
30+
__pycache__/
31+
images/
32+
*.secret
33+
ci/
Lines changed: 16 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,16 @@
1+
apiVersion: v2
2+
name: azimuth-llm-omni-backend
3+
description: |
4+
In-cluster vLLM backends for the Omni multimodal interface
5+
(text-to-text / chat, text-to-speech, text-to-image).
6+
maintainers:
7+
- name: "Victor HANG"
8+
email: victor@stackhpc.com
9+
10+
type: application
11+
12+
# The version and appVersion are updated by the chart build script
13+
version: 0.1.0
14+
appVersion: local
15+
16+
icon: https://huggingface.co/datasets/huggingface/brand-assets/resolve/main/hf-logo.svg
Lines changed: 13 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,13 @@
1+
# CI: only exercise the TTT backend with the smallest possible model.
2+
ttt:
3+
enabled: true
4+
huggingface:
5+
model: HuggingFaceTB/SmolLM2-135M-Instruct
6+
api:
7+
monitoring:
8+
enabled: false
9+
gpus: 0
10+
tts:
11+
enabled: false
12+
tti:
13+
enabled: false
Lines changed: 17 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,17 @@
1+
Azimuth Omni backends provide one or more in-cluster vLLM model deployments
2+
(text-to-text / chat, text-to-speech, text-to-image) for the Omni interface.
3+
4+
Enabled backends in this release:
5+
{{- range $alias := list "ttt" "tts" "tti" }}
6+
{{- $sub := index $.Values $alias }}
7+
{{- if and $sub $sub.enabled }}
8+
- {{ $alias }}: {{ $sub.huggingface.model }} (in-cluster at http://{{ $.Release.Name }}-{{ $alias }}.{{ $.Release.Namespace }}.svc)
9+
{{- else }}
10+
- {{ $alias }}: disabled (toggle with `{{ $alias }}.enabled=true`)
11+
{{- end }}
12+
{{- end }}
13+
14+
Each enabled backend downloads its model weights from HuggingFace on first
15+
start, which can take a while.
16+
17+
Release notes: https://github.com/stackhpc/azimuth-llm/releases

0 commit comments

Comments
 (0)