feat(diffusers): add MPS version (#6121)

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

.github/workflows/backend_build_darwin.yml

@@ -12,6 +12,10 @@ on:
         description: 'Build type (e.g., mps)'
         default: ''
         type: string
+      use-pip:
+        description: 'Use pip to install dependencies'
+        default: false
+        type: boolean
       go-version:
         description: 'Go version to use'
         default: '1.24.x'
@@ -63,7 +67,7 @@ jobs:
       - name: Build ${{ inputs.backend }}-darwin
         run: |
           make protogen-go
-          BACKEND=${{ inputs.backend }} BUILD_TYPE=${{ inputs.build-type }} make build-darwin-python-backend
+          BACKEND=${{ inputs.backend }} BUILD_TYPE=${{ inputs.build-type }} USE_PIP=${{ inputs.use-pip }} make build-darwin-python-backend
           
       - name: Upload ${{ inputs.backend }}.tar
         uses: actions/upload-artifact@v4

Makefile

@@ -369,6 +369,10 @@ backends/mlx:
 	BACKEND=mlx BUILD_TYPE=mps $(MAKE) build-darwin-python-backend
 	./local-ai backends install "ocifile://$(abspath ./backend-images/mlx.tar)"
 
+backends/diffuser-darwin:
+	USE_PIP=true BACKEND=diffusers BUILD_TYPE=mps $(MAKE) build-darwin-python-backend
+	./local-ai backends install "ocifile://$(abspath ./backend-images/diffusers.tar)"
+
 backend-images:
 	mkdir -p backend-images
 

backend/index.yaml

@@ -184,6 +184,7 @@
     intel: "intel-diffusers"
     amd: "rocm-diffusers"
     nvidia-l4t: "nvidia-l4t-diffusers"
+    metal: "metal-diffusers"
 - &exllama2
   name: "exllama2"
   urls:
@@ -875,6 +876,7 @@
     intel: "intel-diffusers-development"
     amd: "rocm-diffusers-development"
     nvidia-l4t: "nvidia-l4t-diffusers-development"
+    metal: "metal-diffusers-development"
 - !!merge <<: *diffusers
   name: "nvidia-l4t-diffusers"
   uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-nvidia-l4t-diffusers"
@@ -925,6 +927,16 @@
   uri: "quay.io/go-skynet/local-ai-backends:master-gpu-intel-diffusers"
   mirrors:
     - localai/localai-backends:master-gpu-intel-diffusers
+- !!merge <<: *diffusers
+  name: "metal-diffusers"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-metal-darwin-arm64-diffusers"
+  mirrors:
+    - localai/localai-backends:latest-metal-darwin-arm64-diffusers
+- !!merge <<: *diffusers
+  name: "metal-diffusers-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-metal-darwin-arm64-diffusers"
+  mirrors:
+    - localai/localai-backends:master-metal-darwin-arm64-diffusers
   ## exllama2
 - !!merge <<: *exllama2
   name: "exllama2-development"

backend/python/diffusers/requirements-mps.txt

@@ -0,0 +1,10 @@
+torch==2.7.1
+torchvision==0.22.1
+git+https://github.com/huggingface/diffusers
+opencv-python
+transformers
+accelerate
+compel
+peft
+sentencepiece
+optimum-quanto

backend/python/diffusers/run.sh

@@ -12,4 +12,6 @@ if [ -d "/opt/intel" ]; then
     export XPU=1
 fi
 
+export PYTORCH_ENABLE_MPS_FALLBACK=1
+
 startBackend $@