Add multimodal rag recipe

multimodal-rag-with-colpali-llamavision-reranker/.gitattributes

@@ -0,0 +1,2 @@
+# SCM syntax highlighting & preventing 3-way merges
+pixi.lock merge=binary linguist-language=YAML linguist-generated=true

multimodal-rag-with-colpali-llamavision-reranker/.gitignore

@@ -0,0 +1,11 @@
+
+# pixi environments
+.pixi
+*.egg-info
+# magic environments
+.magic
+
+qdrant_storage
+.gradio
+temp_*
+.cache

multimodal-rag-with-colpali-llamavision-reranker/Procfile

@@ -0,0 +1,3 @@
+llm-server: (magic global install max-pipelines && magic global update max-pipelines) || true; MAX_SERVE_PORT=8010 MAX_SERVE_HOST=0.0.0.0 max-pipelines serve --huggingface-repo-id=meta-llama/Llama-3.2-11B-Vision-Instruct --max-length 2048 --max-batch-size 1
+qdrant-server: docker run -p 6333:6333 -p 6334:6334 -v "$(pwd)/qdrant_storage:/qdrant/storage:z" qdrant/qdrant
+app: magic run python app.py

multimodal-rag-with-colpali-llamavision-reranker/Procfile.clean

@@ -0,0 +1,6 @@
+cleanup: pkill -f "max-pipelines serve" || true && pkill -f "docker run -p 6333:6333 -p 6334:6334 -v $(pwd)/qdrant_storage:/qdrant/storage:z qdrant/qdrant" || true
+gpu-cleanup: command -v nvidia-smi >/dev/null && nvidia-smi pmon -c 1 | grep python | awk '{print $2}' | xargs -r kill -9 2>/dev/null || true
+port-cleanup: lsof -ti:8010,6333,6334,7860 | xargs -r kill -9 2>/dev/null || true
+qdrat-cleanup: rm -r qdrant_storage 2>/dev/null || true
+gradio-cleanup: rm -r .gradio 2>/dev/null || true
+temp-cleanup: rm -r temp_images 2>/dev/null || true