feat: Adds Qdrant vector search

AGENTS.md

@@ -334,6 +334,7 @@ if __name__ == "__main__":
 - Index backends are pluggable via factory:
   - `faiss`
   - `milvus`
+  - `qdrant`
 - Web search backends are pluggable via factory:
   - `exa`
   - `tavily`

docs/llms.txt

@@ -1596,7 +1596,7 @@ async def retriever_init(
 **Function**
 - Initializes retrieval service.
 - Embedding Backend (backend): Responsible for converting text/images into vectors (Infinity, SentenceTransformers, OpenAI, BM25).
-- Index Backend (index_backend): Responsible for vector storage and retrieval (FAISS, Milvus).
+- Index Backend (index_backend): Responsible for vector storage and retrieval (FAISS, Milvus, Qdrant).
 - Demo Mode: If is_demo=True, forces OpenAI + Milvus configuration, ignoring some parameters.
 
 ---
@@ -1634,7 +1634,7 @@ async def retriever_index(
 **Function**
 - Builds retrieval index.
 - FAISS: Reads embedding_path (.npy) to build local index file.
-- Milvus / Demo: Reads corpus_path (.jsonl), generates vectors and inserts into specified collection_name.
+- Milvus / Qdrant / Demo: Reads corpus_path (.jsonl), generates vectors and inserts into specified collection_name.
 
 ---
 
@@ -1652,7 +1652,7 @@ async def retriever_search(
 
 **Function**
 - Retrieves single or multiple queries.
-- Automatically handles query vectorization (adds query_instruction) and finds Top-K in specified collection_name (for Milvus) or default index.
+- Automatically handles query vectorization (adds query_instruction) and finds Top-K in specified collection_name (for Milvus or Qdrant) or default index.
 
 **Output Format (JSON)**
 ```json
@@ -1820,7 +1820,7 @@ backend_configs:
     save_path: index/bm25
 
 # Index Backend Configuration
-index_backend: faiss # options: faiss, milvus
+index_backend: faiss # options: faiss, milvus, qdrant
 index_backend_configs:
   faiss:
     index_use_gpu: True
@@ -1878,9 +1878,9 @@ Parameter Description:
 | `model_name_or_path` | str | Retrieval model path or name (e.g., HuggingFace model ID) |
 | `corpus_path` | str | Input corpus JSONL file path |
 | `embedding_path` | str | Vector file save path (`.npy`) |
-| `collection_name` | str | Milvus collection name |
+| `collection_name` | str | Milvus or Qdrant collection name |
 | `backend` | str | Select retrieval backend: `infinity`, `sentence_transformers`, `openai`, `bm25` |
-| `index_backend` | str | Index backend: `faiss`, `milvus` |
+| `index_backend` | str | Index backend: `faiss`, `milvus`, `qdrant` |
 | `backend_configs` | dict | Parameter configuration for each backend (see table below) |
 | `index_backend_configs` | dict | Parameter configuration for each index backend (see table below) |
 | `websearch_backend` | str | Web search backend: `tavily`, `exa`, `zhipuai` |
@@ -1941,6 +1941,12 @@ Parameter Description:
 ||index_params|	Dict| Index construction parameters (e.g., index_type: AUTOINDEX)|
 ||search_params	|Dict |Retrieval parameters (e.g., nprobe etc.)|
 ||index_chunk_size|	int| Batch size when inserting data|
+|qdrant|url|	str| Qdrant server URL (e.g., http://localhost:6333); leave null to use local path|
+||path|	str |Local persistent storage path; used when url is null|
+||api_key|	str |API key for authentication (if needed)|
+||text_field_name|	str |Payload field name for document text (default contents)|
+||distance|	str |Distance metric: Cosine, Dot, Euclid, Manhattan (default Cosine)|
+||index_chunk_size|	int |Batch size when upserting vectors|
 
 ================
 File: pages/en/api/router.mdx

examples/demos/parameter/AgentCPM-Report_parameter.yaml

@@ -60,6 +60,13 @@ retriever:
         metric_type: IP
         params: {}
       index_chunk_size: 1000
+    qdrant:
+      url: null
+      path: index/qdrant
+      api_key: null
+      text_field_name: contents
+      distance: Cosine
+      index_chunk_size: 50000
   is_demo: false
   collection_name: wiki
   top_k: 5

examples/demos/parameter/AgentCPM-Report_web_parameter.yaml

@@ -60,6 +60,13 @@ retriever:
         metric_type: IP
         params: {}
       index_chunk_size: 1000
+    qdrant:
+      url: null
+      path: index/qdrant
+      api_key: null
+      text_field_name: contents
+      distance: Cosine
+      index_chunk_size: 50000
   is_demo: false
   collection_name: wiki
   top_k: 5

examples/demos/parameter/LLM_memory_parameter.yaml

@@ -62,6 +62,13 @@ retriever:
         metric_type: IP
         params: {}
       index_chunk_size: 1000
+    qdrant:
+      url: null
+      path: index/qdrant
+      api_key: null
+      text_field_name: contents
+      distance: Cosine
+      index_chunk_size: 50000
   is_demo: false
   collection_name: wiki
   top_k: 5

examples/demos/parameter/LightResearch_parameter.yaml

@@ -94,6 +94,13 @@ retriever:
       token: null
       uri: index/milvus_demo.db
       vector_field_name: vector
+    qdrant:
+      url: null
+      path: index/qdrant
+      api_key: null
+      text_field_name: contents
+      distance: Cosine
+      index_chunk_size: 50000
   is_demo: false
   is_multimodal: false
   model_name_or_path: openbmb/MiniCPM-Embedding-Light

examples/demos/parameter/RAG_memory_parameter.yaml

@@ -62,6 +62,13 @@ retriever:
         metric_type: IP
         params: {}
       index_chunk_size: 1000
+    qdrant:
+      url: null
+      path: index/qdrant
+      api_key: null
+      text_field_name: contents
+      distance: Cosine
+      index_chunk_size: 50000
   is_demo: false
   collection_name: wiki
   top_k: 5

examples/demos/parameter/RAG_parameter.yaml

@@ -90,6 +90,13 @@ retriever:
       token: null
       uri: index/milvus_demo.db
       vector_field_name: vector
+    qdrant:
+      url: null
+      path: index/qdrant
+      api_key: null
+      text_field_name: contents
+      distance: Cosine
+      index_chunk_size: 50000
   is_demo: false
   is_multimodal: false
   model_name_or_path: openbmb/MiniCPM-Embedding-Light

examples/demos/parameter/RAG_web_parameter.yaml

@@ -60,6 +60,13 @@ retriever:
         metric_type: IP
         params: {}
       index_chunk_size: 1000
+    qdrant:
+      url: null
+      path: index/qdrant
+      api_key: null
+      text_field_name: contents
+      distance: Cosine
+      index_chunk_size: 50000
   is_demo: false
   collection_name: wiki
   top_k: 5

examples/demos/parameter/milvus_index_parameter.yaml

@@ -50,6 +50,13 @@ retriever:
         metric_type: IP
         params: {}
       index_chunk_size: 1000
+    qdrant:
+      url: null
+      path: index/qdrant
+      api_key: null
+      text_field_name: contents
+      distance: Cosine
+      index_chunk_size: 50000
   is_demo: false
   collection_name: wiki
   embedding_path: embedding/embedding.npy

examples/demos/qdrant_index.yaml

@@ -0,0 +1,10 @@
+# Qdrant Embedding and Indexing Demo for UltraRAG UI
+
+# MCP Server
+servers:
+  retriever: servers/retriever
+
+# MCP Client Pipeline
+pipeline:
+- retriever.retriever_init
+- retriever.retriever_index

examples/experiments/qdrant_index.yaml

@@ -0,0 +1,10 @@
+# Qdrant Embedding and Indexing Demo for UltraRAG UI
+
+# MCP Server
+servers:
+  retriever: servers/retriever
+
+# MCP Client Pipeline
+pipeline:
+- retriever.retriever_init
+- retriever.retriever_index

pyproject.toml

@@ -43,10 +43,12 @@ retriever = [
     "sentence-transformers",
     "openai",
     "bm25s",
-    "faiss-gpu-cu12",
+    "faiss-gpu-cu12; sys_platform != 'darwin'",
+    "faiss-cpu; sys_platform == 'darwin'",
     "exa_py",
     "tavily-python",
     "pymilvus",
+    "qdrant-client",
     "numba",
     "torch",
     "fastapi",

script/deploy_retriever_config.json

@@ -58,6 +58,15 @@
         "params": {}
       },
       "index_chunk_size": 50000
+    },
+    "qdrant": {
+      "url": null,
+      "path": "index/qdrant",
+      "api_key": null,
+      "collection_name": "ultrarag_embeddings",
+      "text_field_name": "contents",
+      "distance": "Cosine",
+      "index_chunk_size": 50000
     }
   },
 

servers/retriever/parameter.yaml

@@ -31,7 +31,7 @@ backend_configs:
     tokenizer: auto # options: auto, default, jieba, character; auto uses jieba for Chinese
     save_path: index/bm25
 
-index_backend: faiss # options: faiss, milvus
+index_backend: faiss # options: faiss, milvus, qdrant
 index_backend_configs:
   faiss:
     index_use_gpu: True
@@ -53,6 +53,13 @@ index_backend_configs:
       metric_type: IP
       params: {}
     index_chunk_size: 1000
+  qdrant:
+    url: null # http://localhost:6333 for server, or leave null for local path
+    path: index/qdrant # local persistent storage; ignored when url is set
+    api_key: null
+    text_field_name: contents
+    distance: Cosine # options: Cosine, Dot, Euclid, Manhattan
+    index_chunk_size: 50000
 
 websearch_backend: tavily # options: tavily, exa, zhipuai
 websearch_backend_configs:

servers/retriever/src/index_backends/__init__.py

@@ -8,6 +8,7 @@
 _INDEX_BACKENDS: Dict[str, str] = {
     "faiss": ".faiss_backend.FaissIndexBackend",
     "milvus": ".milvus_backend.MilvusIndexBackend",
+    "qdrant": ".qdrant_backend.QdrantIndexBackend",
 }