docs: README + config cleanup

.env.example

@@ -14,9 +14,9 @@ NEO4J__DATABASE=neo4j
 QDRANT__URL=your_qdrant_url_here
 QDRANT__API_KEY=your_qdrant_api_key_here
 QDRANT__COLLECTION_NAME=biomedical_papers
-QDRANT__EMBEDDING_MODEL=text-embedding-3-large
-QDRANT__EMBEDDING_DIMENSION=1536
-QDRANT__RERANKER_EMBEDDING_DIMENSION=3072
+QDRANT__EMBEDDING_MODEL=text-embedding-3-small
+QDRANT__EMBEDDING_DIMENSION=1024
+QDRANT__RERANKER_EMBEDDING_DIMENSION=1536
 QDRANT__ESTIMATE_BM25_AVG_LEN_ON_X_DOCS=300
 QDRANT__CLOUD_INFERENCE=true
 
@@ -26,4 +26,4 @@ PUBMED__EMAIL=your_email@example.com
 
 # JSON Data Paths (optional — defaults are data/pubmed_dataset.json and data/gene_dataset.json)
 JSON_DATA__PUBMED_JSON_PATH=data/pubmed_dataset.json
-JSON_DATA__GENE_JSON_PATH=data/gene_dataset.json
+JSON_DATA__GENE_JSON_PATH=data/gene_dataset.json

README.md

@@ -12,9 +12,9 @@
 
 <!-- Providers -->
 
-[![Qdrant](https://img.shields.io/badge/Qdrant-1.15.1-5A31F4?logo=qdrant&logoColor=white)](https://qdrant.tech/)
-[![Neo4j](https://img.shields.io/badge/Neo4j-5.28.2-008CC1?logo=neo4j&logoColor=white)](https://neo4j.com/)
-[![OpenAI](https://img.shields.io/badge/OpenAI-2.3.0-412991?logo=openai&logoColor=white)](https://openai.com/)
+[![Qdrant](https://img.shields.io/badge/Qdrant-5A31F4?logo=qdrant&logoColor=white)](https://qdrant.tech/)
+[![Neo4j](https://img.shields.io/badge/Neo4j-008CC1?logo=neo4j&logoColor=white)](https://neo4j.com/)
+[![OpenAI](https://img.shields.io/badge/OpenAI-412991?logo=openai&logoColor=white)](https://openai.com/)
 
 </div>
 
@@ -105,7 +105,7 @@ biomedical-graphrag/
 1. Clone the repository:
 
    ```bash
-   git clone git@github.com:thierrypdamiba/biomedical-graphrag.git
+   git clone git@github.com:qdrant-labs/biomedical-graphrag.git
    cd biomedical-graphrag
    ```
 
@@ -156,9 +156,9 @@ NEO4J__DATABASE=neo4j
 QDRANT__URL=http://localhost:6333
 QDRANT__API_KEY=your_qdrant_api_key
 QDRANT__COLLECTION_NAME=biomedical_papers
-QDRANT__EMBEDDING_MODEL=text-embedding-3-large
-QDRANT__EMBEDDING_DIMENSION=1536
-QDRANT__RERANKER_EMBEDDING_DIMENSION=3072
+QDRANT__EMBEDDING_MODEL=text-embedding-3-small
+QDRANT__EMBEDDING_DIMENSION=1024
+QDRANT__RERANKER_EMBEDDING_DIMENSION=1536
 QDRANT__ESTIMATE_BM25_AVG_LEN_ON_X_DOCS=300
 QDRANT__CLOUD_INFERENCE=false
 
@@ -227,9 +227,7 @@ Notes:
 - This project uses OpenAI embeddings **Matryoshka Representation Learning (MRL)** feature:
   - `QDRANT__EMBEDDING_DIMENSION` is the prefix dimension used for **retrieval** (stored in Qdrant as the `Dense` vector).
   - `QDRANT__RERANKER_EMBEDDING_DIMENSION` is the (larger) prefix dimension used for **reranking** (stored in Qdrant as the `Reranker` vector).
-- `make ingest-qdrant-data` currently recreates the collection each run (see `qdrant_ingestion.py`).
-  If you don't want that, change `recreate=True` to `False`. There's also an `only_new` parameter which defaults to `True`, so we ingest only papers whose PMID is not already
-  present in the collection. Set `only_new=False` if you'd prefer to overwrite existing points or on a clean ingestion (then it should be `False`!)
+- `make ingest-qdrant-data` recreates the collection each run (`recreate=True` in `qdrant_ingestion.py`); set it to `False` to keep existing points. The `only_new` flag (default `True`) skips PMIDs already in the collection.
 - The collection is configured by default with **scalar quantization** (compressed dense vectors).
 - `QDRANT__CLOUD_INFERENCE=true` enables **Qdrant Cloud Inference** when embeddings are computed by Qdrant Cloud.
 - `QDRANT__ESTIMATE_BM25_AVG_LEN_ON_X_DOCS` controls how many documents are sampled to estimate the average abstract length used by **BM25**. This helps calibrate BM25-based scoring when using dense+BM25 hybrid retrieval.
@@ -290,12 +288,14 @@ make run-api
 ```bash
 curl -X POST http://localhost:8765/api/graphrag-query \
   -H "Content-Type: application/json" \
-  -d '{"query": "What genes are associated with breast cancer?", "limit": 5}'
+  -d '{"query": "What genes are associated with breast cancer?", "limit": 5, "openai_api_key": "sk-..."}'
 ```
 
+> `openai_api_key` is required on every request. It is used transiently and never stored or logged. A rejected key returns HTTP 401 with `detail: openai_key_rejected`.
+
 ### Frontend
 
-The frontend is maintained in a separate repository: **[biomedical-graphrag-frontend](https://github.com/thierrypdamiba/biomedical-graphrag-frontend)**
+The frontend is maintained in a separate repository: **[biomedical-graphrag-frontend](https://github.com/qdrant-labs/biomedical-graphrag-frontend)**
 
 The quickest way to run it locally:
 
@@ -307,13 +307,13 @@ make run-frontend
 Or manually:
 
 ```bash
-git clone https://github.com/thierrypdamiba/biomedical-graphrag-frontend.git
+git clone https://github.com/qdrant-labs/biomedical-graphrag-frontend.git
 cd biomedical-graphrag-frontend
 pnpm install
 pnpm dev
 ```
 
-The frontend connects to the hosted backend at `https://biomedical-graphrag-9qqm.onrender.com` by default, or you can point it to a local backend via `GRAPHRAG_API_URL=http://localhost:8765`.
+The frontend requires `GRAPHRAG_API_URL` to point at this backend.
 
 ### Docker
 
@@ -330,6 +330,10 @@ docker run --rm -p 8765:8765 --env-file .env biomedical-graphrag:latest
 curl http://localhost:8765/health
 ```
 
+### Deploy to Vercel
+
+Runs as a container via `Dockerfile.vercel`.
+
 ### Troubleshooting
 
 - **Make fails immediately with ".env file is missing"**

src/biomedical_graphrag/config.py

@@ -32,13 +32,13 @@ class QdrantSettings(BaseModel):
         default="biomedical_papers", description="Collection name for Qdrant instance"
     )
     embedding_model: str = Field(
-        default="text-embedding-3-large", description="OpenAI embedding model to use"
+        default="text-embedding-3-small", description="OpenAI embedding model to use"
     )
     embedding_dimension: int = Field(
-        default=1536, description="Dimension of the OpenAI embedding for retrieval (using OpenAI's MRL)"
+        default=1024, description="Dimension of the OpenAI embedding for retrieval (using OpenAI's MRL)"
     )
     reranker_embedding_dimension: int = Field(
-        default=3072,
+        default=1536,
         description="Dimension of the OpenAI embedding for reranking (using OpenAI's MRL)",
     )
     estimate_bm25_avg_len_on_x_docs: int = Field(

tests/integration/test_config.py

@@ -22,9 +22,9 @@ def test_settings_creation_with_defaults() -> None:
     qdrant_settings = QdrantSettings()
     assert qdrant_settings.url == "http://localhost:6333"
     assert qdrant_settings.collection_name == "biomedical_papers"
-    assert qdrant_settings.embedding_model == "text-embedding-3-large"
-    assert qdrant_settings.embedding_dimension == 1536
-    assert qdrant_settings.reranker_embedding_dimension == 3072
+    assert qdrant_settings.embedding_model == "text-embedding-3-small"
+    assert qdrant_settings.embedding_dimension == 1024
+    assert qdrant_settings.reranker_embedding_dimension == 1536
     assert qdrant_settings.estimate_bm25_avg_len_on_x_docs == 300
     assert qdrant_settings.cloud_inference is False
 
@@ -76,9 +76,9 @@ def test_qdrant_settings_validation() -> None:
 
     assert qdrant_settings.url == "http://localhost:6333"
     assert qdrant_settings.collection_name == "biomedical_papers"
-    assert qdrant_settings.embedding_model == "text-embedding-3-large"
-    assert qdrant_settings.embedding_dimension == 1536
-    assert qdrant_settings.reranker_embedding_dimension == 3072
+    assert qdrant_settings.embedding_model == "text-embedding-3-small"
+    assert qdrant_settings.embedding_dimension == 1024
+    assert qdrant_settings.reranker_embedding_dimension == 1536
     assert qdrant_settings.estimate_bm25_avg_len_on_x_docs == 300
     assert qdrant_settings.cloud_inference is False
     assert qdrant_settings.api_key.get_secret_value() == ""

tests/integration/test_databases.py

@@ -125,9 +125,9 @@ def test_qdrant_settings_validation(self) -> None:
 
         assert qdrant_settings.url == "http://localhost:6333"
         assert qdrant_settings.collection_name == "biomedical_papers"
-        assert qdrant_settings.embedding_model == "text-embedding-3-large"
-        assert qdrant_settings.embedding_dimension == 1536
-        assert qdrant_settings.reranker_embedding_dimension == 3072
+        assert qdrant_settings.embedding_model == "text-embedding-3-small"
+        assert qdrant_settings.embedding_dimension == 1024
+        assert qdrant_settings.reranker_embedding_dimension == 1536
         assert qdrant_settings.estimate_bm25_avg_len_on_x_docs == 300
         assert qdrant_settings.cloud_inference is False