Upload 37 files

Dockerfile

@@ -10,7 +10,7 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
 # Copy all files first
 COPY . .
 
-# Install dependencies directly (not editable mode)
+# Install dependencies (lightweight reranking with CPU torch)
 RUN pip install --no-cache-dir \
     fastapi>=0.115.0 \
     uvicorn[standard]>=0.32.0 \
@@ -19,7 +19,10 @@ RUN pip install --no-cache-dir \
     httpx>=0.28.0 \
     aiohttp>=3.11.0 \
     python-dotenv>=1.0.0 \
-    tenacity>=9.0.0
+    tenacity>=9.0.0 \
+    numpy \
+    sentence-transformers \
+    && pip install --no-cache-dir torch --index-url https://download.pytorch.org/whl/cpu
 
 # HuggingFace Spaces uses port 7860
 EXPOSE 7860

app/config.py

@@ -27,9 +27,9 @@ class Settings(BaseSettings):
     llm_provider: Literal["groq", "openrouter"] = "openrouter"
     llm_model: str = "nex-agi/deepseek-v3.1-nex-n1:free"
     
-    # Reranking Models
-    bi_encoder_model: str = "BAAI/bge-small-en-v1.5"
-    cross_encoder_model: str = "BAAI/bge-reranker-base"
+    # Reranking Models (lightweight for HF Spaces)
+    bi_encoder_model: str = "sentence-transformers/all-MiniLM-L6-v2"  # ~90MB
+    cross_encoder_model: str = "cross-encoder/ms-marco-MiniLM-L6-v2"  # ~90MB
     
     # Temporal Settings
     default_freshness_half_life: int = 30  # days