Fix: Update to RoSBERTa model (1024 dims), remove half precision, increase timeout 93cd57d Calcifer0323 commited on 15 days ago
fix: update sentence-transformers>=2.7.0 for ru-en-RoSBERTa compatibility 9ebcd3b Calcifer0323 commited on Dec 19, 2025
v2.2.0: ai-forever/ru-en-RoSBERTa + normalize_embeddings + cache fix 8b656e5 Calcifer0323 commited on Dec 19, 2025
fix: remove model.half() - LayerNorm not supported for Half on CPU 9ca349e Calcifer0323 commited on Dec 14, 2025
Simplify embedding service to stateless, add backend integration docs 345a8d5 Calcifer0323 commited on Dec 12, 2025
Simplify to stateless service - remove storage, keep only embedding generation 4e4150b Calcifer0323 commited on Dec 12, 2025
feat: add HuggingFace deployment files and documentation 6ee1a1d Calcifer0323 commited on Dec 10, 2025
fix: update dependencies and use L12-v2 model (better quality, 16GB RAM is enough) 8e21db2 Calcifer0323 commited on Dec 10, 2025
Fix README.md: proper YAML metadata for HuggingFace Space 8b31749 Calcifer0323 commited on Dec 10, 2025
Deploy FastAPI embedding service: optimized for HuggingFace with Docker, torch 2.1.2, numpy<2 e46589a Calcifer0323 commited on Dec 10, 2025
fix: pin compatible versions - numpy<2, torch 2.1.2, transformers 4.36.2 for Render free tier b51d0f9 Calcifer0323 commited on Dec 10, 2025
fix: optimize for Render free tier - pin numpy<2, use L6-v2 model, CPU-only torch a180bc9 Calcifer0323 commited on Dec 10, 2025
Fix: Configure Render for Python runtime instead of Docker, add Swagger documentation info 1c5dbef Calcifer0323 commited on Dec 10, 2025
Initial commit: Embedding service ready for Render deployment 9cf5488 Calcifer0323 commited on Dec 10, 2025
