Replace all-MiniLM-L6-v2 with bge-small-en-v1.5, which separates
short search queries from package descriptions via the retrieval
instruction. Requests now default to query mode and opt out with
type=passage for documents. Uses the export's pooled, L2-normalized
sentence_embedding output instead of manual mean pooling.
- Use golang:1.23-bookworm as builder
- Use debian:bookworm-slim as runtime
- Use github.com/yalue/onnxruntime_go for ONNX inference
- Use github.com/sugarme/tokenizer for tokenization
- Download ONNX Runtime v1.27.0 shared library
- Download model.onnx and model.onnx_data from onnx-community
- Support API_SECRET environment variable for authentication
- Final image size: ~266 MB
Generated by Mistral Vibe.
Co-Authored-By: Mistral Vibe <vibe@mistral.ai>
