YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
BGE-Reranker-v2-m3（OpenVINO + ONNX）
Unified Hugging Face release for RerankServer / llm-init.
Layout matches beclab/embeddinggemma-300m: one logical model, two complete format trees. FP16 exports are not included.
| Subdirectory | Format | Runtime | Key files |
|---|---|---|---|
| openvino/ | OpenVINO IR FP32 | RERANK_RUNTIME=openvino (ACCELERATOR=intel-gpu) |
openvino_model.xml, openvino_model.bin |
| onnx/ | ONNX FP32 | RERANK_RUNTIME=onnx (cpu / nvidia / nvidia-gb10) |
model.onnx, model.onnx_data |
Logical model name (MODEL_ID / OpenAI model): bge-reranker-v2-m3.
RerankServer
MODEL_DIR must point at the format subdirectory (openvino/ or onnx/), not this repo root.
Tokenizer lives in the same subdirectory (tokenizer.json).
llm-init
MODEL_SOURCE=hf://beclab/bge-reranker-v2-m3
MODEL_NAME=bge-reranker-v2-m3
MODEL_MODE=rerank
License
Same as BAAI/bge-reranker-v2-m3.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support