Update Hugging Face org references: llm-semantic-router → vllm-sr

Improve Vela Reranker compact heads and refresh evaluation scores

Keep ROCm efficient attention inputs contiguous in the Vela loader

Improve multilingual and long-context reranking

Improve multilingual retrieval and long-context reranking

Add ROCm-accelerated ONNX representations

Add portable FP32 ONNX models

Add original mmBERT evaluation comparison for Vela Reranker

Load Vela Reranker directly with Transformers

Simplify the Vela Reranker model card

Release Vela on the unified Encoder base

Unify Vela model cards with Semantic Router branding

Add Vela Embedding to the model family

Add Vela Feedback and update Reranker performance reference

Release Vela 1.0 Reranker

initial commit