Specialized Reranker Variants For Latency Accuracy Trade Offs

via “fine-tuning and domain adaptation via transfer learning”

sentence-similarity model by undefined. 1,28,43,377 downloads.

Unique: Supports both LoRA (parameter-efficient, 10-15% latency overhead) and full fine-tuning while preserving 2048-token context and matryoshka properties, enabling domain adaptation without architectural changes or retraining from scratch

vs others: More efficient fine-tuning than OpenAI embeddings API (no per-token costs, full control over training) and preserves long-context capability that most sentence-transformers lose during fine-tuning due to position interpolation