Lightonai Embeddings Fine Tuning Reranked V1: Teacher-Rescored Query-Document Pairs

Cross-encoder created this dataset by rescoring the lightonai/embeddings-fine-tuning source with mixedbread-ai/mxbai-rerank-large-v2. The dataset provides teacher targets for distilling reranker models, specifically used upstream for cross-encoder/ettin-reranker-v1-data. It was last updated on May 19, 2026.

Use Cases

• Training reranker student models based on teacher scores from a large reranker model.
• Evaluating reranking performance based on rescored query-document pairs.
• Fine-tuning embedding models based on reranker-rescored relevance judgments.
• Benchmarking information retrieval pipelines using pre-scored candidate documents.

Strengths

• Provides teacher scores from a specific, named reranker model (mxbai-rerank-large-v2).
• Created by a known author (cross-encoder) and linked to a downstream model artifact (ettin-reranker-v1-data).

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

lightonai/embeddings-fine-tuning

Collection Method

Rescored using mixedbread-ai/mxbai-rerank-large-v2.

Freshness

Last updated 2026-05-19 15:04:23; freshness should be verified.