M3-Embedding is a single model for multi-lingual, multi-functional, and multi-granular text embeddings trained via self-knowledge distillation that achieves new state-of-the-art results on multilingual, cross-lingual, and long-document retrieval benchmarks.
arXiv preprint arXiv:2110.03611 , year=
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
verdicts
UNVERDICTED 3representative citing papers
DACLR improves multimodal evidence retrieval for fact checking via MLLM text conversion, event-level feature extraction, two-stage recall-rerank, and dynamic contrastive losses at semantic and event levels with hard negative mining.
Stratified sampling preserving teacher score distribution outperforms hard-negative mining as a robust baseline for knowledge distillation in dense retrieval.
citing papers explorer
-
M3-Embedding: Multi-Linguality, Multi-Functionality, Multi-Granularity Text Embeddings Through Self-Knowledge Distillation
M3-Embedding is a single model for multi-lingual, multi-functional, and multi-granular text embeddings trained via self-knowledge distillation that achieves new state-of-the-art results on multilingual, cross-lingual, and long-document retrieval benchmarks.
-
Checking Fact with Better Retrieval: Dynamic Contrastive Learning for Evidence Retrieval
DACLR improves multimodal evidence retrieval for fact checking via MLLM text conversion, event-level feature extraction, two-stage recall-rerank, and dynamic contrastive losses at semantic and event levels with hard negative mining.
-
Beyond Hard Negatives: The Importance of Score Distribution in Knowledge Distillation for Dense Retrieval
Stratified sampling preserving teacher score distribution outperforms hard-negative mining as a robust baseline for knowledge distillation in dense retrieval.