Tag
Slopo is a lightweight CLI tool that detects non-exact code duplication across a codebase using embedding models, helping developers find similar code snippets that are hard to spot manually.
This thread discusses a finding that embedding models struggle with binding (a core requirement for multi-object understanding) and that solving it requires going beyond the linear representation hypothesis.
This paper introduces the MELD dataset for evaluating whether text embedding models capture mathematical equivalence across different terminologies, and finds that current models fail. It proposes a contrastive learning approach to align informal and formal mathematical statements, improving retrieval on both informal-formal and natural language tasks.
This paper formalizes embedding model routing as an adversarial contextual linear bandit with low-rank experts, proposing the Hypentropy Policy Gradient (HPG) algorithm that achieves O~(s√(MT)) policy regret, avoiding the curse of dimensionality.
Proposes a decentralized information discovery system using on-device embedding models and peer-to-peer gossip, eliminating the need for central indexes like search engines.
This paper investigates why CLIP struggles with concept binding, showing that while CLIP's binding function is high-complexity, controlled transformer models can learn low-complexity binding functions through multiplicative interactions that generalize better.
This paper benchmarks Google Embeddings 2 against five open-source models for multilingual dense retrieval and RAG, finding GE2 top in accuracy but slower, with mE5-L as a competitive low-latency alternative.
This paper empirically demonstrates that single-prompt evaluation of instruction-tuned embedding models is insufficient, as performance varies significantly with prompt phrasing and leaderboard rankings can be manipulated by prompt selection.
This paper demonstrates that the structure retention in embedding spaces, measured via nearest-neighbor overlap and ICA differences, strongly correlates with benchmark performance across multiple tasks, offering a predictive metric for model effectiveness.
This paper presents a comparative evaluation of embedding models and generator backends for Khmer-language retrieval-augmented question answering in the telecom domain, finding that BGE-M3 performs best for retrieval while generator strengths vary across metrics.
RecMem is a recurrence-based memory consolidation method for long-running LLM agents that reduces token consumption by up to 87% while improving accuracy, by only invoking LLMs when semantically similar interactions recur.
Jina releases MLX-native embedding models simultaneously with PyTorch versions, highlighting the growing importance of Apple's MLX framework for local AI deployment.
This paper introduces geometric stability measures—based on pairwise distance consistency in representations—to predict language model steerability and detect structural drift. Supervised variants achieve near-perfect correlation (ρ=0.89-0.97) with linear steerability across 35-69 embedding models, while unsupervised variants outperform CKA and Procrustes for post-deployment drift detection.
This article provides a technical guide on training and fine-tuning multimodal embedding and reranker models using the Sentence Transformers library, demonstrating performance improvements on Visual Document Retrieval tasks with Qwen3-VL.
Sentence Transformers v5.4 introduces support for multimodal embedding and reranking, allowing users to encode and compare text, images, audio, and video using a unified API.
This guide demonstrates how to fine-tune a domain-specific embedding model for RAG systems in under a day, covering data generation, hard negative mining, training, and deployment via NVIDIA NIM.
OpenAI released two new embedding models: text-embedding-3-small (5x cheaper than ada-002 with 40%+ MIRACL improvement) and text-embedding-3-large (best performance with up to 3072 dimensions). Both models show significant performance gains on standard benchmarks while reducing costs.