Papersby GlyphRetrieval and data
Embeddings, RAG, search over documents and the data models learn from. 29 papers from 19 labs.
Retrieval Augmented Generation CompositionalityThis study shows that individual RAG techniques do not always work well together and provides a framework to evaluate full retrieval pipelines.
IBM34 upvotesRead →UNREAL: Unifying Retrieval and Long-Context with a Single ModelUNREAL uses a single model-internal mechanism to perform both corpus retrieval and long-context evidence selection, improving performance while reducing computational costs.
NVIDIA31 upvotesRead →RPTune: Learned Context Curation for LLM Catalog SearchRPTune improves LLM search accuracy by learning how to curate product catalogs and adapting the model to those curated contexts.
Google40 upvotesRead →Follow the Entities: A Corpus Map for Agentic SearchCorpusMap improves agentic search by organizing documents into a graph based on recurring entities, allowing agents to traverse relationships between files.
Microsoft100 upvotesRead →Dream-RSI: Recursive Self-Improvement through Evolving WorldsDream-RSI enables scalable recursive self-improvement by using historical discovery replay to evaluate exploration policies offline, reducing costly online evaluations.Google24 citations246 upvotesRead →It Takes Two to Match: Co-Evolving Generative Retriever with Reinforcement LearningCoGR trains LLMs to generate compact keywords for both queries and items, enabling direct inverted-index retrieval optimized via co-evolving reinforcement learning.
Apple2 citations65 upvotesRead →StarHarness: Evolving Harnesses with Stratified Search for Enterprise EnvironmentsStarHarness evolves fixed-weight agent harnesses via stratified task pools and hidden selection to improve enterprise tool-use performance and cross-model transfer.
ServiceNow3 citations41 upvotesRead →ResearchStudio-Idea: An Evidence-Grounded Research-Ideation Skill Suite from ML Conference OutcomesResearchStudio-Idea provides a skill suite for effective research ideation that combines literature search, novelty checking, and pattern-guided generation to produce traceable research proposals.Microsoft3 citations62 upvotesRead →CUE-R: Beyond the Final Answer in Retrieval-Augmented GenerationResearchers developed a framework to measure the operational utility of individual retrieved items in retrieval-augmented generation systems by perturbing evidence and analyzing changes in correctness, grounding faithfulness, and confidence error.
Intuit3 citations3 upvotesRead →Qianfan-OCR: A Unified End-to-End Model for Document IntelligenceQianfan-OCR is a 4B-parameter vision-language model that unifies document parsing, layout analysis, and understanding while maintaining strong performance across multiple OCR benchmarks through its Layout-as-Thought mechanism.
Baidu25 citations155 upvotesRead →jina-embeddings-v5-text: Task-Targeted Embedding DistillationCompact text embedding models are developed through a combined training approach using distillation and contrastive loss, achieving state-of-the-art performance while supporting long-context sequences and efficient quantization.
Jina AI28 citations28 upvotesRead →Learning to Discover at Test TimeTest-time training enables AI systems to discover optimal solutions for specific scientific problems through continual learning focused on individual challenges rather than generalization.
Stanford University118 citations45 upvotesRead →Action100M: A Large-scale Video Action DatasetAction100M is a large-scale video action dataset constructed from internet instructional videos using automated pipelines with V-JEPA embeddings and GPT-based reasoning for structured annotations.
Meta16 citations33 upvotesRead →Applying Embedding-Based Retrieval to Airbnb SearchAn embedding-based retrieval system was developed for Airbnb search to efficiently match guests with accommodations while addressing challenges like dynamic inventory and multi-stage user journeys.
Airbnb2 citations1 upvotesRead →Qwen3-VL-Embedding and Qwen3-VL-Reranker: A Unified Framework for State-of-the-Art Multimodal Retrieval and RankingThe Qwen3-VL-Embedding and Qwen3-VL-Reranker models form an end-to-end multimodal search pipeline, leveraging multi-stage training and cross-attention mechanisms to achieve high-precision retrieval across diverse modalities.
Alibaba (Qwen)315 citations59 upvotesRead →Pico-Banana-400K: A Large-Scale Dataset for Text-Guided Image EditingPico-Banana-400K is a large-scale, high-quality dataset for instruction-based image editing, featuring diverse edit pairs, multi-turn editing, preference subsets, and long-short instruction pairs, enabling comprehensive research and benchmarking.Apple68 citations30 upvotesRead →The Art of Asking: Multilingual Prompt Optimization for Synthetic DataPrompt-space optimization enhances multilingual large language models by systematically transforming prompts for naturalness, cultural adaptation, and difficulty, leading to improved performance across various metrics.
Cohere4 citations1 upvotesRead →DeepCodeSeek: Real-Time API Retrieval for Context-Aware Code GenerationA novel technique for predicting APIs and generating code in real-time using a compact reranker outperforms larger models with reduced latency, addressing API leaks and unclear usage intent in enterprise code.ServiceNow5 upvotesRead →Optimizing What Matters: AUC-Driven Learning for Robust Neural RetrievalA new training objective, MW loss, is introduced to improve retriever calibration and ranking quality by directly optimizing the Area under the ROC Curve (AUC), outperforming Contrastive Loss in retrieval-augmented generation tasks.ServiceNow4 upvotesRead →jina-reranker-v3: Last but Not Late Interaction for Document RerankingA multilingual document reranker using causal self-attention achieves state-of-the-art performance with a compact architecture.Jina AI45 citations13 upvotesRead →LEAF: Knowledge Distillation of Text Embedding Models with Teacher-Aligned RepresentationsLEAF, a knowledge distillation framework for text embeddings, enables flexible asymmetric architectures and robustness, achieving state-of-the-art performance in information retrieval and multi-task settings with minimal training requirements.
MongoDB9 citationsRead →DINOv3DINOv3, a self-supervised learning model, achieves superior performance across various vision tasks by scaling datasets and models, addressing dense feature degradation, and enhancing flexibility with post-hoc strategies.Meta1,932 citations320 upvotesRead →Planning In Natural Language Improves LLM Search For Code GenerationPLANSEARCH, a novel search algorithm, improves performance in coding benchmarks by generating a diverse set of natural language plans, outperforming existing methods through increased diversity in solutions.
Scale AI101 citationsRead →DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree SearchDeepSeek-Prover-V1.5 improves theorem proving by optimizing training and inference, utilizing reinforcement learning, and proposing RMaxTS for diverse proof paths, achieving state-of-the-art results on miniF2F and ProofNet benchmarks.
DeepSeek229 citations64 upvotesRead →DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic DataGenerating extensive Lean 4 proof data from mathematical competition problems improved DeepSeekMath 7B's theorem-proving capabilities over GPT-4 and other methods.DeepSeek284 citations47 upvotesRead →Arctic-Embed: Scalable, Efficient, and Accurate Text Embedding ModelsOpen text embedding models from 22M to 334M parameters that led MTEB retrieval for their size at release.
Snowflake84 citations2 upvotesRead →From Local to Global: A Graph RAG Approach to Query-Focused SummarizationGraphRAG builds a knowledge graph of a document set so a model can answer questions about the whole collection.Microsoft2,561 citations7 upvotesRead →Aya Dataset: An Open-Access Collection for Multilingual Instruction TuningThe initiative builds a human-curated instruction-following dataset spanning 65 languages and creates the largest multilingual collection of instruction-following instances through templating and translating existing datasets across 114 languages, contributing datasets and platforms for participatory research.Cohere253 citations57 upvotesRead →BLOOM: A 176B-Parameter Open-Access Multilingual Language ModelBLOOM, a 176B-parameter open-access language model, achieves competitive performance on various benchmarks and is released publicly with a Responsible AI License.
Hugging Face3,165 citations41 upvotesRead →