{"slug":"retrieval-augmented-generation","label":"Retrieval-Augmented Generation","item_count":3,"day_count":3,"source_count":3,"first_seen":"2026-10-29T16:40:02+00:00","last_updated":"2026-11-06T17:56:37+00:00","generated_at":"2026-11-10T00:13:24.787725+00:00","sources":["arxiv_cs_ai","arxiv_cs_cl","arxiv_llm_reliability"],"days":[{"date":"2026-10-29","items":[{"title":"Retrieving from Your Own Mistakes: On-Policy Training of Embedding Models for Multi-Hop Retrieval-Augmented Generation","url":"http://arxiv.org/abs/2610.40409v1","source":"arxiv_cs_ai","type":"paper","summary_1line":"Complex questions in retrieval-augmented generation require evidence spread over several documents, and an answer is reliable only when the retriever finds all of it. Chain embedding retrievers collect such evidence s...","why_it_matters":"Matches feed focus: agent, eval.","sid":"cc7fe353e9f92660","published":"2026-10-29T16:40:02+00:00","editor_note":"Targets multi-hop RAG by training the embedding retriever on its own failures."}]},{"date":"2026-11-02","items":[{"title":"TRIO: Task Refinement with Individual Optimization for Retrieval-Augmented Generation","url":"http://arxiv.org/abs/2611.02810v1","source":"arxiv_llm_reliability","type":"paper","summary_1line":"This paper studies retrieval-augmented generation (RAG), which leverages external knowledge to improve the performance of language models. Despite the remarkable progress, current RAG approaches are still far from sat...","why_it_matters":"Matches feed focus: agent, eval.","sid":"401ca1e322883a2d","published":"2026-11-02T09:17:22+00:00","editor_note":"Frames current RAG as unsatisfactory and proposes TRIO for task refinement."}]},{"date":"2026-11-06","items":[{"title":"Compaction, decay and reinforcement in a two-tier conversational memory: a paired evaluation against retrieval-augmented generation","url":"http://arxiv.org/abs/2611.09807v1","source":"arxiv_cs_cl","type":"paper","summary_1line":"A conversational agent that remembers has to decide what to keep. Retrieval-augmented generation keeps the transcript and searches it; the alternative is to compact each turn into atomic facts and search those. We com...","why_it_matters":"Matches feed focus: agent, evaluation.","sid":"49e633d61cb9577d","published":"2026-11-06T17:56:37+00:00","editor_note":"Moves the question from retrieval to memory: transcript search versus compacted atomic facts."}]}],"editorial":{"tldr":"Late-October research targeted the retriever: a paper trained embedding models on their own mistakes so multi-hop questions find all the evidence they need. A November 2 paper, TRIO, then argued that current RAG approaches are still far from satisfactory.","stale":false,"whats_new":"A Nov 6 paper compares RAG over a raw transcript against a two-tier memory that compacts each turn into atomic facts, testing whether conversational agents should keep the transcript at all.","why_it_matters":"RAG design choices (what you index, how the retriever is trained, what an agent remembers) are now being tested against alternatives, so the default pipeline is no longer a safe assumption.","take_for_builders":"If you run conversational memory over a raw transcript, rerun your own recall evals against a compacted-facts variant before changing anything; treat these papers as hypotheses until reproduced on your data.","status":{"state":"Developing","tone":"rising","changed":"2026-11-06","detail":"Three arXiv papers in nine days probe different RAG layers: retriever training, task refinement, and agent memory."},"beats":[{"kicker":"RETRIEVER","tone":"launch","headline":"Training embedding models on their own retrieval mistakes for multi-hop RAG","summary":"Multi-hop answers are only reliable if the retriever finds all the evidence.","sids":["cc7fe353e9f92660"]},{"kicker":"PIPELINE","tone":"rising","headline":"TRIO proposes task refinement with individual optimization for RAG","sids":["401ca1e322883a2d"]},{"kicker":"NOW","tone":"now","headline":"Two-tier conversational memory evaluated head-to-head against RAG","summary":"Compacting turns into atomic facts versus searching the stored transcript.","sids":["49e633d61cb9577d"]}],"open_questions":["Do the multi-hop retriever gains hold on benchmarks other than those the paper reports?","Does compacted fact memory beat transcript RAG outside the paper's paired evaluation?","Will any of these methods ship in a widely used RAG framework?"],"generated_at":"2026-11-10T00:30:00+00:00"}}