Weekly IR papers: Huawei's SearchJev offloads search-agent judgments to sub-4B models
_reachsumit · x · 2026-10-12
Vol. 177 of "Top Information Retrieval Papers of the Week" rounds up 10 studies:
- SearchJev (Huawei): fine-tunes Qwen3.5-0.8B/4B to handle search agents' routine judgments (relevance, sufficiency, next action) by reading first-token logits as calibrated probabilities — a fast System-1 approach
- Programmatic Search Agents (Tencent): extends agentic search beyond query reformulation
- Training-Free Semantic IDs for generative recommendation (Amazon)
- Measuring accuracy and cost of LLM-driven retrieval (NVIDIA)
- RL over embeddings for dense retrieval (Renmin University)
- An LLM-agent optimizer for RAG pipelines with per-question failure diagnosis (ETH Zurich)
- Plus a survey of in-parameter memory, when retrieval diversification helps vs hurts, an open agentic retriever, and learned sparsity controls for sparse retrievers
More from coding & agent
- Gemini audits an Audemars Piguet product page: title tags, hreflang, JS rendering and more — gaganghotra_ · 2026-10-12
- Claude Code tiered workflow: Haiku swarms, Sonnet builds, Opus reviews at 3 checkpoints — Arindam_1729 · 2026-10-12
- The Opus 5.5 Flywheel: Turning Every Agent Correction into Reusable Skills — ramagetime · 2026-10-12
- Burkov uses Codex to translate an 8-year-old Spanish microprocessors textbook into English with interactive AI-tutor exercises — burkov · 2026-10-12
- Grok plays WoW autonomously, now free and heading to Stormwind — djcows · 2026-10-12
- Insomnia: open-source tool keeps Mac agents running with the lid closed — _AustinCalvert_ · 2026-10-12