DEPT Unifies Query Expansion and Retrieval via Embedding Preservation
_reachsumit · x · 2026-08-19
New research proposes DEPT (Document Embedding Preservation Tuning), training a single LLM for both query expansion and text encoding. By anchoring document embeddings to initial cached versions, it solves the "moving target" problem in joint training, allowing index reuse and online hard-negative mining. Experiments on BEIR benchmarks show DEPT outperforms training-free, independent, and staged unified baselines.
More from Research
- KDD Cup champions win using only DeepSeek web chat interface — jiqizhixin · 2026-08-19
- Awesome AI4AI: A Living, Weekly-Updated Map of AI Improving AI — No-Strawberry-2588 · 2026-08-19
- GLM-5.3 gains driven by post-training, not base model — teortaxesTex · 2026-08-19
- Chemist defends Claude's binder design: you've never actually run comp-chem software — chaitjo · 2026-08-19
- Why humanoids keep crashing into barriers: onboard vision can't keep up with sprinting legs — 2C_ornot2C · 2026-08-19
- Why LLMs can't make your code simpler, per Naur's "Programming as Theory Building" — math_rachel · 2026-08-19