ProactiveCoach: Hierarchical Guidance Boosts Proactive AI Assistants by 57.1 Points
skku · hf · 2026-10-08
SKKU researchers introduce ProactiveCoach, a suite for building proactive AI assistants that know when to speak and at what granularity.
- Gap: existing datasets either focus on detection-based proactive understanding or offer fixed-granularity procedural guidance, making it hard to determine task completion and adapt guidance level.
- Suite: ProactiveCoach-Instruct provides hierarchically structured guidance at phase/step/action levels; ProactiveCoachBench evaluates whether models give appropriate guidance at the right time and adapt when the requested level changes.
- Results: hierarchical supervision improves performance across backbones by up to 9.6 points over fixed-granularity; combining the fine-tuned VLM with a lightweight guidance router outperforms in-context adaptation baselines by 57.1 points across four guidance-level transitions, with no extra fine-tuning.
Project page: jinsuby.github.io/ProactiveCoach/.
More from Research
- FractAL introduces soft acquisition-strategy selection for batch-mode active learning — anshulkundaje · 2026-10-08
- DeLM's decentralized multi-agent system runs 2.49x faster, but MAS evals pick wildly different metrics — jyangballin · 2026-10-08
- Study: RAG retrieval diversification helps only on redundant multi-evidence pools, paper proposes per-query rule — _reachsumit · 2026-10-08
- Quantize by Drift: label-free mixed-precision quantization for text embedders hits 0.911 Spearman — _reachsumit · 2026-10-08
- RunningTab: environment-side task ledger consistently beats in-model tracking across 3 benchmarks and 3 LLMs — RexDouglass · 2026-10-08
- ColPali-style visual document indices can be inverted: 47% of words recovered, source page ranked first 98.4% — _reachsumit · 2026-10-08