technique

Evidence-Diagnosed Intervention Training

techniqueactiveprovisionalevidence-diagnosed-intervention-training-82ed4c69·1 events·first seen 12d ago

Aliases: Evidence-Diagnosed Intervention Training

Co-occurring entities

EDIT: Evidence-Diagnosed Intervention Training for Rule-Faithful LLM Grading

More like this (12)

EDIT: Evidence-Diagnosed Intervention Training for Rule-Faithful LLM Grading inference-time intervention Tool-Integrated Reasoning AASIST collaborative distributed training ART (Agent Reinforcement Trainer)instruction-based multitask pretraining Technology Innovation Institute T-Eval ICO (Intervention, Comparator, Outcome) framework proactive assistance evaluation adversarial training

Recent events (1)

4arXiv · cs.CL·12d ago·source ↗

EDIT framework trains more rubric-faithful LLM graders via internal-state diagnostics

Researchers introduce Evidence-Diagnosed Intervention Training (EDIT), a two-phase framework for improving LLM-based rubric grading. The first phase (EDIT-SFT) identifies problematic reasoning steps using posterior belief signals and input-grounding scores, then revises only those steps with rubric checklists; the second phase (EDIT-RL) uses belief-guided reward shaping to penalize harmful belief drifts during RL. Experiments on two real-world multi-subject grading benchmarks show consistent improvements over SFT and RL baselines on both in-domain and out-of-domain splits.

Evaluation and Benchmarking Alignment and RLHF Evidence-Diagnosed Intervention Training EDIT: Evidence-Diagnosed Intervention Training for Rule-Faithful LLM Grading