New frameworks evaluate models by skills and dialogue success rather than single scores, while data selection and clarification improve fine-tuning and RAG reliability.
An embedding-based method using Whisper's encoder selects smaller, representative training subsets for ASR fine-tuning. In Hebrew and Danish, k-center selection often outperforms random sampling, especially at smaller budgets. Lexical diversity analy...
A fully automated framework shows that on 5,000+ simulated dialogues, Remediation Success Rate moves from 42.1% under exact-match to 81.3% under human-calibrated LLM judges. The 39.2-point gap exceeds the combined effect of model, teaching strategy, ...
IBM Research releases ClariEdge, a dataset of about 3,900 examples for spotting ambiguous queries in RAG. A lightweight LoRA obtained through fine-tuning reaches 96.6% accuracy and outperforms larger baselines. The signal: specialized local component...
IBM Research has released ClariEdge, a dataset of about 3,900 examples for recognizing ambiguous queries in RAG systems and generating clarification requests. A lightweight LoRA reaches 96.6% classification accuracy and beats prompted baselines from ...
SkillEval introduces an evaluation framework based on skill profiles automatically discovered by LLMs and estimated with cognitive diagnosis models. Tested on 3,811 models and 9,523 items, it produces more calibrated and predictive profiles. Applicat...
