Three-Step Nav uses a three-view MLLM protocol to achieve state-of-the-art zero-shot VLN performance on R2R-CE and RxR-CE by global planning, local alignment, and trajectory auditing.
[Not Applicable] (b) Complete proofs of all theoretical results
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
years
2026 3representative citing papers
Sycophantic GRPO fine-tuning degrades LLM calibration, raising ECE by 0.006 and MCE by 0.010, with a persistent residual after post-hoc scaling.
Optimizing a single scalar temperature improves semantic calibration, discrimination, and entropy in language model question-answering over heuristic baselines and token-level recalibration methods.
citing papers explorer
-
Three-Step Nav: A Hierarchical Global-Local Planner for Zero-Shot Vision-and-Language Navigation
Three-Step Nav uses a three-view MLLM protocol to achieve state-of-the-art zero-shot VLN performance on R2R-CE and RxR-CE by global planning, local alignment, and trajectory auditing.
-
Calibration Collapse Under Sycophancy Fine-Tuning: How Reward Hacking Breaks Uncertainty Quantification in LLMs
Sycophantic GRPO fine-tuning degrades LLM calibration, raising ECE by 0.006 and MCE by 0.010, with a persistent residual after post-hoc scaling.
-
Improving Semantic Uncertainty Quantification in Language Model Question-Answering via Token-Level Temperature Scaling
Optimizing a single scalar temperature improves semantic calibration, discrimination, and entropy in language model question-answering over heuristic baselines and token-level recalibration methods.