#model calibration
9 papers · 1 filter
QuantWAMs: Calibrating at the Right Granularity for World Action Models
Jiacheng Zhou, Jinfan Lv, Ruixuan Li +4
The paper proposes QuantWAMs, a post‑training quantization framework that tailors quantization decisions to the structure, rollout distribution, and task objectives of World Action…
Calibrate Before Reason: Robust Visual Token Reduction against Semantic Drift in VLMs
Jiasheng Li, Zhong Ji, Yan Zhang +1
The paper proposes CaRe, a training‑free method that calibrates compact visual representations before reasoning to keep semantic consistency when reducing visual tokens in large vi…
Objective-Aligned Direct Answer SFT for Robust Multi-Frame Medical VQA
Site Li, Jianyi Hao, Xiaofeng Liu
The paper evaluates different adaptation strategies for multi-frame medical visual question answering and finds that a simple objective-aligned direct answer supervised fine-tuning…
CalTwin: Towards Calibrated, Shift-Robust Medical World Models via Fisher-Information Regularisation
Behraj Khan, Shabir Ahmad, Syed Ahmad Chan Bukhari +1
The paper introduces CalTwin, a lightweight regularization that combines a Fisher‑information‑based shift penalty with a confidence‑misalignment penalty to make GRU‑based medical w…
Beyond scalar losses: calibrating segmentation models via gradient vector field surgery
Laurin Lux, Alexander H. Berger, Moritz Knolle +2
The paper introduces a gradient‑based modification to region‑based loss functions that scales the gradient magnitude with prediction error, improving calibration of medical image s…
When Top-K Misses the Decision: Tool-Call Drift in Multi-Teacher On-Policy Distillation
Jiabin Shen, Guang Chen, Chengjun Mao
The paper studies how multi-teacher on-policy distillation can cause language models to over-call tools, and introduces Soft Clamp, a token-level divergence calibration method that…