model

MedGemma 4B IT

modelactiveprovisionalmedgemma-4b-it-d2ae6dba·1 events·first seen 2d ago

Aliases: MedGemma 4B IT

Co-occurring entities

Just how sure are you? Improving Verbalized Uncertainty Calibration in Medical VQA Qwen2.5-7B-Instruct-1M

More like this (12)

Gemma-4 E4B-it MedGemma Gemma-3-4B-IT Gemma 4 T5Gemma Gemma 3n Gemma3-270M Gemma 3 Gemma 2 9B Gemma 3 270M Gemma Scope 2 Gemini-3 Pro

Recent events (1)

5arXiv · cs.CL·2d ago·source ↗

Training framework reduces calibration error 60%+ in Medical VQA multimodal LLMs

A new arXiv preprint proposes a finetuning framework to improve verbalized uncertainty calibration in multimodal LLMs applied to Medical Visual Question Answering. The composite loss function combines Brier-style calibration, anchor regularization, contrastive image-text alignment, and KL-based stabilization, evaluated on MedGemma 4B IT and Qwen2-VL 7B Instruct across three medical VQA benchmarks. The method reduces calibration error by 60% or more and improves discrimination by 26% or more while preserving predictive accuracy, outperforming prompting-, sampling-, and training-based baselines.

Evaluation and Benchmarking AI Safety Research Just how sure are you? Improving Verbalized Uncertainty Calibration in Medical VQA Qwen2.5-7B-Instruct-1M MedGemma 4B IT +1 more