List your company

Medical Visual Question Answering on SLAKE Closed 1.0 (test)

0ECEPlatt Scaling
-0.015080.086710.18850.29029Apr 2, 2026

Evaluation results

MethodLinks
Platt ScalingBackbone=Qwen3-VL-32B, Confidence Source=Sampling, 5-fold CV=true2026.040
Isotonic Regr.Backbone=Qwen3-VL-32B, Confidence Source=Sampling, 5-fold CV=true2026.040
Hist. BinningBackbone=Qwen3-VL-32B, Confidence Source=Sampling, 5-fold CV=true2026.040
Hist. BinningBackbone=Qwen3-VL-2B, Confidence Source=Verbalized, 5-fold CV=true2026.040.008
Hist. BinningBackbone=InternVL3-2B, Confidence Source=Verbalized, 5-fold CV=true2026.040.012
HAC-GateBackbone=InternVL3-8B, Confidence Source=Verbalized, 5-fold CV=true2026.040.013
Isotonic Regr.Backbone=Qwen3-VL-2B, Confidence Source=Verbalized, 5-fold CV=true2026.040.018
Isotonic Regr.Backbone=Qwen3-VL-32B, Confidence Source=Verbalized, 5-fold CV=true2026.040.018
Platt ScalingBackbone=Qwen3-VL-2B, Confidence Source=Verbalized, 5-fold CV=true2026.040.019
HAC-PlattBackbone=Qwen3-VL-2B, Confidence Source=Verbalized, 5-fold CV=true2026.040.022
Platt ScalingBackbone=LLaVA-NeXT-7B, Confidence Source=Verbalized, 5-fold CV=true2026.040.023
HAC-PlattBackbone=LLaVA-NeXT-7B, Confidence Source=Verbalized, 5-fold CV=true2026.040.023
Isotonic Regr.Backbone=InternVL3-8B, Confidence Source=Verbalized, 5-fold CV=true2026.040.024
Hist. BinningBackbone=Qwen3-VL-32B, Confidence Source=Verbalized, 5-fold CV=true2026.040.03
Hist. BinningBackbone=InternVL3-8B, Confidence Source=Verbalized, 5-fold CV=true2026.040.03
Isotonic Regr.Backbone=Qwen3-VL-8B, Confidence Source=Sampling, 5-fold CV=true2026.040.031
Isotonic Regr.Backbone=Qwen3-VL-8B, Confidence Source=Verbalized, 5-fold CV=true2026.040.031
Platt ScalingBackbone=Qwen3-VL-8B, Confidence Source=Sampling, 5-fold CV=true2026.040.032
Hist. BinningBackbone=Qwen3-VL-8B, Confidence Source=Sampling, 5-fold CV=true2026.040.033
Isotonic Regr.Backbone=InternVL3-2B, Confidence Source=Verbalized, 5-fold CV=true2026.040.033
HAC-GateBackbone=Qwen3-VL-32B, Confidence Source=Verbalized, 5-fold CV=true2026.040.034
Hist. BinningBackbone=InternVL3-38B, Confidence Source=Verbalized, 5-fold CV=true2026.040.037
Isotonic Regr.Backbone=LLaVA-NeXT-34B, Confidence Source=Verbalized, 5-fold CV=true2026.040.037
Platt ScalingBackbone=InternVL3-8B, Confidence Source=Verbalized, 5-fold CV=true2026.040.038
Isotonic Regr.Backbone=Qwen3-VL-2B, Confidence Source=Sampling, 5-fold CV=true2026.040.04
Hist. BinningBackbone=LLaVA-NeXT-34B, Confidence Source=Verbalized, 5-fold CV=true2026.040.04
Hist. BinningBackbone=Qwen3-VL-8B, Confidence Source=Verbalized, 5-fold CV=true2026.040.041
HAC-GateBackbone=Qwen3-VL-2B, Confidence Source=Verbalized, 5-fold CV=true2026.040.042
Hist. BinningBackbone=Qwen3-VL-2B, Confidence Source=Sampling, 5-fold CV=true2026.040.044
Isotonic Regr.Backbone=InternVL3-38B, Confidence Source=Verbalized, 5-fold CV=true2026.040.046
Isotonic Regr.Backbone=LLaVA-NeXT-34B, Confidence Source=Sampling, 5-fold CV=true2026.040.046
Platt ScalingBackbone=Qwen3-VL-2B, Confidence Source=Sampling, 5-fold CV=true2026.040.047
HAC-PlattBackbone=LLaVA-NeXT-34B, Confidence Source=Sampling, 5-fold CV=true2026.040.051
Hist. BinningBackbone=LLaVA-NeXT-34B, Confidence Source=Sampling, 5-fold CV=true2026.040.052
HAC-PlattBackbone=InternVL3-8B, Confidence Source=Verbalized, 5-fold CV=true2026.040.053
HAC-PlattBackbone=InternVL3-38B, Confidence Source=Sampling, 5-fold CV=true2026.040.055
Isotonic Regr.Backbone=InternVL3-38B, Confidence Source=Sampling, 5-fold CV=true2026.040.058
Platt ScalingBackbone=LLaVA-NeXT-34B, Confidence Source=Sampling, 5-fold CV=true2026.040.059
Hist. BinningBackbone=InternVL3-38B, Confidence Source=Sampling, 5-fold CV=true2026.040.061
Platt ScalingBackbone=LLaVA-NeXT-34B, Confidence Source=Verbalized, 5-fold CV=true2026.040.063
HAC-PlattBackbone=LLaVA-NeXT-34B, Confidence Source=Verbalized, 5-fold CV=true2026.040.063
HAC-GateBackbone=InternVL3-2B, Confidence Source=Verbalized, 5-fold CV=true2026.040.064
Platt ScalingBackbone=InternVL3-2B, Confidence Source=Verbalized, 5-fold CV=true2026.040.067
HAC-PlattBackbone=Qwen3-VL-2B, Confidence Source=Sampling, 5-fold CV=true2026.040.068
Platt ScalingBackbone=InternVL3-38B, Confidence Source=Sampling, 5-fold CV=true2026.040.068
HAC-GateBackbone=Qwen3-VL-8B, Confidence Source=Verbalized, 5-fold CV=true2026.040.069
Platt ScalingBackbone=Qwen3-VL-8B, Confidence Source=Verbalized, 5-fold CV=true2026.040.07
HAC-PlattBackbone=Qwen3-VL-8B, Confidence Source=Verbalized, 5-fold CV=true2026.040.07
Platt ScalingBackbone=Qwen3-VL-32B, Confidence Source=Verbalized, 5-fold CV=true2026.040.07
HAC-GateBackbone=InternVL3-38B, Confidence Source=Verbalized, 5-fold CV=true2026.040.07
Platt ScalingBackbone=InternVL3-38B, Confidence Source=Verbalized, 5-fold CV=true2026.040.071
HAC-PlattBackbone=Qwen3-VL-32B, Confidence Source=Verbalized, 5-fold CV=true2026.040.072
Platt ScalingBackbone=LLaVA-NeXT-7B, Confidence Source=Sampling, 5-fold CV=true2026.040.072
Hist. BinningBackbone=LLaVA-NeXT-7B, Confidence Source=Sampling, 5-fold CV=true2026.040.073
HAC-GateBackbone=Qwen3-VL-2B, Confidence Source=Sampling, 5-fold CV=true2026.040.077
HAC-GateBackbone=InternVL3-38B, Confidence Source=Sampling, 5-fold CV=true2026.040.077
HAC-PlattBackbone=LLaVA-NeXT-7B, Confidence Source=Sampling, 5-fold CV=true2026.040.078
Hist. BinningBackbone=InternVL3-8B, Confidence Source=Sampling, 5-fold CV=true2026.040.079
Hist. BinningBackbone=InternVL3-2B, Confidence Source=Sampling, 5-fold CV=true2026.040.08
Hist. BinningBackbone=LLaVA-NeXT-7B, Confidence Source=Verbalized, 5-fold CV=true2026.040.082
Isotonic Regr.Backbone=InternVL3-8B, Confidence Source=Sampling, 5-fold CV=true2026.040.083
HAC-GateBackbone=LLaVA-NeXT-34B, Confidence Source=Sampling, 5-fold CV=true2026.040.083
Isotonic Regr.Backbone=LLaVA-NeXT-7B, Confidence Source=Sampling, 5-fold CV=true2026.040.084
Isotonic Regr.Backbone=LLaVA-NeXT-7B, Confidence Source=Verbalized, 5-fold CV=true2026.040.085
HAC-PlattBackbone=Qwen3-VL-32B, Confidence Source=Sampling, 5-fold CV=true2026.040.086
HAC-GateBackbone=Qwen3-VL-32B, Confidence Source=Sampling, 5-fold CV=true2026.040.086
Platt ScalingBackbone=InternVL3-8B, Confidence Source=Sampling, 5-fold CV=true2026.040.087
HAC-PlattBackbone=InternVL3-8B, Confidence Source=Sampling, 5-fold CV=true2026.040.089
HAC-PlattBackbone=InternVL3-38B, Confidence Source=Verbalized, 5-fold CV=true2026.040.089
HAC-GateBackbone=LLaVA-NeXT-34B, Confidence Source=Verbalized, 5-fold CV=true2026.040.09
HAC-GateBackbone=Qwen3-VL-8B, Confidence Source=Sampling, 5-fold CV=true2026.040.091
HAC-GateBackbone=InternVL3-8B, Confidence Source=Sampling, 5-fold CV=true2026.040.092
HAC-PlattBackbone=Qwen3-VL-8B, Confidence Source=Sampling, 5-fold CV=true2026.040.094
Isotonic Regr.Backbone=InternVL3-2B, Confidence Source=Sampling, 5-fold CV=true2026.040.1
HAC-PlattBackbone=InternVL3-2B, Confidence Source=Verbalized, 5-fold CV=true2026.040.101
UncalibratedBackbone=InternVL3-38B, Confidence Source=Sampling, 5-fold CV=true2026.040.107
HAC-GateBackbone=LLaVA-NeXT-7B, Confidence Source=Sampling, 5-fold CV=true2026.040.117
Platt ScalingBackbone=InternVL3-2B, Confidence Source=Sampling, 5-fold CV=true2026.040.123
UncalibratedBackbone=LLaVA-NeXT-7B, Confidence Source=Sampling, 5-fold CV=true2026.040.132
HAC-GateBackbone=InternVL3-2B, Confidence Source=Sampling, 5-fold CV=true2026.040.137
HAC-GateBackbone=LLaVA-NeXT-7B, Confidence Source=Verbalized, 5-fold CV=true2026.040.138
UncalibratedBackbone=Qwen3-VL-32B, Confidence Source=Verbalized, 5-fold CV=true2026.040.143
UncalibratedBackbone=Qwen3-VL-2B, Confidence Source=Verbalized, 5-fold CV=true2026.040.15
UncalibratedBackbone=InternVL3-38B, Confidence Source=Verbalized, 5-fold CV=true2026.040.154
HAC-PlattBackbone=InternVL3-2B, Confidence Source=Sampling, 5-fold CV=true2026.040.156
UncalibratedBackbone=InternVL3-8B, Confidence Source=Sampling, 5-fold CV=true2026.040.158
UncalibratedBackbone=InternVL3-8B, Confidence Source=Verbalized, 5-fold CV=true2026.040.165
UncalibratedBackbone=InternVL3-2B, Confidence Source=Verbalized, 5-fold CV=true2026.040.17
UncalibratedBackbone=Qwen3-VL-8B, Confidence Source=Verbalized, 5-fold CV=true2026.040.171
UncalibratedBackbone=InternVL3-2B, Confidence Source=Sampling, 5-fold CV=true2026.040.175
UncalibratedBackbone=Qwen3-VL-32B, Confidence Source=Sampling, 5-fold CV=true2026.040.197
UncalibratedBackbone=Qwen3-VL-2B, Confidence Source=Sampling, 5-fold CV=true2026.040.199
UncalibratedBackbone=LLaVA-NeXT-34B, Confidence Source=Sampling, 5-fold CV=true2026.040.202
UncalibratedBackbone=Qwen3-VL-8B, Confidence Source=Sampling, 5-fold CV=true2026.040.206
UncalibratedBackbone=LLaVA-NeXT-34B, Confidence Source=Verbalized, 5-fold CV=true2026.040.234
UncalibratedBackbone=LLaVA-NeXT-7B, Confidence Source=Verbalized, 5-fold CV=true2026.040.377