List your company

Faithful Calibration on Faithful Calibration Dataset Suite (test)

85PQARLMF
48.658.0567.576.95Jun 30, 2026

Evaluation results

MethodLinks
RLMFBackbone=Llama3.1-8B-Ins, Method=Reinforcement Learning with Metacognitive Feedback (Numerical)2026.068581838281848483868684410.26
RLMFBackbone=Qwen3-8B, Method=Reinforcement Learning with Metacognitive Feedback (Numerical)2026.068582868284828382838483570.19
+RLBackbone=Llama3.1-8B-Ins, Method=RL Ablation (No Metacognitive Scaling)2026.068278807975738180737277400.2
RLMF + Rewr.Backbone=Qwen3-8B, Method=Reinforcement Learning with Metacognitive Feedback (Linguistic Rewriting)2026.068286808480808787828283570.19
RLMF + Rewr.Backbone=Llama3.1-8B-Ins, Method=Reinforcement Learning with Metacognitive Feedback (Linguistic Rewriting)2026.068186808180818281878382410.26
+RLBackbone=Qwen3-8B, Method=RL Ablation (No Metacognitive Scaling)2026.067566692055545844323851590.26
FUTBackbone=Llama3.1-8B-Ins, Method=Faithful Uncertainty Tuning2026.066967686663706363686766310.29
MetaFaithBackbone=Llama3.1-8B-Ins, Method=Metacognitive Prompting2026.066871656767646466687267280.36
Gemini-3.1-ProModel Type=Frontier Model2026.066271706872686671738270780.15
Llama3.1-8B-InsBackbone=Llama3.1-8B-Ins, Configuration=Base Model2026.066061615065624861597160310.33
Gemini-3-FlashModel Type=Frontier Model2026.065964556667706566777166720.16
FUTBackbone=Qwen3-8B, Method=Faithful Uncertainty Tuning2026.065775487472716667717467380.41
Qwen3-8BBackbone=Qwen3-8B, Configuration=Base Model2026.06536357546359595976254550.31
MetaFaithBackbone=Qwen3-8B, Method=Metacognitive Prompting2026.065366476867727049706763510.29
GPT-5Model Type=Frontier Model, Prompting=MetaFaith prompting2026.065061526659576057687761690.19