[KongchangAI]
ModelMagistral

Magistral

A language model used as one of three models in the evaluation of U-Lens for uncertainty estimation benchmarking.

Timeline (last 90 days)

Oct 12

In evaluations across 3 seeds, 3 models, and 4 benchmarks (involving Gemma 4, Qwen3.5, and Magistral), U-Lens is the best estimator, ahead of 5-round sampling, 10-round sampling, and supervised baselines.

Unverified50%

All Facts (1)

Unverified

In evaluations across 3 seeds, 3 models, and 4 benchmarks (involving Gemma 4, Qwen3.5, and Magistral), U-Lens is the best estimator, ahead of 5-round sampling, 10-round sampling, and supervised baselines.

50%

Source Articles

U-Space Reveals When Language Models Become UncertainOct 12