Length-Controlled AlpacaEval: A Simple Way to Debias Automatic Evaluators Dubois et al.: https://
arxiv.org/abs/2404.04475 #ArtificialIntelligence #DeepLearning #MachineLearning
RESEARCH
-

Length-Controlled AlpacaEval: Debiasing Automatic AI Evaluators
By
–
-
Europe’s 2025 Digitalisation Report Released
By
–
Digitalisation in Europe – 2025 edition out now https://
ec.europa.eu/eurostat/produ
ct?code=WDN-20250429-1
…
#digitaleu #digital #europe #innovation #technology @DigitalEU @ArturHabant @elaniazito @BetaMoroney @Khulood_Almani @Timothy_Hughes @IanLJones98 @Fabriziobustama @CurieuxExplorer @Shi4Tech @enilev @AkwyZ -
Critique du classement LMArena : illusion des performances modèles
By
–
There's a new paper circulating looking in detail at LMArena leaderboard: "The Leaderboard Illusion" https://arxiv.org/abs/2504.20879 I first became a bit suspicious when at one point a while back, a Gemini model scored #1 way above the second best, but when I tried to switch for a few
-

Xiaomi launches MiMo-7B-RL AI model
By
–
Xiaomi is entering AI game with a new MiMo-7B-RL open source model!
-

DeepSeek Releases Prover-V2-671B Math Model
By
–
DeepSeek just uploaded a new model — not R2, but DeepSeek-Prover-V2-671B, a powerful math model built on DeepSeek-V3.
Link: https://
huggingface.co/deepseek-ai/De
epSeek-Prover-V2-671B/tree/main
… -
BAIR Faculty Member Elected to National Academy of Sciences
By
–
Congratulations to BAIR faculty member @AlisonGopnik who has been elected to the National Academy of Sciences (
@theNASciences ). -
RoboBee Achieves Safe Landings with Insect-Inspired Leg Design
By
–
Une nouvelle étude de @SciRobotics met en évidence comment une conception et une stratégie de contrôle des pattes inspirées des insectes permettent au RoboBee de @Harvard d'exécuter des atterrissages sûrs, contrôlés et précis lorsqu'il vole entre les feuilles d'une plante.… pic.twitter.com/eGbRF8INdr
— VISION IA (@vision_ia) 30 avril 2025Une nouvelle étude de @SciRobotics met en évidence comment une conception et une stratégie de contrôle des pattes inspirées des insectes permettent au RoboBee de @Harvard d'exécuter des atterrissages sûrs, contrôlés et précis lorsqu'il vole entre les feuilles d'une plante.
-
Negative Prompting and CoT Countermeasures in AI Models
By
–
(I have used “You are bad at X; please don’t” to some effect this last year but it had not occurred to me to sketch out suggested countermeasures, in part because have spent less time in CoT prompting experiments than I should have.)
-
Prompt Engineering Techniques for More Effective AI Model Responses
By
–
(That prompt has some interesting things going on incidentally and annotating it with “this technique might get models to be more effective than simply sending a tweet-sized expression of desired outcome to them” would be a useful thing to have on Internet somewhere.)
