Could AI models start self-improving recursively, scaling up performance to the moon in an infinite loop? A friend said that it's impossible, I wasn't that sure. I argued that with agentic methods like Absolute Zero (basically an LLM generates training data by itself by
LLMS
-
Position Bias in LLMs: How Architecture and Training Data Matter
By
–
LLMs sometimes exhibit position bias, where they overemphasize the beginning or end of a document or conversation while neglecting the middle. A theoretical framework from MIT revealed that model architectures & training data contribute to this problem: http://
bit.ly/44mW9V8 -
US Court Rules LLM Training on Copyrighted Books Fair Use
By
–
On Monday, a United States District Court ruled that training LLMs on copyrighted books constitutes fair use. A number of authors had filed suit against Anthropic for training its models on their books without permission. Just as we allow people to read books and learn from them
-
ChatGPT cognitive effects confirmed by MIT research study
By
–
ChatGPT nous abrutit ? Ça se confirme J’en parlais dans Mes Flashs de la Semaine #28 à la suite d’une étude choc signée Microsoft–Carnegie Mellon. Une nouvelle recherche du MIT confirme la tendance, avec une mise en garde encore plus claire. Ceuxqui ont utilisé ChatGPT ont
-
LLMs Applied to Investigation Training at Europol
By
–
Last week, I had the privilege of spending three intense days in Paris giving our training to Europol’s Ops-Tech LLM core group: an “LLMs Applied to Investigation” boot camp. Here’s what we squeezed into 15 hours (+ many more in 1:1 discussions): – Foundational knowledge – why
-
Kimi-Dev-72B shows impressive SWE-Bench Verified performance, no tech report yet
By
–
Why isn't anyone talking about Kimi-Dev-72B, released ~1 week ago on the Hub?
The tech report is not out yet, and I'll love to see more benchmarks, but performance on SWE-Bench Verified seems very impressive! -
Transformers Prefer Simpler Hypotheses In-Context Occam’s Razor
By
–
In-Context Occam's Razor: How Transformers Prefer Simpler Hypotheses on the Fly Deora et al.: https://
arxiv.org/abs/2506.19351 #ArtificialIntelligence #DeepLearning #MachineLearning -
OpenAI to Soon Release Open Source GPT Model
By
–
Breaking News : J’ai eu la confirmation secrète qu’OpenAI va bientôt publier son modèle open source ! Ça annonce un gros bouleversement dans le paysage de l’IA ! Parce qu’on va pouvoir faire du développement local et totalement sécurisé avec un modèle GPT !! À voir si ce sera
-
KerasHub enables Hugging Face checkpoints across multiple frameworks
By
–
KerasHub lets you use any Hugging Face checkpoint for all top models like Llama, Gemma, Mistral, etc… Run your workflows in JAX, PyTorch, TensorFlow – inference, LoRA fine-tuning, large-scale training from scratch Blog post: https://
developers.googleblog.com/en/load-model-
weights-from-safetensors-into-kerashub-multi-framework-machine-learning/
… -
OpenRouter AI Secures $40M Funding for Multi-Model LLM Platform
By
–
OpenRouter AI announced $40M seed + series A funding from led 16z & Menlo Ventures
— Rowan Cheung (@rowancheung) 26 juin 2025
The company has built a unified interface for LLMs, offering 450+ models with easy switching and no lock-in
1M+ developers use their APIpic.twitter.com/51uR8o7J2OOpenRouter AI announced $40M seed + series A funding from led 16z & Menlo Ventures The company has built a unified interface for LLMs, offering 450+ models with easy switching and no lock-in 1M+ developers use their API