LLMs still struggle with deep reasoning LUFFY is a new framework bridging imitation and exploration by injecting off-policy guidance (like DeepSeek-R1) into zero-RL Boosts math reasoning by 7 points Improves weaker base models like Llama and Qwen Trending on alphaXiv
MACHINE LEARNING
-
Gemini 2.5 Pro tops leaderboard but faces trust and ‘woke’ issues
By
–
People are either sleeping on Gemini – or they just don't trust Google. Google's Gemini-2.5 Pro still tops the Chatbot Arena Leaderboard being the highest quality LLM available as of this morning – and yet no one talks about it. Is that because Google's 'woke' image generator
-
Yann LeCun Explains Why AGI Is Not Possible With LLMs
By
–
Apprendre comment fonctionne le monde est difficile.
— VISION IA (@vision_ia) 28 avril 2025
Yann LeCun explique (traduit en 🇫🇷) pourquoi l'AGI n'est pas possible avec les LLM. pic.twitter.com/OF76FUxOWlApprendre comment fonctionne le monde est difficile. Yann LeCun explique (traduit en ) pourquoi l'AGI n'est pas possible avec les LLM.
-
Panel Discussion on World Models at ICLR 2025
By
–
Had a really fun panel discussion with @jeffclune and other heroic minds at the World Models workshop at #ICLR2025
-
Moonshot AI Launches Kimi-Audio Open-Source Universal Audio Model
By
–
Moonshot AI launched Kimi-Audio, a new open-source universal audio model —Supports speech recognition, audio-to-text, and speech-to-speech
—Pre-trained on >13M hrs of audio data
—Exels in 10+ audio benchmarks, including MMAU, VoiceBench, and LibriSpeech -
Liquid Sciences Launches Hyena Edge: Efficient Hybrid AI Architecture
By
–
Liquid Sciences dropped Hyena Edge, a hybrid AI with a ‘convolution’ architecture It delivers faster processing and improved benchmarks, beating Transformer-based baselines in computational efficiency and model quality on edge hardware
-
Orchestrating Synthetic Data With Reasoning in AI
By
–
ORCHESTRATING SYNTHETIC DATA WITH REASONING Davidson et al.: https://
openreview.net/pdf?id=VOoeogZ
bMb
… #ArtificialIntelligence #DeepLearning #MachineLearning -
Databricks ML Migration: Optimization and Model Lifecycle Tips
By
–
Here are some great tips from the #DatabricksCommunity on migrating ML workloads to Databricks. Explore optimization techniques, cluster management, and model lifecycle management using Unity Catalog: https://
dbricks.co/4cKAgSd -
Subscribe to Weekly AI Engineering Tutorials on ML and LLMs
By
–
If you're interested in ML, LLMs, RAG, and AI Agents and want to receive tutorials every week, subscribe to AI Engineering (for free):
-
Scaling Laws in LLMs: Beyond Classic Models and Inference Costs
By
–
Great set of slides by my Gemini colleague @FeinbergVlad on scaling considerations in large language models, addressing the fact that the classic "scaling law" work does not take into account inference cost (!), distillation, learning rate schedules, etc. https://
vladfeinberg.com/2025/04/24/gem
ini-flash-pretraining.html
…