🤔 How can we systematically enhance LLMs for complex medical coding tasks? 🚀 Introducing MedAgentGym, an interactive gym-style platform designed specifically for training LLM agents in coding-based medical reasoning! 🧬💻 🎯 Comprehensive Code-based Medical Reasoning

New Google paper trains LLM judges to use small bits of code alongside reasoning, so their decisions become precise. So judging stops being guesswork and becomes checkable. Text only judges often miscount, miss structure rules, or accept shaky logic that a simple program would

We just built and released the largest dataset for supervised fine-tuning of agentic LMs, 1.27M trajectories (~36B tokens)! Up until now, large-scale SFT for agents is rare - not for lack of data, but because of fragmentation across heterogeneous formats, tools, and interfaces.

🚨 Eigen-1 gets 48.3% (Pass@1) & 61.74% (Pass@5) on "Humanity's Last Exam" (HLE) gold subset @FutureHouseSF using DeepSeek V3.1. Prev. Grok4->30.2%, GPT-5->22.8%, Gemini 2.5 Pro->18.8% 📎 https://t.co/4Fhcp8VTBG The future isn't bigger models, it's smarter agentic design! 🚀

With deep research revolutionizing research/data analysis, why are we still stuck in manually crafting data viz? Meet CoDA ( https://t.co/g8yjnHiMHM): The ultimate multi-agent LLM powerhouse for auto-generating stunning plots from NL queries! Handles complex data, self-refines

New @GoogleResearch paper builds a personal health assistant that reads a user’s data, answers health questions, and coaches daily habits. It evaluates the system on 10 tasks with 7,000+ human annotations and 1,100 hours from experts and users. The assistant covers 4 needs,

RLAD (Reinforcement Learning with Abstraction and Deduction) trains models via RL using a 2-player setup: ▪️ An abstraction generator – proposes short, natural-language “reasoning hints” (abstractions) summarizing key facts and strategies. ▪️ A solution generator – uses them to

🚀 Tired of choosing between speed and accuracy for Diffusion Large Language Models? Meet FreeDave ( https://t.co/KzdsX3TrnX) — the lossless parallel decoding algorithm that fixes DLLMs’ inference pain points perfectly! No extra draft models, no model tweaks — just smart parallel

🚀 Introducing BroRL: Scaling Reinforcement Learning via Broadened Exploration When step-scaling hits a plateau, scale rollouts, not steps. BroRL takes reinforcement learning beyond saturation—reviving stalled models by expanding exploration with large-N rollouts. 👇 (1/n)

🚫 This @Microsoft paper brings really bad news for medical AI models. Exposes some serious flaws. AI models just aren’t ready yet for reliable medical reasoning. 🤯 Paper finds that medical AI model pass tests by exploiting patterns in the data, not by actually combining

Cool research paper from Google. This is what clever context engineering looks like. It proposes Tool-Use-Mixture (TUMIX), leveraging diverse tool-use strategies to improve reasoning. This work shows how to get better reasoning from LLMs by running a bunch of diverse agents

The paper teaches small LLMs to reason better by training with built in tree search. i.e. Smarter exploration beats longer training runs. It reaches 62.95% average accuracy while using 5.7x fewer GPU hours. Typical reinforcement learning with verifiable rewards stalls because

🚨 Happy to share AceSearcher accepted to #NeurIPS2025 #Spotlight! 🔹 One LLM, two roles: Decomposer (split queries) + Solver (combine context) 🔹 +7.6% on QA & fact verification 🔹 32B ≈ DeepSeek-V3 on DocMath 📂 Code: https://t.co/lQU12Dm7vb 📑 arXiv: https://t.co/JI0kOh0yDk

ReasoningBank: memory for self-evolving LLM agents • Distills strategies from both successes & failures • Enables agents to learn, reuse, and improve over time • Outperforms prior memory methods on web & SWE tasks (+34.2% eff., –16% steps)

Baichuan-M2: Scaling Medical Capability with Large Verifier System Baichuan has released what is probably right now the best open-source LLM for medicine! Second only to GPT-5! "Despite its relatively small number of parameters (only 32B), Baichuan-M2 outperformed all other

🎉Our paper "Interpretation Meets Safety: A Survey on Interpretation Methods and Tools for Improving LLM Safety" has been accepted to EMNLP 2025 Main Track! @emnlpmeeting 👉First survey connecting LLM interpretation & safety

✨ Alongside NVIDIA-Nemotron-Nano-v2-9B, we’re also open-sourcing its pre-training dataset. At NVIDIA, we remain committed to openness — models + datasets. As the global open-source ecosystem rapidly evolves (with remarkable momentum emerging from Asia and beyond), we stand

👋 Welcome to CTCAE6 GO — fast CTCAE v6/v5 reference, built by a clinician. 🎓 Trainees always free 💡 Share how you're using Pro to learn, teach, or save time in clinic. Tag us or DM your best workflow or use case! =>longer access as CTCAE6 GO Elite User . Free PRO CODE below

Pass@k Training for Adaptively Balancing Exploration and Exploitation of Large Reasoning Models Reasoning RL using the pass@k metric. It promotes exploration and generalization.

🚀 DeepCode: Open Agentic Coding is Here! We dropped DeepCode - an AI-powered coding platform that transforms research papers and technical documents into production-ready code! 🔗 Fully Open Source: https://t.co/vBzRhcVAsN ✨ Current Features: • Paper2Code: Convert research