Beyond Math and Code: How SpyRL Turns Party Games Into Verifiable Training Signals for LLMs
Reinforcement Learning with Verifiable Rewards (RLVR) has become the engine behind modern reasoning models. The recipe is straightforward: let a
Tag: Machine Learning
Metis: The First Memory Foundation Model That Learns to Remember
AI agents have gotten remarkably good at reasoning, perceiving, and acting. But ask one to remember what you told it
Continue readingMetis: The First Memory Foundation Model That Learns to Remember
Xiaomi-Robotics-1: When Scaling Laws Finally Arrive in Robotics
Robotics has a data problem. While language and vision models have ridden scaling laws to ever-higher capabilities, robot learning has
Continue readingXiaomi-Robotics-1: When Scaling Laws Finally Arrive in Robotics
Inkling: Thinking Machines Lab’s 975B Open-Weights Multimodal Model
The open-weights LLM landscape just gained a significant new entrant. Inkling, released on July 15 by Thinking Machines Lab, is
Continue readingInkling: Thinking Machines Lab’s 975B Open-Weights Multimodal Model
Training-Inference Mismatch: Why Your LLM Reinforcement Learning Is Optimizing the Wrong Policy
Reinforcement learning has become the defining ingredient of modern LLM post-training. GRPO, PPO, and their variants drive the reasoning capabilities
Program-as-Weights: Compiling Natural Language Into Local Neural Programs
There’s a class of programming tasks that resists clean implementation: deciding whether a log line is “important,” repairing malformed JSON
Continue readingProgram-as-Weights: Compiling Natural Language Into Local Neural Programs
Why Verification Is Harder Than Generation for AI Coding Agents
There’s a classical intuition in computer science that verifying a solution is easier than finding one. For NP-complete problems, this
Continue readingWhy Verification Is Harder Than Generation for AI Coding Agents
LoopCoder-v2: Why Two Loops Beat Four in Test-Time Compute Scaling
The dominant scaling narrative in large language models has been straightforward: more parameters, more data, more compute. But there’s a
Continue readingLoopCoder-v2: Why Two Loops Beat Four in Test-Time Compute Scaling
How MiniMax Sparse Attention Achieves 28x Compute Reduction at 1M Context Length
The attention mechanism is the backbone of every transformer model, but it carries a brutal cost: quadratic complexity with respect
Continue readingHow MiniMax Sparse Attention Achieves 28x Compute Reduction at 1M Context Length
SkillOpt: Training AI Agent Skills Like Neural Networks
AI agents have a skill problem. You give a language model a system prompt — or “skill” — and it
Continue readingSkillOpt: Training AI Agent Skills Like Neural Networks