May 2026

Reinforcement Learning with Verifiable Rewards (RLVR) has achieved great success in developing Large Language Models (LLMs) with chain-of-thought rollouts for many tasks such as math and coding. Nevertheless, RLVR struggles with sample efficiency on difficult problems where correct rollouts are hard to generate. Prior works propose to address this issue via demonstration-guided RLVR, i.e., to conduct Supervised FineTuning (SFT) when RL fails; however, SFT often requires a lot of data, which can be expensive to acquire. In this […]

Ver mais

Gromov–Wasserstein (GW) distances compare graphs, shapes, and point clouds through internal distances, without requiring a common coordinate system. This invariance is powerful, but discrete GW is a nonconvex quadratic optimal transport problem and is difficult to estimate at scale. We propose emph{Distance-Matrix Wasserstein} (DMW), a hierarchy of Wasserstein statistics comparing laws of random finite distance matrices. Rather than optimizing a global point-level alignment, DMW samples $n$ points from each space, records their pairwise distances, and transports the resulting […]

Ver mais

well known benchmarks(dm-control, og-bench, humanoid-bench, etc) are based on cpu-simulator, and they are extremely slow. for publish paper with novel rl-algorithm, we need to use multiple seeds(at least 5) for each benchmarks, and we have to also do some ablations. I think it is too long to test the hyperparameter tuning and conduct ablation tests for cpu-based simulator benchmarks. But, recent GPU-based simulator benchmarks(mujoco-mjx, isaac gym, isaac lab, mujoco-playground) makes all training so fast. These alternatives are good […]

Ver mais

Large language models (LLMs) have become integral to enterprise applications across industries. Under the hood, customers’ inputs to the models are usually augmented with prompts that encode intricate business logic, regulatory requirements, and domain expertise: a healthcare system must use language compliant with the Health Insurance Portability and Accountability Act, for instance, and a financial trading system must follow risk tolerance rules. These prompts are typically crafted by domain experts over weeks or months. Yet business demands continue […]

Ver mais

In this quiz, you’ll test your understanding of Python Metaclasses. Metaclasses sit behind every class you write in Python, and they’re one of the language’s deeper object-oriented concepts. By working through this quiz, you’ll revisit how classes are themselves objects, how type creates them, and how a custom metaclass lets you customize class creation. You’ll also reflect on when a custom metaclass is actually the right tool and when a simpler technique does the job better. [ Improve […]

Ver mais

In this quiz, you’ll test your understanding of Cursor vs Windsurf: Which AI Code Editor Is Best for Python? By working through these questions, you’ll revisit how the two editors differ across code completion, agentic multi-file editing, and debugging. You’ll also reconnect with the audit points worth applying whenever an AI agent writes Python on your behalf. [ Improve Your Python With 🐍 Python Tricks 💌 – Get a short & sweet Python Trick delivered to your inbox […]

Ver mais

I keep hearing some version of this: “A paper that got accepted years ago wouldn’t stand a chance today.” Honestly, for a lot of ML subfields, this doesn’t sound crazy anymore. A paper that once looked solid can now look under-evaluated, under-ablated, weak on baselines, or just too obvious. So maybe the real claim is: A mediocre accepted ML paper from years ago would probably get rejected today. Do people agree? Has the bar actually gone up, or […]

Ver mais