Stars
The simplest, fastest repository for training/finetuning small-sized VLMs.
Fully open data curation for reasoning models
Open-source observability for your GenAI or LLM application, based on OpenTelemetry
A minimal GPU design in Verilog to learn how GPUs work from the ground up
AI Agent that handles engineering tasks end-to-end: integrates with developers’ tools, plans, executes, and iterates until it achieves a successful result.
MMaDA - Open-Sourced Multimodal Large Diffusion Language Models (dLLMs with block diffusion, mixed-CoT, unified RL)
Transformer vs Recurrent Neural Operator (RNO) on 1D unit cell problem with viscoelastic constitutive relation. Benchmarked on XPU.
tutorial for training a BPE tokenizer on multiple GPUs
Pocket-Sized Multimodal AI for content understanding and generation across multilingual texts, images, and 🔜 video, up to 5x faster than OpenAI CLIP and LLaVA 🖼️ & 🖋️
Attention Kernels for Symmetric Power Transformers
Pruna is a model optimization framework built for developers, enabling you to deliver faster, more efficient models with minimal overhead.
[NeurIPS 2025] Reinforcement Learning for Reasoning in Large Language Models with One Training Example
Framework for creating high fidelity and complex RL environments and evaluation tasks
Qwen3 is the large language model series developed by Qwen team, Alibaba Cloud.
Code for "Scaling Language-Free Visual Representation Learning" paper (Web-SSL).
[NeurIPS 2025] TTRL: Test-Time Reinforcement Learning
A TTS model capable of generating ultra-realistic dialogue in one pass.
Take-home interview questions for Poly. E-mail us at work@withpoly.com with a link to your repo!
Preswald is a WASM packager for Python-based interactive data apps: bundle full complex data workflows, particularly visualizations, into single files, runnable completely in-browser, using Pyodide…
The API to search, scrape, and interact with the web at scale. 🔥
🌐 Make websites accessible for AI agents. Automate tasks online with ease.