[NeurIPS 2025 Spotlight] A Native Multimodal LLM for 3D Generation and Understanding

Load MDX content from anywhere

An efficient implementation of the NSA (Native Sparse Attention) kernel

Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3, Qwen3-MoE, DeepSeek-R1, GLM4.5, InternLM3, Llama4, ...) and 300+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, GLM4.5v, Llava, …

🚀 Efficient implementations of state-of-the-art linear attention models

beep boop personal website hosted at tinabmai.com

Repo for "LoLCATs: On Low-Rank Linearizing of Large Language Models"

🐳 Efficient Triton implementations for "Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention"

Residual Quantization Autoencoder, used for interpreting LLMs

repo for code for paper on general theory associative memory models

Recipes to scale inference-time compute of open models

Curated list of datasets and tools for post-training.

Run Slurm in Kubernetes

A toolkit for describing model features and intervening on those features to steer behavior.

Recreating and refactoring weareninja.com's "space warp" effect

Entropy Based Sampling and Parallel CoT Decoding

Repository for Manifold Integrated Gradients: Riemannian Geometry for Feature Attribution

A library for making RepE control vectors

Go ahead and axolotl questions

Thorn in a HaizeStack test for evaluating long-context adversarial robustness.

The simplest, fastest repository for training/finetuning medium-sized GPTs.

Simple, powerful and flexible site generation framework with everything you love from Next.js.

Orchestrate zero-shot computer vision models

LLM training in simple, raw C/CUDA

LlamaIndex is the leading framework for building LLM-powered agents over your data.