Models

A pair of Agent Skills packages that teach coding agents about building with Google's Gemma models and about training or fine-tuning them locally, installable through the Vercel or Context7 skills CLIs.

• Agent skills
• Models

onnx2tf converts ONNX model files to LiteRT, TensorFlow Lite, TensorFlow, native PyTorch modules, TorchScript, PyTorch state dictionaries and exported PyTorch programs, and can also convert directly from LiteRT back to PyTorch.

• Models
• CLI tools

Nemo Skills is NVIDIA's collection of pipelines for improving large language models: synthetic data generation, model training, and evaluation across benchmarks in math, code, science and instruction following, scaling from a workstation to a Slurm cluster.

• Models
• AI evals

Optimum Quanto is a PyTorch quantization backend for Hugging Face's optimum library, now in maintenance mode, that inserts quantization and dequantization automatically and supports int2, int4, int8 and float8 weights and activations across CPU, CUDA and MPS devices.

• Models
• Libraries

rizzo-pii is a small, CPU-friendly model that detects 22 categories of personal data in Italian legal text, including Italian tax and property identifiers, and drives a local, reversible anonymize-then-restore workflow so sensitive documents can still be sent to a cloud LLM.

• Models
• Security

NeoHorse is a family of open-weight language models sized for agent workflows, with 4B and 9B versions tuned for tool use, coding and instruction following, plus a companion decision model that uses prefill-only inference to pick an action or score a condition.

• Models

FINN is an experimental open-source compiler from AMD Research that generates dataflow-style FPGA accelerators for quantized neural networks, aiming for high inference throughput and low latency.

• Models

skrl is an open-source, modular reinforcement-learning library implemented in PyTorch, JAX and NVIDIA Warp, designed for readable, transparent algorithms, with support for Gymnasium, Gym, PettingZoo, Isaac Lab and MuJoCo Playground environments.

• Models

YOLOs-CPP is a C++ inference library that gives every YOLO model version from v5 through v12 and YOLO26 one unified API for detection, segmentation, pose estimation, oriented bounding boxes, classification and depth estimation, built on ONNX Runtime and OpenCV.

• Models
• Inference

VeRL-Omni is a reinforcement-learning training framework for diffusion, image, video, audio and other multimodal generative models, built as a dedicated extension of the verl RL library.

• Models
• Generative media

A 2-bit quantized language model, and the training pipeline for it, small enough to run as a 40KB CP/M binary on a Z80 processor with 64KB of RAM.

• Models

A high-performance operator library for large language models, implemented in the Triton language, that registers with PyTorch's ATen backend to accelerate training and inference across many hardware platforms.

• Models
• Inference

Open research code exploring the Engram memory-injection architecture as an alternative to LoRA for fine-tuning language and vision models, extended from Qwen to Stable Diffusion.

• Models

A research system and reference implementation of the VINDEX3 format for executing, representing and studying a machine learning model as an inspectable, queryable artifact.

• Models

A collection of machine learning models optimized and ready to deploy on Qualcomm devices, with a CLI to browse, compile, quantize, profile and run them on real hardware.

• Models

A Python framework of post-training and training-time compression algorithms, such as quantization and pruning, for optimizing neural network inference in OpenVINO with minimal accuracy loss.

• Models
• Inference

A Python framework for building, iterating on and deploying machine learning workloads on Kubernetes, bringing cluster compute into a local development workflow with fast iteration.

• Models
• Containers

An extensible LLM quantization toolkit supporting GPTQ, AWQ, GGUF, FP8, and other formats across NVIDIA, AMD, Huawei Ascend, and Intel hardware, integrated with Transformers, vLLM, and SGLang.

• Models
• Inference

An autonomous agent that runs deep-learning experiments around the clock on a local machine, a remote server over SSH, or a Slurm cluster, tracking real job outcomes with constant-size memory and zero-cost monitoring.

• Agent frameworks
• Models

An end-to-end platform for building computer vision models, from dataset preparation and training through optimization and deployment, tuned for Intel's CPU, GPU, and other accelerator hardware.

• Models

A research framework covering the full lifecycle of GUI agents: online reinforcement-learning training, standardized evaluation across benchmarks, and deployment to control real mobile devices by natural language.

• Agent frameworks
• Models

An end-to-end autonomous research agent that carries a research idea through literature search, experiment design, real GPU training, result analysis, and a complete LaTeX paper with figures drawn from actual experiment data.

• Agent frameworks
• Models

An Unsloth-compatible fine-tuning library, for large language, vision, speech and OCR models, that runs natively on Apple Silicon through Apple's MLX framework.

• Models

A family of autoregressive vision models from Apple's AI/ML research team, with released code and pre-trained checkpoints for AIMv1 and AIMv2, aimed at large-scale visual and multimodal representation learning.

• Models