Models

REAP is a research codebase for pruning experts from large sparse Mixture-of-Experts language models, using a criterion that combines router gate values and expert activation norms to compress models with less accuracy loss than merging.

• Models

A curated collection of tutorials, papers, tools and best practices for fine-tuning large language models such as GPT, BERT and RoBERTa.

• Models
• Curated lists

A from-scratch LLM stack that distills a large teacher model into a small student and serves it efficiently, with custom Triton and CUDA kernels, sharded training and paged-KV serving.

• Models
• Inference

CodeFuse Embeddings is an index of CodeFuse's text and code embedding research repositories, spanning several named embedding model projects and a benchmark.

• Models
• Curated lists

CodeRL is the official research code for a NeurIPS 2022 paper that trains a pretrained code-generation language model with deep reinforcement learning, using a critic network to judge functional correctness and guide a regeneration step at inference time.

• Models
• Papers

Open-AgentRL is an open-source collection of reinforcement-learning research for large language model agents, including a system that jointly trains policy and reward models, a study of what makes agentic RL training effective, and a method for dynamic tool selection.

• Models

Relax is an asynchronous reinforcement-learning framework for post-training large multimodal language models at scale, decoupling training and inference so agents can learn from text, image, video and audio interactions.

• Models

Open-dLLM is an open diffusion-based large language model for code generation, releasing its full stack: pretraining pipeline and datasets, inference scripts, an evaluation suite, and model checkpoints, rather than just weights and inference code.

• Models

A vector post-training quantization method that compresses large language models to under two bits per parameter while keeping them usable, with pre-quantized community models available for models as large as Llama 3.1 405B.

• Models
• Inference

The Hailo Model Zoo provides pre-trained deep learning models plus a full environment for building, quantizing and evaluating them for deployment on Hailo AI accelerator devices.

• Models
• Inference

An open-source training framework for large language and multimodal models that runs the same codebase from one GPU to multiple nodes, supporting pretraining, supervised fine-tuning, preference optimization and asynchronous multi-turn reinforcement learning.

• Models

A series of open-weight 11-billion-parameter vision-language models for long-form and real-time video understanding, including a variant that watches a continuous video stream and decides on its own when to respond.

• Models

A toolkit for compressing large generative AI models, including language models, vision-language models and video-generation models, using quantization, pruning and token-reduction algorithms to cut size while preserving performance.

• Models
• Inference

A tool that converts an existing open large language model into a calibrated decision model, returning a typed answer and a real probability for a choice, a yes/no or a score, without any additional training.

• Models
• Inference

A command-line tool that estimates whether a Hugging Face model will fit and fine-tune on a given consumer GPU, and recommends batch size, sequence length, LoRA rank and quantization before any weights are downloaded.

• Models
• CLI tools

A C++ and CUDA toolkit that trains and serves large language models on NVIDIA GPUs, combining pretraining, LoRA or QLoRA fine-tuning, GRPO or DPO optimization and HTTP-based serving in one system.

• Models
• Inference

A Swift package for building tools and applications with large language models and vision-language models using MLX Swift, with LoRA and full fine-tuning support.

• Models
• Libraries

Research code accompanying a paper on masked diffusion language models for code generation, including inference examples and a Coupled-GRPO post-training method.

• Models
• Papers

An open-source AI platform that runs entirely on local hardware for building RAG applications, training custom classifiers, detecting anomalies and processing documents, with no cloud dependency.

• RAG
• Models

A library for training speculative-decoding draft models that deploy directly into LLM inference engines such as vLLM, speeding up inference without changing output quality.

• Models
• Inference

A config-based command-line tool for running and comparing LLM fine-tuning experiments, controlling prompts, models, optimization strategy and testing from a single YAML file.

• Models

A tool for generating synthetic training data for language models and agent evaluations, combining reasoning traces with tool-calling patterns through topic-graph-guided sample generation.

• Models
• AI evals

A PyTorch-native distributed training library for fine-tuning large language and vision-language models directly from their Hugging Face checkpoints, with recipes for many current open-weight model families.

• Models

A reinforcement-learning framework for post-training unified multimodal models, coordinating rollout generation, scoring and policy updates across pluggable trainers, rollout engines and algorithms.

• Models