whitecircle/halo

This page introduces an external open-source repository. It is not an HDATF product.

What it does

A training framework for language and multimodal models supporting pretraining, supervised fine-tuning, preference optimization, and asynchronous multi-turn reinforcement learning. It runs Hugging Face models on one GPU or across nodes.

How it helps ATF

LabChin can compare the shared training infrastructure when planning experiments across model-training methods.

License

Custom license Custom or non-standard license. Read the license file before any reuse.

More in this category

• jaredpalmer/kev
  Small Qwen3.5-based decision models with pretrained weights, training code, and evaluation data. The API supports yes/no, multiple-choice, and rating questions and matches TypeSafe System One.
• dexmal/dexbotic
  A PyTorch toolbox for vision-language-action research covering pretraining, fine-tuning, inference, and evaluation. It provides multiple model configurations, a unified robot training data format, and deployment scripts.
• mizorewww/laya-mlx
  A native MLX runtime for Laya typed decision models on Apple Silicon. It performs local inference without text generation, a PyTorch runtime, or a cloud API.
• NandhaKishorM/laya
  A multilingual decision engine that answers typed questions about text, email, tickets, or JSON in a single forward pass. A router selects a checkpoint for each request.
• mizorewww/laya-coreml
  A port of Laya typed decision models to Apple Core ML and the Neural Engine, running on device without generating tokens, published with speed and energy benchmarks and a Snake demo.

Only repositories in the ranked Trendshift lists are included, and the lists are used only to find candidates. We do not copy their ranks. Descriptions, licenses and star counts come from each GitHub repository. The notes are our own reading. We have not tested these projects, and a place on a trending list does not prove quality.