A Modular Open-Source Framework for Systematic Training, Inference, and Deployment of World-Action Models

📦 Huggingface Collections

• 🚀 OpenWAM_Alpha: Public Release of our pretrained checkpoint (OpenWAM-Alpha), plus finetuned variants across downstream benchmarks.

• 🔬 OpenWAM-Study — checkpoints from our controlled study on WAM design principles, split by axis:

  • 🏗️ OpenWAM_Study_Architecture: Architecture Variant Checkpoints for OpenWAM-Study
  • 🌱 OpenWAM_Study_Pretrain: Pretrain Experiments Checkpoints for OpenWAM-Study
  • 👁️ OpenWAM_Study_Visual_Encoder: Visual Encoder Variant Checkpoints for OpenWAM-Study
  • 🎬 OpenWAM_Study_Video_Backbone: Video Backbone Variant Checkpoints for OpenWAM-Study
  • 🎭 OpenWAM_Study_Attention_Mask: Attention Mask Variant Checkpoints for OpenWAM-Study