Releasing ViTok-v2: open-source ViT auto-encoder codebase + pretrained weights ...

Philippe Hansen-Estruch (@tokenpilled65B)

2026-01-20 | ❤️ 413 | 🔁 57 | 💬 15

---

Releasing ViTok-v2: open-source ViT auto-encoder codebase + pretrained weights

Train your own ViT auto-encoder on any streamed (hf://) or local webdataset. NaFlex pipeline handles any resolution and aspect ratio

Includes reproduced 350M and 4.5B models weights competitive at 256p, SOTA at high-res (512p+)

미디어

---

🔗 Related

• what-if-we-could-model-vision-like-a-wave-moving-through — 주제: AI-ML, Dev-Tools
• video-models-serve-as-a-good-pretrained-backbone-for-robot — 주제: AI-ML, Dev-Tools
• introducing-shaper-a-method-for-robust-conditional-3d-shape — 주제: AI-ML, Dev-Tools
• what-if-we-could-train-ai-robots-in-a-perfect-physics — 주제: AI-ML, Dev-Tools
• if-youve-ever-tried-to-create-3dgs-scenes-from-photos-taken — 주제: AI-ML, Web/Graphics

Tags

AI-ML Dev-Tools Web-Graphics