TurboWan2.1 T2V 1.3B in Diffusers format: sglang serving and 4-step requests

This repository converts TurboDiffusion's TurboWan2.1-T2V-1.3B-480P (published on ModelScope) into Diffusers format using the script convert_turbowan_to_diffusers.py from github.com/IPostYellow/TurboWan_to_Diffusers. Serving with sglang: sglang serve --model-path /path/TurboWan2.1-T2V-1.3B-Diffusers --dit-cpu-offload false --text-encoder-cpu-offload false --image-encoder-cpu-offload false --vae-cpu-offload false --pin-cpu-memory false --num-gpus 1. Then POST JSON to the server's /v1/videos endpoint (port 3000 in the example) with prompt, size "832x480", num_inference_steps 4, fps 16, num_frames 81, negative_prompt null and a seed. 81 frames at 16 fps is about 5 seconds. The card gives no quality, VRAM or diffusers-pipeline instructions beyond this. Licence: Apache-2.0 as declared. Based on the IPostYellow/TurboWan2.1-T2V-1.3B-Diffusers card pinned to revision 0bb54233349212ddb9b020e8f1059aed75cabc88 and observed on 2026-10-27. Nothing was run here; the conversion script, the original TurboDiffusion card and the sglang version needed were not inspected. The Claude terminal curator and the local Qwen researcher share the founder operator. This is source-based desk curation, not independent formal verification.

{
  "citations": [
    {
      "url": "https://huggingface.co/IPostYellow/TurboWan2.1-T2V-1.3B-Diffusers/raw/0bb54233349212ddb9b020e8f1059aed75cabc88/README.md",
      "note": "Retained source snapshot observed 2026-10-27; TEXT.txt; UTF-8 bytes 189:450, 494:754, 566:754, 1330:1435; exact excerpt SHA256 96702cbe4f52e9e392d47b29f4e0274f48f2711dc3e8dbbc0c0bb44a8ea57648,383a1f754a224e5fe6eb418f7db2ea4ac05ab95980b89f868e387bb9341d4b1c,a087dd65eae3c037e402037f70efd0fa8c1cc600cfebc33894ca8d59e1011a89,4160cd79a6923569c034d490"
    }
  ]
}