HunyuanDiT v1.1 distilled: 25-step diffusers pipeline, Chinese and English prompts

Hunyuan-DiT is Tencent's multi-resolution diffusion transformer with fine-grained Chinese understanding (arXiv 2405.08748). This repository holds a distilled v1.1 in diffusers format that the card says supports 25-step text-to-image generation. Setup from the card: install PyTorch, pip install --upgrade transformers, and diffusers from GitHub. Then HunyuanDiTPipeline.from_pretrained("Tencent-Hunyuan/HunyuanDiT-v1.1-Diffusers-Distilled", torch_dtype=torch.float16).to("cuda") and pipe(prompt).images[0]; prompts can be Chinese or English. The example does not set a step count, so pass num_inference_steps=25 to get the distilled step budget rather than relying on the pipeline default. The card's comparison table (human evaluation against SDXL, PixArt-alpha, Playground 2.5, SD 3, Midjourney v6 and DALL-E 3) describes the original Hunyuan-DiT, not this distilled checkpoint. Licence: Tencent Hunyuan Community License (licence file in the Tencent-Hunyuan/HunyuanDiT repository). Based on the Tencent-Hunyuan/HunyuanDiT-v1.1-Diffusers-Distilled card pinned to revision 527cf2ecce7c04021975938f8b0e44e35d2b1ed9 and observed on 2026-10-27. Nothing was run here; the advice to set 25 steps explicitly is a curator note because the example omits the step count, and the licence text was not reviewed. 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/Tencent-Hunyuan/HunyuanDiT-v1.1-Diffusers-Distilled/raw/527cf2ecce7c04021975938f8b0e44e35d2b1ed9/README.md",
      "note": "Retained source snapshot observed 2026-10-27; TEXT.txt; UTF-8 bytes 647:761, 763:808, 1270:1402, 19:190; exact excerpt SHA256 f6c1889367e59eb2cf0446fca6f391194216d177ea8c76390a6d9583580ba39a,18ca149a7ca64c5ea432753a98de5422fda2187e8b5475cb4abda33357a0729e,16d7382cf8bf332edb3d234d7426078cf6af6f75f9c20173962b9e01493bf2b8,4ac27a1baf9822b60c3b0fe99"
    }
  ]
}