[mlzerotoheroyoutube]

What does this PR do?

Fixes #14886 (go-ahead from @pratikgx in the issue).

Replaces the 22 dead external links in docs/source/en (16 distinct URLs across 15 files). I re-checked every target today, as suggested. Every URL on a changed line now returns 200.

Group	Old → new
Moved upstream docs	torch_compile_tutorial_.html → docs.pytorch.org/.../torch_compile_tutorial.html; torchao api_ref_quantization.html#… → AOBaseConfig page; generated/…Int4WeightOnlyConfig → api_reference/generated/…Int4WeightOnlyConfig.html; NeMo guides/diffusion/dataset.html → datasets/diffusion-dataset, guides/diffusion/finetune.html → recipes-e2e-examples/diffusion-fine-tuning; kernels skills/cuda-kernels/… → kernel-builder/skills/cuda-kernels/… (2)
Renamed repos	Tencent/HunyuanVideo1-1.5 → Tencent-Hunyuan/HunyuanVideo-1.5 (2 files); pruna-ai/pruna → PrunaAI/pruna
Removed files/pages	Lumina-Next PDF → arXiv 2406.18583; CogVideoX PDF → arXiv 2408.06072 (2 files); webdataset docs site → github.com/webdataset/webdataset; Stable Audio 3 → new news-updates/… slug (2 files); InstructPix2Pix project page → arXiv 2211.09800
Stale internal reference	training/t2i_adapters.md: CI preview moon-ci-docs…/pr_5512/…/create_dataset → relative create_dataset, as in the other training guides

No links inside quoted paper abstracts or code samples were changed. Docs-only; no code or tests affected.

Testing

Every URL on a changed line, checked live (following redirects):

git diff main...HEAD -U0 | grep '^+' | grep -v '^+++' | grep -oE 'https?://[^ )"]+' | sort -u \
  | while read u; do echo "$(curl -s -o /dev/null -w '%{http_code}' -L "$u") $u"; done

200 https://arxiv.org/abs/2211.09800
200 https://arxiv.org/abs/2406.18583
200 https://arxiv.org/abs/2408.06072
200 https://docs.nvidia.com/nemo/automodel/latest/datasets/diffusion-dataset
200 https://docs.nvidia.com/nemo/automodel/latest/recipes-e2e-examples/diffusion-fine-tuning
200 https://docs.pytorch.org/ao/main/api_reference/generated/torchao.core.config.AOBaseConfig.html
200 https://docs.pytorch.org/ao/main/api_reference/generated/torchao.quantization.Int4WeightOnlyConfig.html
200 https://docs.pytorch.org/ao/main/api_reference/generated/torchao.quantization.Int8DynamicActivationInt8WeightConfig.html
200 https://docs.pytorch.org/tutorials/intermediate/torch_compile_tutorial.html
200 https://github.com/PrunaAI/pruna
200 https://github.com/THUDM/CogVideo
200 https://github.com/Tencent-Hunyuan/HunyuanVideo-1.5
200 https://github.com/huggingface/kernels/blob/main/kernel-builder/skills/cuda-kernels/SKILL.md
200 https://github.com/huggingface/kernels/blob/main/kernel-builder/skills/cuda-kernels/references/diffusers-integration.md
200 https://github.com/pytorch/data
200 https://github.com/timothybrooks/instruct-pix2pix
200 https://github.com/webdataset/webdataset
200 https://huggingface.co/blog/custom-cuda-kernels-agent-skills
200 https://huggingface.co/datasets/fusing/fill50k
200 https://huggingface.co/spaces/timbrooks/instruct-pix2pix
200 https://stability.ai/news-updates/meet-stable-audio-3-the-model-family-built-for-artistic-experimentation-with-open-weight-models
200 https://www.tensorflow.org/datasets/tfless_tfds

All 22 return 200, covering the 16 replacements plus the unchanged links that share a line with them. Docs only, so no unit tests apply.

Self-review notes (final round)

Ran the self-review skill against .ai/references/review-rules.md on the whole diff.

• Blocking: none.
• Dead code: none (Markdown only).
• Leave for review (judgement calls):
  1. optimization/fp16.md: the PyTorch dynamic-quantization recipe was removed upstream with no direct successor, so both links now point at torchao's Int8DynamicActivationInt8WeightConfig API page. That is the API the example below them uses. Happy to point somewhere else if you prefer.
  2. api/pipelines/pix2pix.md: the InstructPix2Pix project page is gone, and only a Wayback copy exists. "project page" is now the arXiv "paper"; the codebase and demo links are unchanged.
• Verdict: READY

Before submitting

• Did you use an AI agent (Claude Code, Codex, Cursor, etc.) to help with this PR? If so:
  • Did you read the Coding with AI agents guide?
  • Did you run the self-review skill on the diff?
  • Did you share the final self-review notes in the PR description or a comment?
• Did you read the contributor guideline?
• Did you read our philosophy doc? (important for complex PRs)
• Was this discussed/approved via a GitHub issue or the forum? Please add a link to it if that's the case. [docs] 22 dead external links across the English docs #14886
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests? (not applicable: docs only)
• Are you the author (or part of the team) of the model/pipeline (only applicable for model/pipeline related PRs)?

Who can review?

@pratikgx @stevhliu

🤖 Generated with Claude Code

[stevhliu]

thanks for fixing! only comment is to lets use the huggingface.co/papers link :)

[stevhliu]

Suggested change

	The 3D variational autoencoder (VAE) model with KL loss used in [CogVideoX](https://github.com/THUDM/CogVideo) was introduced in [CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer](https://arxiv.org/abs/2408.06072) by Tsinghua University & ZhipuAI.
	The 3D variational autoencoder (VAE) model with KL loss used in [CogVideoX](https://github.com/THUDM/CogVideo) was introduced in [CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer](https://huggingface.co/papers/2408.06072) by Tsinghua University & ZhipuAI.

[stevhliu]

Suggested change

	[Lumina-Next : Making Lumina-T2X Stronger and Faster with Next-DiT](https://arxiv.org/abs/2406.18583) from Alpha-VLLM, OpenGVLab, Shanghai AI Laboratory.
	[Lumina-Next : Making Lumina-T2X Stronger and Faster with Next-DiT](https://huggingface.co/papers/2406.18583) from Alpha-VLLM, OpenGVLab, Shanghai AI Laboratory.

[stevhliu]

Suggested change

	You can find additional information about InstructPix2Pix in the [paper](https://arxiv.org/abs/2211.09800), [original codebase](https://github.com/timothybrooks/instruct-pix2pix), and try it out in a [demo](https://huggingface.co/spaces/timbrooks/instruct-pix2pix).
	You can find additional information about InstructPix2Pix in the [paper](https://huggingface.co/papers/2211.09800), [original codebase](https://github.com/timothybrooks/instruct-pix2pix), and try it out in a [demo](https://huggingface.co/spaces/timbrooks/instruct-pix2pix).

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

Pages changed in this PR:

• en/api/models/autoencoder_kl_hunyuan_video15
• en/api/models/autoencoder_same
• en/api/models/autoencoderkl_cogvideox
• en/api/models/cogvideox_transformer3d
• en/api/models/hunyuan_video15_transformer_3d
• en/api/models/stable_audio_3_transformer
• en/api/pipelines/lumina
• en/api/pipelines/pix2pix
• en/optimization/fp16
• en/optimization/pruna
• en/quantization/torchao
• en/training/controlnet
• en/training/nemo_automodel
• en/training/t2i_adapters
• en/using-diffusers/text-img2vid

[mlzerotoheroyoutube]

Thanks @stevhliu! I switched all four to huggingface.co/papers: your three suggestions, plus cogvideox_transformer3d.md, which had the same CogVideoX arXiv link. All four return 200.