TypingMindTypingMindThe best frontend for AI
Diffusion Engineering logo

Diffusion Engineering

Community
AnastasiyaW
diffusion-engineering

Практическая инженерия диффузионных моделей: архитектуры, обучение, инференс, оптимизация памяти. Использовать при любых задачах с диффузионными моделями: проектирование или модификация архитектуры (UNet/DiT/Flow/Flux), выбор и настройка schedulers/samplers, дообучение (LoRA/DreamBooth/full fine-tune), оптимизация памяти (AMP/checkpointing/ZeRO/FSDP/quantization), замена или fusion текст-энкодеров (CLIP/Qwen), работа с Diffusers, отладка диффузионных пайплайнов, оценка качества (FID/CLIPScore/LPIPS), latent diffusion, VAE, guidance/CFG, rectified flow, Stable Diffusion, SDXL, Flux. Также применять при вопросах про GPU-память при обучении генеративных моделей, text-to-image пайплайны, ControlNet, multi-encoder fusion, WebDataset. Do NOT use for writing FLUX.2 Klein generation prompts (use flux2-klein-prompting) или для FLUX.2 Klein / Qwen-Edit LoRA-тренинга и edit-LoRA (use flux2-lora-training); это общая диффузионная инженерия, не специфика FLUX.2 Klein.

Overview

PublisherAnastasiyaW
Repositorycodex-claude-code-config
Skill namediffusion-engineering
Stars
212
Forks
32
Bundled files
6
LicenseMIT
Links
Skill sourceRepository
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 6 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by AnastasiyaW on GitHub. Read the source before you install it.

Installation

Install the Diffusion Engineering AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/AnastasiyaW/codex-claude-code-config/tree/main/skills/ai-ml/diffusion-engineering
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/AnastasiyaW/codex-claude-code-config.git /tmp/codex-claude-code-config
mkdir -p .claude/skills
cp -r /tmp/codex-claude-code-config/skills/ai-ml/diffusion-engineering .claude/skills/diffusion-engineering
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Diffusion Engineering in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Diffusion Engineering on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Diffusion Engineering is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Diffusion Engineering Skill

Быстрая ориентация

Три инженерных решения, которые больше всего влияют на качество/скорость/стоимость:

  1. Где идёт диффузия → пиксели (дорого) или латентное пространство (LDM/SD-семейство — практично)
  2. Backbone денойзера → UNet (классика, проще) или Transformer/DiT/Flow (масштабируется лучше)
  3. Управление сэмплингом → scheduler, число шагов, guidance_scale — часто дают больше, чем правка сети

Reference files — читать по задаче

ТемаФайлКогда читать
Архитектуры и data flowreferences/architectures.mdDDPM/SDE/LDM/DiT/Flux/VAE/SDXL, схема пайплайна
Schedulers и guidancereferences/samplers.mdDDIM/Euler/Heun/DPM-Solver/PNDM, CFG, prediction_type
Обучение и дообучениеreferences/training.mdLoss/цели, LoRA/DreamBooth/full FT, гиперпараметры
Память и распределённостьreferences/memory.mdAMP, checkpointing, ZeRO, FSDP, quantization, FP8
Текст-энкодеры и данныеreferences/encoders-data.mdCLIP/Qwen/multi-encoder, токенизация, data pipeline
Оценка и траблшутингreferences/eval-debug.mdFID/CLIPScore/LPIPS, типовые поломки и фиксы, лицензии

Быстрый чеклист «я строю/модифицирую diffusion»

  • Backbone: UNet (проще) или DiT/Flow (масштабирование)?
  • Модули зафиксированы: tokenizer → text encoder → encoder_hidden_states → denoiser → VAE decode
  • Scheduler выбран: DDIM / Euler / DPM-Solver — A/B на фиксированных seed
  • Дообучение: начинать с LoRA, в full fine-tune только при необходимости
  • Память: AMP включён, при необходимости checkpointing, при масштабе ZeRO/FSDP
  • Данные: стриминг/шардинг (HF streaming, WebDataset), валидировать throughput dataloader
  • Оценка: выбрать метрики под требуемый результат и доступные данные, а не обязательный набор FID/CLIPScore/LPIPS. Для парной ретуши проверять сохранность личности/геометрии/цвета и целевое изменение на фиксированном наборе; FID не заменяет эти проверки. См. references/eval-debug.md.

Trade-offs на один экран

РучкаУвеличитьУменьшить
num_inference_steps↑ время; качество проверять для данной модели↓ время; качество проверять, особенно у distilled моделей
guidance_scale (CFG)↑ adherence к промпту, риск «пережога»↑ разнообразие
LoRA rank↑ выразительность↑ параметры, риск overfitting
Шаги дообучения↑ адаптация↑ риск catastrophic forgetting
Batch size↑ стабильность градиентов↑ VRAM

Мини-рецепты по бюджету GPU

БюджетЧто делать
8–16 GB (1 GPU)LoRA вместо full FT; grad accumulation; BF16/FP16; xFormers/SDPA; 8-bit оптимизатор
24–48 GB (1–4 GPU)LoRA или partial FT; иногда FSDP; большее разрешение
8+ GPU, H100Full FT, ZeRO-3/FSDP, float8, WebDataset стриминг, масштабный датапайплайн

Gotchas

  • Универсальный набор метрик превращал малую парную edit-задачу в ненужный сбор тысяч генераций. Метрика должна проверять конкретное требование; численный балл не заменяет визуальную приёмку.
  • Число workers и оптимизации памяти — кандидаты для замера на данном runtime, не обязательная лестница. Больше workers может замедлить Windows-пайплайн или дублировать iterable-данные.

Troubleshooting

СимптомПроверкаДействие
Оценка не отвечает, исправлена ли ретушьСвязать каждую метрику с требованием и эталонной паройЗаменить нерелевантную метрику проверкой нужного свойства; сохранить визуальный контроль
Загрузка данных медленнее после настройкиСравнить throughput, память и уникальность данных с num_workers=0Оставить измеренно подходящие параметры; шардировать iterable dataset при нескольких workers

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Diffusion Engineering AI skill do?

Практическая инженерия диффузионных моделей: архитектуры, обучение, инференс, оптимизация памяти. Использовать при любых задачах с диффузионными моделями: проектирование или модификация архитектуры (UNet/DiT/Flow/Flux), выбор и настройка schedulers/samplers, дообучение (LoRA/DreamBooth/full fine-tune), оптимизация памяти (AMP/checkpointing/ZeRO/FSDP/quantization), замена или fusion текст-энкодеров (CLIP/Qwen), работа с Diffusers, отладка диффузионных пайплайнов, оценка качества (FID/CLIPScore/LPIPS), latent diffusion, VAE, guidance/CFG, rectified flow, Stable Diffusion, SDXL, Flux. Также пр...

Why use Diffusion Engineering on TypingMind?

Because you install it once and use it with any model. Diffusion Engineering is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Diffusion Engineering in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/AnastasiyaW/codex-claude-code-config/tree/main/skills/ai-ml/diffusion-engineering. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Diffusion Engineering?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Diffusion Engineering?

As many as you like. As long as a model supports skills, you can use Diffusion Engineering with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Diffusion Engineering AI skill free?

Yes. It is published on GitHub by AnastasiyaW under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

Related AI Skills

View all
garrytan logo

Landing Report

garrytan
CommunityPopular

Read-only queue dashboard for workspace-aware ship. (gstack)

19.9K
133.5K
1 files
View
garrytan logo

Gstack Openclaw Ceo Review

garrytan
CommunityPopular

Use when asked to review a plan, challenge a proposal, run a CEO review, poke holes in an approach, think bigger about scope, or decide whether to expand or reduce the plan.

19.9K
133.5K
Instructions
View
JuliusBrussee logo

Caveman Stats

JuliusBrussee
CommunityPopular

Show recorded output and cache-read token usage and mode attribution for the current Claude Code session, or locate the host's native usage report. Trigger: /caveman-stats.

6.2K
106.3K
Instructions
View
pytorch logo

Ci Metrics

pytorch
OrganizationPopular

Query PyTorch CI, GitHub Actions, HUD, Grafana, and infrastructure metrics. Use when users ask about CI duration, job failures, queue times, workflow trends, runner health, dashboard data, or PyTorch…

31K
103.6K
1 files
View

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇