# Training Krea 2: What matters in generative model training — Sangwu Lee, Krea.ai

Source: [AI Engineer](https://www.youtube.com/watch?v=-tviRdpmHvs)  
Feed7 permalink: https://feed7.dev/p/training-krea-2-what-matters-in-generative-model-training-sangwu-lee-kre-0m4n8q5  
Published: 2026-08-18T14:00:06.000Z  
Trust: Source Linked (source_linked)

## Why Included

Krea 2’s training notes put data curation and iteration speed ahead of architecture novelty, with explicit safeguards against filtering away unusual visual styles.

## Source Summary

Krea trained Krea 2 on **2–10 billion images**, using basic hashes before semantic deduplication and roughly **30–40 in-house filters**. Its pipeline progressed from **256px to 1K** before mid-training, supervised tuning, preference optimization, and reinforcement learning.

## Practical Implication

For image systems, treat dataset composition as a product decision. Audit aesthetic filters for style collapse, distill expensive vision-model judgments into cheap classifiers, and keep the training stack simple enough to test changes quickly.

## Agent-Ready Context

Krea trained Krea 2 on **2–10 billion images**, using basic hashes before semantic deduplication and roughly **30–40 in-house filters**. Its pipeline progressed from **256px to 1K** before mid-training, supervised tuning, preference optimization, and reinforcement learning.

For image systems, treat dataset composition as a product decision. Audit aesthetic filters for style collapse, distill expensive vision-model judgments into cheap classifiers, and keep the training stack simple enough to test changes quickly.

These are research notes from one model rather than controlled comparisons of every lever. More reliable outputs may still trade away diversity, and the talk does not quantify Krea 2’s gains against alternatives.

## Context Map

- Layer: model
- Domains: image
- Topics: generative-media, open-models

## Uncertainty

- These are research notes from one model rather than controlled comparisons of every lever. More reliable outputs may still trade away diversity, and the talk does not quantify Krea 2’s gains against alternatives.

## Agent Instruction

Use this item as source-backed context. Do not invent claims beyond the linked source. If this item conflicts with another source, call out the conflict.