Exploring LoRA Training Settings for Ideogram 4: 1700 Steps with 24 Images
Zealousideal-Car4724 · reddit · 2026-07-30
A user shared practical experience on training a LoRA model using Ideogram 4. The model is configured with Rank 32 and Alpha 16, utilizing a dataset of 24 images.
The author found that training for 1,300 steps resulted in an underfitted model with many generation errors, but increasing to 1,700 steps showed noticeable improvements. They are now seeking community advice on whether pushing the training to 2,000 or 2,500 steps would yield better results.
More from Multimodal
- MiniMax H3 Tested: Impressive Results with 3x3 Grid Prompts — techhalla · 2026-07-30
- From Single Image to Cinematic Video: Complete AI Workflow Revealed — SimplyAnnisa · 2026-07-30
- Users Anticipate Open-Source FLUX 3 Video Model, Complain About LTX 2.3 Glitches — EverythingMacPro · 2026-07-30
- Sarvam AI Voice Model Demo: Cinematic Expressiveness Shocks Testers — Scobleizer · 2026-07-30
- CivitAI Massively Removes Adult and Face LoRAs, Shocking Returning Users — literally-me-bro · 2026-07-30
- Restoring Black & White Videos with LTX 2.3 AI Colorization Workflow — BrilliantIll5962 · 2026-07-30