Completed
Overfitting in FLUX training and training image quality
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
FLUX LoRA Training Tutorial: From Zero to Hero with Kohya SS GUI - 8GB GPU, Windows
Automatically move to the next video in the Classroom when playback concludes
- 1 Full FLUX LoRA Training Tutorial
- 2 Guide on downloading and extracting Kohya GUI
- 3 System requirements: Python, FFmpeg, CUDA, C++ tools, and Git
- 4 Verifying installations using the command prompt
- 5 Kohya GUI installation process and error-checking
- 6 Setting the Accelerate option in Kohya GUI, with a discussion of choices
- 7 Use of the bat file update to upgrade libraries and scripts
- 8 Speed differences between Torch 2.4.0 and 2.5, particularly on Windows and Linux
- 9 Starting Kohya GUI via the gui.bat or automatic starter file
- 10 Kohya GUI interface and selecting LoRA training mode
- 11 LoRA vs. DreamBooth training, with pros and cons
- 12 Emphasis on extensive research, with over 72 training sessions
- 13 Ongoing research on hyperparameters and future updates
- 14 Selecting configurations based on GPU VRAM size
- 15 Different configurations and their impact on training quality
- 16 "Better colors" configuration for improved image coloring
- 17 Setting the pre-trained model path and links for downloading models
- 18 Significance of training images and potential errors
- 19 Dataset preparation, emphasizing image captioning, cropping, and resizing
- 20 Repeating and regularization images for balanced datasets
- 21 Impact of regularization images and their optional use in FLUX training
- 22 Instance and class prompts and their importance in training
- 23 Setting the destination directory for saving training data
- 24 Preparing training data in Kohya GUI and generated folder structure
- 25 Joy Caption for batch captioning images, with key features
- 26 Joy Caption interface for batch captioning
- 27 Impact of captioning on likeness, with tips for training styles
- 28 Adding an activation token to prompts
- 29 Image caption editor for manual caption editing
- 30 Batch edit options in the caption editor
- 31 Verifying captions for activation token inclusion
- 32 Kohya GUI and copying info to respective fields
- 33 "Train images image" folder path and its relevance
- 34 Setting different repeating numbers for multiple concepts
- 35 Setting the output name for generated checkpoints
- 36 Parameters: epochs, training dataset, and VAE path
- 37 Epochs and recommended numbers based on images
- 38 Training dataset quality, including diversity
- 39 Importance of image focus, sharpness, and lighting
- 40 Saving checkpoints at specific intervals
- 41 Caption file extension option default: TXT
- 42 VAE path setting and selecting the appropriate VA.saveTensor file
- 43 Clip large model setting and selecting the appropriate file
- 44 T5 XXL setting and selecting the appropriate file
- 45 Saving and reloading configurations in Kohya GUI
- 46 Ongoing research on clip large training and VRAM usage
- 47 Checking VRAM usage before training and tips to reduce it
- 48 Starting training in Kohya GUI and explanation of messages
- 49 Messages during training: steps, batch size, and regularization factor
- 50 How to set virtual RAM memory to prevent errors
- 51 Checkpoint saving process and their location
- 52 Output directory setting and changing it for specific locations
- 53 Checkpoint size and saving them in FP16 format for smaller files
- 54 Swarm UI for using trained models and its features
- 55 Moving LoRA files to the Swarm UI folder
- 56 Speed up Swarm UI on RTX 4000 series GPUs
- 57 Generating images using FLUX in Swarm UI
- 58 Generating an image without a LoRA using test prompts
- 59 VRAM usage with FLUX and using multiple GPUs for faster generation
- 60 Using LoRAs in Swarm UI and selecting a LoRA
- 61 Generating an image using a LoRA in Swarm UI
- 62 Optional in-painting face feature in Swarm UI
- 63 Overfitting in FLUX training and training image quality
- 64 Finding the best checkpoint using the Grid Generator tool in Swarm UI
- 65 Grid Generator tool for selecting LoRAs and prompts
- 66 Generating the grid and expected results
- 67 Analyzing grid results in Swarm UI
- 68 Finding the best LoRA checkpoint based on grid results
- 69 Generating images with wildcards in Swarm UI
- 70 Save models on Hugging Face with a link to a tutorial
- 71 Training SDXL and SD1.5 models using Kohya GUI
- 72 Using regularization images for SDXL training
- 73 Saving checkpoints during SDXL training
- 74 Extracting LoRAs from SDXL models
