DarkViewAI

flux text encoder training not working

3

Traceback (most recent call last): File "/home/ai/Downloads/kohya_ss/sd-scripts/flux_train_network.py", line 411, in trainer.train(args) File "/home/ai/Downloads/kohya_ss/sd-scripts/train_network.py", line 330, in train self.assert_extra_args(args, train_dataset_group) File "/home/ai/Downloads/kohya_ss/sd-scripts/flux_train_network.py", line 43, in assert_extra_args assert ( AssertionError: network for...

Possible Enhancements/optimizations. - Is quantization possible for kohya? ai toolkit has it. Serious speed boost for 1024 training, and bucketing multi resolutions.

3

1. Is it possible to add quantization? with ai toolkit i get 22.1gb vram doing 1024 training, and 1400 steps done in 45 mins in kohya 1024 training, takes me...

Extracted flux lora not working

16

2 Issues 1. sometimes it works, but it looks nothing like the fine tuned model. 2. Doesnt work and i get the error below During handling of the above exception,...

Flux Dreambooth crashes after trying to save checkpoint

2

I am able to successful db training and samples look good, but after 600 steps it gives me this error 2024-08-19 07:57:06 INFO train_util.py:5123 INFO saving checkpoint: train_util.py:5124 /home/Ubuntu/Downloads/model 4/Rick-step00000600.safetensors...

Apply T5 Attention Mask - Horrible Results

4

   I applied t5 attention mask, and the results are quite bad

Text encoder training support FLux

1

i saw that on here https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/README_flux.md you can train the text encoder. Is it possible to add it to sd-scripts

Support For T5XXL Unchained

6

Someone made an updated t5 with new vocab. https://github.com/Kaoru8/T5XXL-Unchained https://github.com/kohya-ss/sd-scripts/issues/1987 https://huggingface.co/Kaoru8/T5XXL-Unchained/tree/main This is a fork of the T5-XXL encoder that uses an extended tokenizer to facilitate Flux uncensoring and offer...