Trained for 192 epochs and 2500 steps.

Trained with datasets ['text-embeds', 'guimiv2']
Learning rate 0.0001, batch size 4, and 1 gradient accumulation steps.
Used DDPM noise scheduler for training with epsilon prediction type and rescaled_betas_zero_snr=False
Using 'trailing' timestep spacing.
Base model: /data/yihao/FLUX.1-dev
VAE: None

pytorch_lora_weights.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4eee07aa1d8a6e1f78e612d25529dc852755bd53d55e76256ef90a5ded103888
-size 89840872
+oid sha256:e796cbc8eecf1afca260ecfcaaecdcf18c18b2370eb3b7699adb92d7e04060be
+size 89841632