Age | Commit message (Collapse) | Author |
|
Exclude clip index from merge
|
|
|
|
|
|
election
rework VAE resolving code to be more simple
|
|
subject) and moves live preview settings to its own tab
|
|
add checkpoint hash to footer
|
|
|
|
|
|
|
|
|
|
|
|
|
|
add gradient settings to training settings log files
|
|
Fix extension parameters not being saved to last used parameters
|
|
Add script callback for fixing infotext parameters
|
|
fix an error when generating a picture with embedding in it
|
|
Enable batch_size>1 for mixed-sized training
|
|
|
|
|
|
|
|
|
|
always show all tools each in own tab
|
|
add textinfo to progress api response
|
|
Set TQDM progress bar and state textinfo description
|
|
Fix prompt parser default step transformer
|
|
TypeError: expected str, bytes or os.PathLike object, not NoneType
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Support loading textual inversion embeddings from safetensors files
|
|
|
|
|
|
|
|
|
|
|
|
|
|
add fallback for when quick model creation fails
|
|
prevent .vae.safetensors files from being listed as stable diffusion models
|
|
to speedup creating sd model from config
|
|
|
|
|
|
|
|
|
|
|
|
Implements variable dropout rate from #4549
Fixes hypernetwork multiplier being able to modified during training, also fixes user-errors by setting multiplier value to lower values for training.
Changes function name to match torch.nn.module standard
Fixes RNG reset issue when generating previews by restoring RNG state
|