Fix SD upscale Batch count (#2950)

Add support for fp8 scaled (#2946)

Multithreading softinpainting (#2927)

Add Chroma (#2925)

Repair ancestral sampling for FLUX (#2915)

Resolve warnings of datetime library (#2874)

Add BF16 to GGUF (#2877)

Support wd_on_output for DoRA (#2856)

re-add line to update sampling step (#2853)

Add separate cuda stream for live preview VAE (#2844)

update bitsandbytes version for cuda128 support (#2712)

some extra lora support, inc. new glora (#2715)

fix zero'd cond_t5 device (#2701)

this is useful (spiece.model for SD3 T5 tokenizer)(#2699)

SD3+ (#2688)