master-8f6b4a3

fix: enhance the tokenizer's handing of Unicode (#120)

master-9842a3f

fix: add support for int32_t on other compilers (#114)

master-ca33304

fix: remove dangling pointer to work_output in CLIPTextModel (#111)

master-ac8f5a0

feat: add SD-Turbo support

master-69efe3c

chore: make code cleaner

master-2eac844

fix: generate image correctly in img2img mode

master-134883a

feat: add TAESD implementation - faster autoencoder (#88)

* add taesd implementation

* taesd gpu offloading

* show seed when generating image with -s -1

* less restrictive with larger images

* cuda: im2col speedup x2

* cuda: group norm speedup x90

* quantized models now works in cuda :)

* fix cal mem size

---------

Co-authored-by: leejet <leejet714@gmail.com>

master-f99bcd1

fix: detect model format base on file content

master-d7af2c2

feat: load weights from safetensors and ckpt (#101)

master-8a87b27

fix: allow model and vae using different format