Releases · Haidra-Org/hordelib

ci: fix PR tests image deployment by @jug-dev in #3
feat: Re-aded clip/blip based on external clipfree library by @db0 in #2
fix: clipfree compatability fixes by @tazlin in #5
fix: make finding the model directory backwards compatible by @jug-dev in #7
docs: update readme by @jug-dev in #8
ci: re-enable pypi release publishing by @jug-dev in #9
chore: use clipfree as a pypi package ('horde_clipfree') by @tazlin in #11
fix: don't cache incorrect location of model directory by @jug-dev in #14
fix: remove some unneeded logger messages by @tazlin in #16
refactor: The pytest rework no-one asked for, adjustment to packaging behavior by @tazlin in #17
fix: actually attempt to redownload on hash mismatch by @tazlin in #19
Fixes lost issue of image_is_control not working by @spinagon in #18
fix: remove rlimit hack from 3rd party module by @tazlin in #21
refactor: Improved type hints, various fixes, removes loguru hijacks by @tazlin in #22
fix: taint models readded, more accurate dependency reqs by @tazlin in #23
fix: quiet log spam from get_mm_pointers(...) by @tazlin in #25
Tests: negative lora model strength values by @db0 in #50
refactor: Code cleanup, py 3.11 support, better type hints, mypy assisted bug squashing by @tazlin in #53
Loras: "any" trigger picks one randomly, "all" trigger uses all triggers joined by ", " by @Efreak in #56
fix: store lora.json with other model reference files by @tazlin in #58
fix: release yaml oversight by @tazlin in #61
feat: Automatic downloading of TIs by @db0 in #55
fix: TI now correctly compares to intended sha256 from hordeling by @tazlin in #65
fix: inject negative embeddings correctly by @tazlin in #70
fix: use pre-commit for lint checks by @tazlin in #73
refactor: hordelib now single threaded, no more need for .patch; feat: SDXL support by @tazlin in #68
feat: add AIWORKER_LORA_CACHE_SIZE env var by @tazlin in #75
fix: handle lora env var None by @tazlin in #77
fix: better signal to comfyui lowvram mode is desired by @tazlin in #78
feat: torch 2.1; better memory management in high vram scenarios by @tazlin in #80
docs: fix reference to old cuda version by @tazlin in #82
feat: graceful fail if we detect a login redirect from civitai for loras by @tazlin in #89
feat: use comfy dd116abfc48e8023bb425c2dd5bd954ee99d7a9c by @tazlin in #90
feat: use latest comfyui (version as of Nov 10, 2023) by @tazlin in #92
tests: updated expected images for lora+hires and k_dpmpp_sde by @tazlin in #94
tests: fix more expected images by @tazlin in #95
feat: Inference and PP return objects instead of Images by @db0 in #96
chore: add dependabot support by @tazlin in #97
build(deps): bump the python-packages group with 6 updates by @dependabot[bot] in #98
fix: don't count tainted models as available by @tazlin in #99
fix: allow success messages (level 25) to show at verbosity 2 by @tazlin in #109
feat: update to comfy version e13454 (SDV support; version as of Dec 6 2023) by @tazlin in #114
fix: Handle missing lora IDs by @db0 in #116
fix: must initiate final_rawpng by @db0 in #118
fix: do download timeout checks for TIs too by @tazlin in #122
fix: retry hordeling more often; try fewer times on 500s by @tazlin in #124
Allows multiple lora versions by @db0 in #127
feat: update to comfy version c78214 by @tazlin in #134
fix: support comfyui args; detect pipeline changes between runs by @tazlin in #135
fix: purge caches for comfyui executor by @tazlin in #137
fix: downloading loras on fast systems no longer causes duplicate images by @db0 in #133
ci: run lora tests sooner; samplers test later by @tazlin in #139
ci: lora fixture no longer hangs by @tazlin in #141
fix: avoid sending old images on faults; ci: do more tests on comfyui output by @db0 in #144
Fixes requesting generic lora not always returning latest version. by @db0 in #147
fix: orig name not searched in lowercase by @db0 in #153
fix: try and fallback to on-disk model ref when can't download by @tazlin in #159
feat: Allow img2img to use n_iter by @db0 in #158
build(deps): bump the python-packages group with 7 updates by @dependabot[bot] in #150
ci: use CIVIT_API_TOKEN in pr/main tests by @tazlin in #169
feat: Add CivitAI token while downloading Loras and Ti's by @GabrielJanczak in #162
fix: Attempt to stop loras going randomly missing by @db0 in #167
feat: update to comfy version c5a369a by @tazlin in #177
feat: pyinstaller support by @tazlin in #181
build(deps-dev): bump the python-packages group with 4 updates by @dependabot[bot] in #179
fix: remove torchaudio from req. deps by @tazlin in #183
fix: attempt to reload the json on fail by @db0 in #191
Feat: Adds support for Stable Cascade by @db0 in #195
feat: support comfyui f81dbe2 by @tazlin in #197
build(deps): bump the python-packages group with 6 updates by @dependabot[bot] in #194
ci: split pre-commit into own job; set HORDELIB_CI_ONGOING by @tazlin in #198
feat: free resources less often; fix: handle non-standard inpainting payloads by @tazlin in #199
feat: reduce cfg on sampler_stage_b makes it faster by @db0 in #200
feat: Passes CivitAI token on checkpoint download as well by @db0 in #202
fix: handle 401s for lora dls as terminal; spend less time retrying by @tazlin in #208
fix: increase timeout rate of lora metadata/model downloads by @tazlin in #211
feat: callbacks for comfyui inference/post process progress; update comfyui to 2a813c3b by @tazlin in #216
feat: img2img on Stable Cascade, update comfyui to 40e124c6; fix: minor bug fixes by @db0 in #219
build(deps-dev): bump the python-packages group with 5 updates by @dependabot[bot] in #218
feat: support comfyui a28a9dc by @tazlin in #226
fix: missing extra_source_images conversion to PIL by @db0 in #228
fix: avoid crash on long lora names by @db0 in #232
feat: update to comfyui a7dd82e by @tazlin in #235
feat: use comfyui 16eabdf7 by @tazlin in #244
feat: Allow providing custom models (comfyui 16eabdf7) by @tazlin in #245
feat: Removes the requirement for a design pipeline by @db0 in #251
feat: adds stable_cascade_2pass by @db0 in https://github.com/Haidra-Or...

No results found