Skip to content

Pull requests: ostris/ai-toolkit

New pull request New
65 Open 251 Closed
65 Open 251 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

prompt cache: preserve frozen tensor dtypes across serialization
#961 opened Jul 17, 2026 by Zironic Contributor Loading…
captioner: fix Ideogram4 prompt Unicode escape syntax
#960 opened Jul 17, 2026 by Zironic Contributor Loading…
run: preserve original error when job creation fails
#959 opened Jul 17, 2026 by Zironic Contributor Loading…
Skip empty cancelled timers
#958 opened Jul 17, 2026 by Zironic Contributor Loading…
Honor model-declared patch size
#957 opened Jul 17, 2026 by Zironic Contributor Loading…
Disable exhaustive tuning for quantized compile
#951 opened Jul 15, 2026 by Zironic Contributor Loading…
A smarter, dynamic, quant and model agnostic Memory Manager.
#948 opened Jul 14, 2026 by Zironic Contributor Loading…
krea2: fix timestep-shift training
#947 opened Jul 14, 2026 by tcldr Loading…
Add support for reordering active jobs. Desktop+Mobile
#942 opened Jul 11, 2026 by fatalis Contributor Loading…
Use the generate config sampler instead of the undefined train_config in GenerateProcess
#936 opened Jul 6, 2026 by rajakshay Loading…
1
Send T5 tokens to the T5 encoder's own device when caching text embeddings
#935 opened Jul 6, 2026 by rajakshay Loading…
2
Log total training params for every optimizer
#932 opened Jul 4, 2026 by fatalis Contributor Loading…
Detect AMD GPUs with amd-smi if nvidia-smi fails.
#931 opened Jul 3, 2026 by sjuxax Loading…
Add qfloat8 offloading
#930 opened Jul 3, 2026 by PlagueKind Contributor Loading…
Expose pretrained LoRA path in simple UI
#928 opened Jul 3, 2026 by mini-mind Loading…
Fix CUDA OOM during Krea2 block-by-block quantization
#927 opened Jul 2, 2026 by thedivergentai Loading…
3 tasks done
Add offline mode to settings. Sets HF_HUB_OFFLINE=1 when enabled
#926 opened Jul 2, 2026 by fatalis Contributor Loading…
1
Relocate Krea 2 to Images section
#923 opened Jun 29, 2026 by astreloff Loading…
Fix several cases of silent crashing on UI
#922 opened Jun 29, 2026 by fatalis Contributor Loading…
Fix Krea2/Qwen-Image flat samples: use a VAE with non-zero quant convs
#919 opened Jun 28, 2026 by skydam Loading…
fix(ui): resolve 9 npm security vulnerabilities
#917 opened Jun 27, 2026 by 7y2v62s55s-maker Loading…
Fix Krea 2 README categorization from Video to Image
#915 opened Jun 26, 2026 by alpaim Loading…
1
vram reduction for all models and slight it/s boost
#913 opened Jun 24, 2026 by PlagueKind Contributor Loading…
Fix most typos in the repo
#912 opened Jun 24, 2026 by fatalis Contributor Loading…
Honor low_vram in ACE-Step load_model (fix OOM during latent caching)
#911 opened Jun 24, 2026 by SanDiegoDude Loading…
1 of 2 tasks
ProTip! Follow long discussions with comments:>50.