Repository navigation

All pull requests

Labels (9)Milestones (0)New pull request

Open

(29)

Closed

(87)

Fix Low VRAM Audio VAE loader returning an unusable model
#562Henry Kroll III (themanyone) opened Sep 19, 2026
1
fix: replace invalid kornia pad import with F.pad in pyramid_blending
#556Jhonny Rosa (lordekeen) opened Sep 1, 2026
1
fix(stg): drop unsupported momentum_buffer kwarg from APG call (#492)
#555Tai An (Anai-Guo) opened Aug 30, 2026
Preserve non-tensor conditioning options across save and load
#554しの (vallicgrr) opened Aug 25, 2026
Fix: clone latent samples in LTXVImgToVideoConditionOnly to prevent in-place mutation across list/batch execution
#552kb999-coder opened Aug 22, 2026
Fix broken pad import by switching to torch.nn.functional.pad
#551afkaf opened Aug 22, 2026
Fix NaN output from LTXVAdainLatent
#540Jean J. de Jong (jjdejong) opened Jul 31, 20264 tasks
fix: cap tiled VAE decode output at 8GB RAM, disk-back above that, an…
#530Juan Fidel (johnnycaifan) opened Jul 22, 2026
Use torch padding with modern kornia
#526Julien D (jdaln) opened Jul 11, 2026
Fix ImportError: kornia removed pad re-export from geometry.transform.pyramid
#516Vivi (VixenVillainous) opened Jun 18, 2026
3
perf: run LTXVDilateVideoMask pooling on GPU for massive speedup
#514Denarius40 opened Jun 17, 2026
fix: use torch padding for kornia 0.8 compatibility
#508Nianze Wu (wunianze666-netizen) opened Jun 9, 2026
1
fix: initialize multimodal guider post-hook fallbacks at cfg=1
#507Nianze Wu (wunianze666-netizen) opened Jun 9, 2026
1
fix(text-embeddings): compare text_encoder_norm_type case-insensitively
#504James D. Finlon (jim-finlon) opened Jun 5, 2026
Fix STG crash/index miscount with guide-mask self-attention
#503Jean J. de Jong (jjdejong) opened Jun 2, 2026
fix(stg): skip self-attention by module name instead of call index
#501Nekodificador (Nekodificador) opened May 26, 20263 tasks done
Add LTXAV model detection monkey-patch for checkpoint compatibility
#482Marcel Hofer (goodguy1963) opened May 9, 2026
Fix ImportError: precompute_freqs_cis is now _precompute_freqs_cis on LTXBaseModel
#481Marcel Hofer (goodguy1963) opened May 9, 2026
fix(text-embeddings): compare text_encoder_norm_type case-insensitively
#478Albatross1382 opened May 2, 2026
AV latent support for LTXVLoopingSampler and LTXVExtendSampler
#472Jean J. de Jong (jjdejong) opened Apr 25, 20266 tasks
1
feat: add MiniMax prompt enhancer node
#466Octopus (octo-patch) opened Apr 17, 20264 tasks done
fix: UnboundLocalError for noise_pred_neg/perturbed when cfg=1.0
#460Frank Greico (codeman101) opened Apr 8, 2026
feat: load LTX-2.3 connector weights from GGUF on Apple Silicon
#431Samir Hassen (samirhassen) opened Mar 10, 2026
Refactor: Unify and organize node categories
#401Brent (drphero) opened Jan 31, 2026
Enhance device management and memory optimization for selected GPU
#392FearL0rd opened Jan 27, 2026

Navigation Menu

Repository navigation

All pull requests

Fix Low VRAM Audio VAE loader returning an unusable model

fix: replace invalid kornia pad import with F.pad in pyramid_blending

fix(stg): drop unsupported momentum_buffer kwarg from APG call (#492)

Preserve non-tensor conditioning options across save and load

Fix: clone latent samples in LTXVImgToVideoConditionOnly to prevent in-place mutation across list/batch execution

Fix broken pad import by switching to torch.nn.functional.pad

Fix NaN output from LTXVAdainLatent

fix: cap tiled VAE decode output at 8GB RAM, disk-back above that, an…

Use torch padding with modern kornia

Fix ImportError: kornia removed `pad` re-export from geometry.transform.pyramid

perf: run LTXVDilateVideoMask pooling on GPU for massive speedup

fix: use torch padding for kornia 0.8 compatibility

fix: initialize multimodal guider post-hook fallbacks at cfg=1

fix(text-embeddings): compare text_encoder_norm_type case-insensitively

Fix STG crash/index miscount with guide-mask self-attention

fix(stg): skip self-attention by module name instead of call index

Add LTXAV model detection monkey-patch for checkpoint compatibility

Fix ImportError: precompute_freqs_cis is now _precompute_freqs_cis on LTXBaseModel

fix(text-embeddings): compare text_encoder_norm_type case-insensitively

AV latent support for LTXVLoopingSampler and LTXVExtendSampler

feat: add MiniMax prompt enhancer node

fix: UnboundLocalError for noise_pred_neg/perturbed when cfg=1.0

feat: load LTX-2.3 connector weights from GGUF on Apple Silicon

Refactor: Unify and organize node categories

Enhance device management and memory optimization for selected GPU

Pull requests