Repository navigation
Conversation
Organize training examples by model family, flavour, and VRAM
…cision Port KohakuFA key centering and precision diagnostics
Keep EMA accumulation in FP32 during mixed-precision training
Serialize DINOv2 cache loading across REPA workers
Target repeated MOSS transformer blocks for regional compilation
Require LyCORIS 4.1.3 to fix LoKR kernel dispatch
Refresh Webshart caption and image metadata caches on worker reload
Sync CuTe attention with KohakuFA-CuTe 9280c20
Port KohakuFA-CuTe eb1c2f2 query-column sharing, staged KV statistics, and cache/replay regressions.
Thread completion does not guarantee that multiprocessing queue feeders have published bucket results. Match queue types to worker types and cover delayed feeder writes in the eval metadata regression.
Synchronize CuTe kernels with upstream 181c1c2 and extend projection-layout, gradient cancellation, and full-width dispatch coverage for SM120.
Optimize SM120 CuTe attention with full-width D512 tiles
Sync KohakuFA-CuTe c0786dd, enable SM86 in native and automatic dispatch, and extend architecture, graph replay, and gradient regressions. Update the attention option documentation and translations.
Add SM86 Ampere support and improve Ada CuTe attention
Disable cuDNN SDPA to prevent nonfinite attention gradients
Enable VAE tiling automatically after CUDA OOM
Respect caption dropout in AnyFlow derivative guidance
Compute AnyFlow latent interpolation in FP32
Add optional FP32 storage for trainable adapter parameters
Preserve runtime dataset settings when reloading metadata caches
Sync KohakuFA-CuTe forward tuning and launch-grid fixes
…extras Include CuTe dependencies in CUDA extras and update installation documentation.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.