[pull] main from kohya-ss:main#47
Merged
pull[bot] merged 6 commits intoadmariner:mainfrom Feb 13, 2026
Merged
Conversation
* Support Anima model * Update document and fix bug * Fix latent normlization * Fix typo * Fix cache embedding * fix typo in tests/test_anima_cache.py * Remove redundant argument apply_t5_attn_mask * Improving caching with argument caption_dropout_rate * Fix W&B logging bugs * Fix discrete_flow_shift default value
* fix: update extend-exclude list in _typos.toml to include configs * fix: exclude anima tests from pytest * feat: add entry for 'temperal' in extend-words section of _typos.toml for Qwen-Image VAE * fix: update default value for --discrete_flow_shift in anima training guide * feat: add Qwen-Image VAE * feat: simplify encode_tokens * feat: use unified attention module, add wrapper for state dict compatibility * feat: loading with dynamic fp8 optimization and LoRA support * feat: add anima minimal inference script (WIP) * format: format * feat: simplify target module selection by regular expression patterns * feat: kept caption dropout rate in cache and handle in training script * feat: update train_llm_adapter and verbose default values to string type * fix: use strategy instead of using tokenizers directly * feat: add dtype property and all-zero mask handling in cross-attention in LLMAdapterTransformerBlock * feat: support 5d tensor in get_noisy_model_input_and_timesteps * feat: update loss calculation to support 5d tensor * fix: update argument names in anima_train_utils to align with other archtectures * feat: simplify Anima training script and update empty caption handling * feat: support LoRA format without `net.` prefix * fix: update to work fp8_scaled option * feat: add regex-based learning rates and dimensions handling in create_network * fix: improve regex matching for module selection and learning rates in LoRANetwork * fix: update logging message for regex match in LoRANetwork * fix: keep latents 4D except DiT call * feat: enhance block swap functionality for inference and training in Anima model * feat: refactor Anima training script * feat: optimize VAE processing by adjusting tensor dimensions and data types * fix: wait all block trasfer before siwtching offloader mode * feat: update Anima training guide with new argument specifications and regex-based module selection. Thank you Claude! * feat: support LORA for Qwen3 * feat: update Anima SAI model spec metadata handling * fix: remove unused code * feat: split CFG processing in do_sample function to reduce memory usage * feat: add VAE chunking and caching options to reduce memory usage * feat: optimize RMSNorm forward method and remove unused torch_attention_op * Update library/strategy_anima.py Use torch.all instead of all. Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * Update library/safetensors_utils.py Fix duplicated new_key for concat_hook. Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * Update anima_minimal_inference.py Remove unused code. Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * Update anima_train.py Remove unused import. Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * Update library/anima_train_utils.py Remove unused import. Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * fix: review with Copilot * feat: add script to convert LoRA format to ComfyUI compatible format (WIP, not tested yet) * feat: add process_escape function to handle escape sequences in prompts * feat: enhance LoRA weight handling in model loading and add text encoder loading function * feat: improve ComfyUI conversion script with prefix constants and module name adjustments * feat: update caption dropout documentation to clarify cache regeneration requirement * feat: add clarification on learning rate adjustments * feat: add note on PyTorch version requirement to prevent NaN loss --------- Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
feat: Anima support
Release 0.10.1
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 subscribe to this conversation on GitHub.
Already have an account?
Sign in.
2 participants
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.
See Commits and Changes for more details.
Created by
pull[bot] (v2.0.0-alpha.4)
Can you help keep this open source service alive? 💖 Please sponsor : )