v0.1.0-beta.1 · ace-step ACE-Step-1.5 · Discussion #436 · GitHub

ChuxiJ
Feb 11, 2026
Maintainer

What's Changed

Refact add inference by @ChuxiJ in Refact add inference #1
feat ✨ : add lyrics alignment scores by @keylxiao in feat ✨ : add lyrics alignment scores #2
Fix lrc bugs by @ChuxiJ in Fix lrc bugs #3
cover/repaint test ok by @ChuxiJ in cover/repaint test ok #4
refact understand_music by @ChuxiJ in refact understand_music #5
test ok by @ChuxiJ in test ok #6
Add rewrite lyrics by @ChuxiJ in Add rewrite lyrics #7
support input timesteps by @ChuxiJ in support input timesteps #8
max_model_len 8192 -> 4096 by @ChuxiJ in max_model_len 8192 -> 4096 #9
reduce vRAM usage for vae decode by @ChuxiJ in reduce vRAM usage for vae decode #10
fix: 修复 offload 模式下的设备不匹配错误 by @DumoeDss in fix: 修复 offload 模式下的设备不匹配错误 #11
fix language issue simple mode by @ChuxiJ in fix language issue simple mode #12
support lora trianing & inter by @ChuxiJ in support lora trianing & inter #13
fix cover logic for api_server by @ChuxiJ in fix cover logic for api_server #14
add docs and serve mode by @ChuxiJ in add docs and serve mode #15
add docs and readme by @ChuxiJ in add docs and readme #16
add model zoo by @ChuxiJ in add model zoo #17
fix audio play position reset by @ChuxiJ in fix audio play position reset #18
update hf demo by @ChuxiJ in update hf demo #19
fix duration str bug by @ChuxiJ in fix duration str bug #20
fix vllm bug by @ChuxiJ in fix vllm bug #21
Update abstract with specific performance metrics and unified messaging by @Copilot in Update abstract with specific performance metrics and unified messaging #22
Fix author names in BibTeX citation by @ChuxiJ in Fix author names in BibTeX citation #23
Verify author name order in BibTeX citation by @Copilot in Verify author name order in BibTeX citation #24
fix mac nv install by @ChuxiJ in fix mac nv install #25
feat: update skills by @DumoeDss in feat: update skills #26
improve lora training by @ChuxiJ in improve lora training #28
fix lora reload by @ChuxiJ in fix lora reload #29
add model downloader by @ChuxiJ in add model downloader #30
add model downloader by @ChuxiJ in add model downloader #31
openrouter compatible by @seaniezhao in openrouter compatible #32
fix duration by @ChuxiJ in fix duration #33
fix cfg kv block allocate by @ChuxiJ in fix cfg kv block allocate #34
fix_tiled_vae_encode_bug by @ChuxiJ in fix_tiled_vae_encode_bug #35
fix refer audio shape by @ChuxiJ in fix refer audio shape #36
fix 4 vram by @ChuxiJ in fix 4 vram #38
tutorial by @ChuxiJ in tutorial #39
Fix TypeError: AceStepConditionGenerationModel does not support len() by @saltchicken in Fix TypeError: AceStepConditionGenerationModel does not support len() #53
Fix runtime crash with integer seeds and correct type hints by @iamgrootns in Fix runtime crash with integer seeds and correct type hints #48
Fix torch.compile crash: add len to dynamically loaded models by @Copilot in Fix torch.compile crash: add __len__ to dynamically loaded models #47
Update Readme.md by @Saganaki22 in Update Readme.md #62
fix missing torchao by @ChuxiJ in fix missing torchao #65
Update release status for acestep-5Hz-lm-4B by @ChuxiJ in Update release status for acestep-5Hz-lm-4B #71
fix: reset linux torch version for nano vllm by @DumoeDss in fix: reset linux torch version for nano vllm #72
fix max audio code id by @ChuxiJ in fix max audio code id #78
Revert "fix max audio code id" by @ChuxiJ in Revert "fix max audio code id" #81
add support for intel gpu, tested on U9 285H, with and without 5hz llm by @xushengyuan in add support for intel gpu, tested on U9 285H, with and without 5hz llm #88
Fix max audio code by @ChuxiJ in Fix max audio code #103
Fix AttributeError when llm_handler is None by @fspecii in Fix AttributeError when llm_handler is None #109
Fix brocken HuggingFace link in README by @oumaklaus in Fix brocken HuggingFace link in README #118
Optimize env installation by @DumoeDss in Optimize env installation #128
Fix a Windows error that was unexpected at this time. (Fixes . was unexpected at this time. When trying to run both start .bat files #134) by @start-life in Fix a Windows error that was unexpected at this time. (Fixes #134) #137
fix: bat error by @DumoeDss in fix: bat error #148
feat: Add the enforced use of lm models by @DumoeDss in feat: Add the enforced use of lm models #152
Add tensorboard to training dependencies for LoRA by @tlennon-ie in Add tensorboard to training dependencies for LoRA #155
fix: Windows path compatibility and UI freeze issues (Issue Gradio UI Hang/Freeze After Generations + Audio Not Loading (tested only in Windows) Solved #113) by @DumoeDss in fix: Windows path compatibility and UI freeze issues (Issue #113) #159
docs: add Korean translation for core documentation by @acidsound in docs: add Korean translation for core documentation #187
feat(i18n): add translation support to LoRA training UI by @jayvenn21 in feat(i18n): add translation support to LoRA training UI #205
Adding Hebrew language by @start-life in Adding Hebrew language #201
docs: clarify ROCm / AMD usage and uv behavior by @jayvenn21 in docs: clarify ROCm / AMD usage and uv behavior #179
fix: avoid epoch-boundary stalls during LoRA training on Windows by @jayvenn21 in fix: avoid epoch-boundary stalls during LoRA training on Windows #177
fix path traversal vulnerability in audio endpoint by @Albab-Hasan in fix path traversal vulnerability in audio endpoint #161
fix: guard against missing reference audio in text2music by @jayvenn21 in fix: guard against missing reference audio in text2music #171
docs: add responsible disclosure security policy by @jayvenn21 in docs: add responsible disclosure security policy #169
fix: disable CUDA graph capture for 5Hz LM during LoRA training by @jayvenn21 in fix: disable CUDA graph capture for 5Hz LM during LoRA training #172
CLI by @chigkim in CLI #94
fix: add tolerance to 16GB VRAM detection by @jayvenn21 in fix: add tolerance to 16GB VRAM detection #175
fix: Wrap audio paths in FileData for Gradio 6.x compatibility by @DemetrionWare in fix: Wrap audio paths in FileData for Gradio 6.x compatibility #181
feat: expose denoise control for audio-conditioned generation in Gradio by @jayvenn21 in feat: expose denoise control for audio-conditioned generation in Gradio #176
Extended API to allow Analysis only without generating new music. by @goedzo in Extended API to allow Analysis only without generating new music. #197
Fix Gradio Audio component error by passing file paths directly by @Copilot in Fix Gradio Audio component error by passing file paths directly #219
Fix MPS tiled_decode conv1d output length limit on macOS 14.7 by @Copilot in Fix MPS tiled_decode conv1d output length limit on macOS 14.7 #222
Fix 4B model support for 16GB GPUs by increasing VRAM detection tolerance by @Copilot in Fix 4B model support for 16GB GPUs by increasing VRAM detection tolerance #221
fix: use per-socket timeout instead of global setdefaulttimeout by @Albab-Hasan in fix: use per-socket timeout instead of global setdefaulttimeout #226
Fix bugs and add debuglog by @ChuxiJ in Fix bugs and add debuglog #230
Add musician-friendly guide to documentation by @sigalarm in Add musician-friendly guide to documentation #238
Position Audio time labels lower to avoid scrollbar overlap by @rkfg in Position Audio time labels lower to avoid scrollbar overlap #234
Feat/progress UI: Prevent Gradio slider math error when no audio files are found by @1larity in Feat/progress UI: Prevent Gradio slider math error when no audio files are found #194
Include torchcodec on MacOS to read audio files. by @chigkim in Include torchcodec on MacOS to read audio files. #243
docs: clarify AMD / ROCm usage and uv behavior by @jayvenn21 in docs: clarify AMD / ROCm usage and uv behavior #244
fix(song): ensure lyrics / vocal conditioning is preserved during song generation by @jayvenn21 in fix(song): ensure lyrics / vocal conditioning is preserved during song generation #241
fix: resolve intermittent CUDA assertion error in concurrent serving … by @ChuxiJ in fix: resolve intermittent CUDA assertion error in concurrent serving … #257
fix: remove nano vllm install in toml by @ChuxiJ in fix: remove nano vllm install in toml #260
feat(ui): add experimental Studio UI for REST API by @jayvenn21 in feat(ui): add experimental Studio UI for REST API #258
feat(mps): Comprehensive Apple Silicon MPS backend support by @tonyjohnvan in feat(mps): Comprehensive Apple Silicon MPS backend support #256
fix(mps): Address memory management and performance issues for Apple … by @ChuxiJ in fix(mps): Address memory management and performance issues for Apple … #261
feat: Add AMD ROCm support for Windows (RX 7000/6000 series) by @clowerweb in feat: Add AMD ROCm support for Windows (RX 7000/6000 series) #210
[WIP] Fix uv sync regression with CUDA version mismatches by @Copilot in [WIP] Fix uv sync regression with CUDA version mismatches #268
fix: add anti-clipping normalization to prevent audio clipping/overload by @ChuxiJ in fix: add anti-clipping normalization to prevent audio clipping/overload #276
fix: validate training-incompatible settings before starting LoRA training by @ChuxiJ in fix: validate training-incompatible settings before starting LoRA training #277
Fix check_update.bat: handle line ending changes and remove pager blocking by @Copilot in Fix check_update.bat: handle line ending changes and remove pager blocking #273
Add manual for rocm Linux (Cachy-OS) and additional requirements file by @Neresco in Add manual for rocm Linux (Cachy-OS) and additional requirements file #271
Added advanced Auto-Label functionality and Progress Logging via API by @goedzo in Added advanced Auto-Label functionality and Progress Logging via API #223
Comprehensive debugger by @1larity in Comprehensive debugger #281
fix(ui): stop active audio playback when starting a new generation by @jayvenn21 in fix(ui): stop active audio playback when starting a new generation #288
Remove std-based audio normalization causing distortion artifacts by @Copilot in Remove std-based audio normalization causing distortion artifacts #299
Add tensorboard dependency to pyproject.toml by @Copilot in Add tensorboard dependency to pyproject.toml #300
Fix NameError in auto_label_all function by @Copilot in Fix NameError in auto_label_all function #301
Bring back generated audio codes by @rkfg in Bring back generated audio codes #291
docs: add LoRA training performance notes (Windows / low VRAM) by @jayvenn21 in docs: add LoRA training performance notes (Windows / low VRAM) #285
fix(linux): warn on pre-release Python when using vLLM backend by @jayvenn21 in fix(linux): warn on pre-release Python when using vLLM backend #284
Add diagnostic messaging for AMD GPU detection failures by @Copilot in Add diagnostic messaging for AMD GPU detection failures #302
fix: detect and surface silent audio generation failures by @jayvenn21 in fix: detect and surface silent audio generation failures #306
fix(rocm): add CPU VAE decode workaround for MIOpen workspace failures by @ElysiumDisc in fix(rocm): add CPU VAE decode workaround for MIOpen workspace failures #310
Hebrew language update also for the LORA interface by @start-life in Hebrew language update also for the LORA interface #312
CLI Uses settings from gpu_config.py. by @chigkim in CLI Uses settings from gpu_config.py. #289
Tensor preprocessing & Lora fix by @1larity in Tensor preprocessing & Lora fix #309
Dev/api update by @DumoeDss in Dev/api update #315
docs: enhance module documentation and code comments by @oumaklaus in docs: enhance module documentation and code comments #206
Docs/cpu support clarification by @jayvenn21 in Docs/cpu support clarification #170
feat(MLX) MLX Native Inference Acceleration by @tonyjohnvan in feat(MLX) MLX Native Inference Acceleration #295
CLI usage in readme by @chigkim in CLI usage in readme #317
Fix NameError in training exception handler by @Copilot in Fix NameError in training exception handler #316
fix: detect and error on NaN latents to prevent silent audio output by @jayvenn21 in fix: detect and error on NaN latents to prevent silent audio output #318
fix: prevent arbitrary code execution via unsafe torch.load() by @Albab-Hasan in fix: prevent arbitrary code execution via unsafe torch.load() #319
Fix LoRA scale slider modifying non-LoRA modules when adapter disabled by @Copilot in Fix LoRA scale slider modifying non-LoRA modules when adapter disabled #320
Merge mainline commits as of 2026/02/08 05:16 UTC with MPS optimizations and do optimization checks by @riversedge in Merge mainline commits as of 2026/02/08 05:16 UTC with MPS optimizations and do optimization checks #322
Enhance LoRA scale functionality to prevent unintended modifications … by @ChuxiJ in Enhance LoRA scale functionality to prevent unintended modifications … #324
Fix return value mismatch in handle_create_sample when LLM uninitialized by @Copilot in Fix return value mismatch in handle_create_sample when LLM uninitialized #328
fix: implement platform-specific audio playback reset logic by @ChuxiJ in fix: implement platform-specific audio playback reset logic #329
Revert "fix: implement platform-specific audio playback reset logic" by @ChuxiJ in Revert "fix: implement platform-specific audio playback reset logic" #331
Revert previous audio playback reset logic changes to restore origina… by @ChuxiJ in Revert previous audio playback reset logic changes to restore origina… #336
Fix AttributeError when stderr lacks isatty() during API server init by @Copilot in Fix AttributeError when stderr lacks isatty() during API server init #337
fix: clamp batch_size to GPU limit when loading metadata by @ChuxiJ in fix: clamp batch_size to GPU limit when loading metadata #339
Update dtype handling for MPS and enhance MLX model loading process by @ChuxiJ in Update dtype handling for MPS and enhance MLX model loading process #341
fix path traversal vulnerability in audio file path parameters by @Albab-Hasan in fix path traversal vulnerability in audio file path parameters #343
Update PEFT to >=0.18.0 for TorchAO compatibility by @Copilot in Update PEFT to >=0.18.0 for TorchAO compatibility #347
fix: avoid calling Module.to() on AffineQuantizedTensor parameters by @jayvenn21 in fix: avoid calling Module.to() on AffineQuantizedTensor parameters #354
fix: preserve user-provided seed and inference steps through generation pipeline by @jayvenn21 in fix: preserve user-provided seed and inference steps through generation pipeline #355
Prevent LoRA loading on quantized models to avoid TorchaoLoraLinear initialization error by @Copilot in Prevent LoRA loading on quantized models to avoid TorchaoLoraLinear initialization error #349
Critical: Fix crash on empty config_path and improve model selection messaging by @1larity in Critical: Fix crash on empty config_path and improve model selection messaging #350
Cli mlx by @chigkim in Cli mlx #362
Externalize hardcoded training UI strings to i18n JSON files by @Copilot in Externalize hardcoded training UI strings to i18n JSON files #364
Pin torchao to 0.12.0 for torch 2.7.1+cu128 compatibility by @Copilot in Pin torchao to 0.12.0 for torch 2.7.1+cu128 compatibility #363
Revert "Pin torchao to 0.12.0 for torch 2.7.1+cu128 compatibility" by @ChuxiJ in Revert "Pin torchao to 0.12.0 for torch 2.7.1+cu128 compatibility" #368
fix: avoid calling Module.to() on AffineQuantizedTensor parameters by @jayvenn21 in fix: avoid calling Module.to() on AffineQuantizedTensor parameters #358
fix: add KV cache headroom to gpu_memory_utilization calculation by @ChuxiJ in fix: add KV cache headroom to gpu_memory_utilization calculation #376
Feat/mlx native generation by @ChuxiJ in Feat/mlx native generation #381
Enhance ACE-Step Inference Profiler with new features and improved do… by @ChuxiJ in Enhance ACE-Step Inference Profiler with new features and improved do… #382
refactor: remove device type checks for MPS in AceStepHandler by @ChuxiJ in refactor: remove device type checks for MPS in AceStepHandler #392
refactor: update audio uploads accordion behavior in results section by @ChuxiJ in refactor: update audio uploads accordion behavior in results section #393
Update ACE-Step Inference Profiler with new features and improved doc… by @ChuxiJ in Update ACE-Step Inference Profiler with new features and improved doc… #394
Start script by @DumoeDss in Start script #397
fix: after gpu optimize not to download 0.6B lm by @DumoeDss in fix: after gpu optimize not to download 0.6B lm #410
feat: add fallback to get duration by @DumoeDss in feat: add fallback to get duration #411
fix: service_mode check by @DumoeDss in fix: service_mode check #414
fix: check_update.bat by @DumoeDss in fix: check_update.bat #418
(fix)(API backend): MPS VAE decode crash — reduce tiled_decode chunk size to avoid Conv1d output limit by @tonyjohnvan in (fix)(API backend): MPS VAE decode crash — reduce tiled_decode chunk size to avoid Conv1d output limit #416
Chore: Add domain package skeleton for incremental functional decomposition by @1larity in Chore: Add domain package skeleton for incremental functional decomposition #419
Create CONTRIBUTING.md with contribution guidelines by @1larity in Create CONTRIBUTING.md with contribution guidelines #423
Refactor LoRA handling into LoraService and fix low-scale UI regressions by @1larity in Refactor LoRA handling into LoraService and fix low-scale UI regressions #431
feat: GPU compatibility tier system with boundary testing by @ChuxiJ in feat: GPU compatibility tier system with boundary testing #417

New Contributors

@ChuxiJ made their first contribution in Refact add inference #1
@keylxiao made their first contribution in feat ✨ : add lyrics alignment scores #2
@DumoeDss made their first contribution in fix: 修复 offload 模式下的设备不匹配错误 #11
@Copilot made their first contribution in Update abstract with specific performance metrics and unified messaging #22
@seaniezhao made their first contribution in openrouter compatible #32
@saltchicken made their first contribution in Fix TypeError: AceStepConditionGenerationModel does not support len() #53
@iamgrootns made their first contribution in Fix runtime crash with integer seeds and correct type hints #48
@Saganaki22 made their first contribution in Update Readme.md #62
@xushengyuan made their first contribution in add support for intel gpu, tested on U9 285H, with and without 5hz llm #88
@fspecii made their first contribution in Fix AttributeError when llm_handler is None #109
@oumaklaus made their first contribution in Fix brocken HuggingFace link in README #118
@start-life made their first contribution in Fix a Windows error that was unexpected at this time. (Fixes #134) #137
@tlennon-ie made their first contribution in Add tensorboard to training dependencies for LoRA #155
@acidsound made their first contribution in docs: add Korean translation for core documentation #187
@jayvenn21 made their first contribution in feat(i18n): add translation support to LoRA training UI #205
@Albab-Hasan made their first contribution in fix path traversal vulnerability in audio endpoint #161
@chigkim made their first contribution in CLI #94
@DemetrionWare made their first contribution in fix: Wrap audio paths in FileData for Gradio 6.x compatibility #181
@goedzo made their first contribution in Extended API to allow Analysis only without generating new music. #197
@sigalarm made their first contribution in Add musician-friendly guide to documentation #238
@rkfg made their first contribution in Position Audio time labels lower to avoid scrollbar overlap #234
@1larity made their first contribution in Feat/progress UI: Prevent Gradio slider math error when no audio files are found #194
@tonyjohnvan made their first contribution in feat(mps): Comprehensive Apple Silicon MPS backend support #256
@clowerweb made their first contribution in feat: Add AMD ROCm support for Windows (RX 7000/6000 series) #210
@Neresco made their first contribution in Add manual for rocm Linux (Cachy-OS) and additional requirements file #271
@ElysiumDisc made their first contribution in fix(rocm): add CPU VAE decode workaround for MIOpen workspace failures #310
@riversedge made their first contribution in Merge mainline commits as of 2026/02/08 05:16 UTC with MPS optimizations and do optimization checks #322

Full Changelog: https://github.com/ace-step/ACE-Step-1.5/commits/v0.1.0-beta.1

This discussion was created from the release v0.1.0-beta.1.

Replies: 1 comment

Nurb4000
Feb 13, 2026

Thank you for the improvements.

0 replies

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment