[Request] Parsed ggml-turbo/request0.json [Request] seed=42 lm_batch=1 synth_batch=1 [Request] caption: An upbeat and anthemic pop-rock track driven by bright, slig... [Request] lyrics: 490 bytes [Request] bpm=83 dur=88 key=G major ts=4 lang=fr [Request] lm: temp=0.85 cfg=2.0 top_p=0.90 top_k=0 [Request] dit: steps=8 guidance=0.0 shift=3.0 [Request] task_type: text2music [Request] solver: euler (stork_substeps=10) [Request] lm_mode: generate [Request] output_format: mp3 [Request] synth_model: acestep-v15-turbo-Q8_0.gguf [Request] vae: vae-BF16.gguf [Request] audio_codes: (present) [Ace-Synth] Batch: 1 request(s) [Registry] Qwen3-Embedding-0.6B-BF16.gguf -> Text-Enc [Registry] Qwen3-Embedding-0.6B-Q8_0.gguf -> Text-Enc [Registry] acestep-5Hz-lm-0.6B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-0.6B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-4B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q5_K_M.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q6_K.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q8_0.gguf -> LM [Registry] acestep-v15-base-BF16.gguf -> DiT [Registry] acestep-v15-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-base-Q6_K.gguf -> DiT [Registry] acestep-v15-base-Q8_0.gguf -> DiT [Registry] acestep-v15-sft-BF16.gguf -> DiT [Registry] acestep-v15-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-BF16.gguf -> DiT [Registry] acestep-v15-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-continuous-BF16.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift1-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift3-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-base-BF16.gguf -> DiT [Registry] acestep-v15-xl-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-base-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sft-BF16.gguf -> DiT [Registry] acestep-v15-xl-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-turbo-BF16.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q8_0.gguf -> DiT [Registry] scragvae-BF16.gguf -> VAE [Registry] vae-BF16.gguf -> VAE [Store] Created (policy=STRICT) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q8_0.gguf: 678 tensors, data at offset 56864 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q8_0.gguf: 678 tensors, data at offset 56864 [Synth-Load] Ready: turbo=yes, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=8, guidance=1.0, shift=3.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/Qwen3-Embedding-0.6B-BF16.gguf: 310 tensors, data at offset 5337568 [Load] TextEncoder: 28L, H=1024, Nh=16/8 [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 310 tensors, 1136.5 MB into backend [Store] Load TextEnc: 319 ms [Debug] text_hidden: [70, 1024] first4: 3.705836 2.395382 0.221845 -13.145830 [Debug] lyric_embed: [167, 1024] first4: 0.029175 0.032227 -0.022339 -0.028809 [Store] Unload TextEnc (1136.5 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] CondEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q8_0.gguf: 678 tensors, data at offset 56864 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 140 tensors, 616.6 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L), text_proj, null_cond [Store] Load CondEnc: 328 ms [CondEnc] Lyric sliding mask: 167x167, window=128 [CondEnc] Timbre sliding mask: 1x1, window=128 [Encode] Packed: lyric=167 + timbre=1 + text=70 = 238 tokens [Encode-Text Batch0] 70+167 tokens -> enc_S=238, 18.5 ms [Debug] enc_hidden: [238, 2048] first4: 1.758854 -0.049784 -0.133323 0.058382 [Store] Unload CondEnc (616.6 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] Detokenizer backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q8_0.gguf: 678 tensors, data at offset 56864 [WeightCtx] Loaded 30 tensors, 106.5 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 183 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 110.2 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.124812 1.441172 0.303914 -0.631591 [Store] Unload FSQ-Detok (106.5 MB) [Init-Noise Batch0] Philox noise seed=42, [2170, 64] solver=euler [Debug] noise: [2170, 64] first4: 0.194336 2.156250 -0.171875 0.847656 [Debug] context: [2170, 128] first4: -0.124812 1.441172 0.303914 -0.631591 [Init-Noise] Starting: T=2170, S=1085, enc_S=238, steps=8, batch=1 (cover) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] DiT backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q8_0.gguf: 678 tensors, data at offset 56864 [DiT] Self-attn: Q+K+V fused [DiT] Cross-attn: Q+K+V fused [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 478 tensors, 1600.7 MB into backend [Load] DiT: 24 layers, H=2048, Nh=16/8, D=128 [Store] Load DiT: 904 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2129 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [12288] first4: 0.260124 -0.161873 -0.097043 0.052039 [Debug] temb: [2048] first4: 0.000130 -0.132501 -0.035452 0.064788 [Debug] temb_t: [2048] first4: 0.001145 0.026826 -0.052770 0.063722 [Debug] temb_r: [2048] first4: -0.001015 -0.159327 0.017318 0.001066 [Debug] sinusoidal_t: [256] first4: 0.562486 0.789701 0.439822 -0.023583 [Debug] sinusoidal_r: [256] first4: 1.000000 1.000000 1.000000 1.000000 [Debug] temb_lin1_t: [2048] first4: -0.048950 -0.051683 -0.015299 -0.038721 [Debug] temb_lin1_r: [2048] first4: -0.013066 -0.018836 -0.015732 0.008463 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.038574 -0.958557 0.537598 0.446259 [Debug] proj_in_input: [192, 2170] first4: -0.124812 1.441172 0.303914 -0.631591 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.169418 0.817307 0.329163 -0.557957 [Debug] layer0_sa_input: [2048, 1085] first4: -0.718028 -0.757975 -0.047152 0.261293 [Debug] layer0_q_after_rope: [128, 16] first4: -1.541547 -1.031567 0.196720 0.455789 [Debug] layer0_k_after_rope: [128, 8] first4: -0.169418 0.817307 0.329163 -0.557957 [Debug] layer0_sa_output: [2048, 1085] first4: -1.507812 0.164551 -0.349121 0.512695 [Debug] layer0_attn_out: [2048, 1085] first4: -1.541547 -1.031567 0.196720 0.455789 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.541547 -1.031567 0.196720 0.455789 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.580853 -0.792920 -0.281795 0.496927 [Debug] hidden_after_layer0: [2048, 1085] first4: -8.927026 0.559497 51.180729 -0.873846 [Debug] hidden_after_layer6: [2048, 1085] first4: -20.899237 -0.194543 34.219177 -4.382227 [Debug] hidden_after_layer12: [2048, 1085] first4: -14.615265 -18.322664 73.964745 30.124640 [Debug] hidden_after_layer18: [2048, 1085] first4: -26.774414 14.169523 62.177979 19.658535 [Debug] hidden_after_layer23: [2048, 1085] first4: -2.562471 39.824661 192.395004 144.278229 [Debug] dit_step0_vt: [2170, 64] first4: 0.032806 1.118179 0.345366 2.341881 [Debug] dit_step0_xt: [2170, 64] first4: 0.192845 2.105424 -0.187573 0.741207 [DiT] Step 1/8 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.180931 1.297287 -0.116950 1.917435 [Debug] dit_step1_xt: [2170, 64] first4: 0.202714 2.034662 -0.181194 0.636620 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: -0.016251 1.219034 0.094940 2.367386 [Debug] dit_step2_xt: [2170, 64] first4: 0.203797 1.953394 -0.187524 0.478794 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: 0.248707 1.073059 0.253395 2.622116 [Debug] dit_step3_xt: [2170, 64] first4: 0.183072 1.863972 -0.208640 0.260284 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: 0.290359 1.013794 0.116058 2.692276 [Debug] dit_step4_xt: [2170, 64] first4: 0.151962 1.755351 -0.221075 -0.028174 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: 0.265259 0.923615 -0.240723 2.716354 [Debug] dit_step5_xt: [2170, 64] first4: 0.114067 1.623406 -0.186686 -0.416224 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: 0.161392 0.644836 -0.735420 2.781006 [Debug] dit_step6_xt: [2170, 64] first4: 0.081789 1.494439 -0.039602 -0.972426 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: -0.016296 0.178223 -1.388428 3.070557 [Debug] dit_x0: [2170, 64] first4: 0.086678 1.440972 0.376927 -1.893593 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 281.0 ms (281.0 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.086678 1.440972 0.376927 -1.893593 [Store] Unload DiT (1600.7 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: Vulkan0 (CPU threads: 16) [VAE] Backend: Vulkan0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 668 ms [VAE] Tiled decode: 3 tiles (chunk=1024, overlap=64, stride=896) [VAE] Graph: 479 nodes, T_latent=960 [VAE] Upsample factor: 1920.00 (expected ~1920) [VAE] Graph: 479 nodes, T_latent=1024 [VAE] Graph: 479 nodes, T_latent=442 [VAE] Tiled decode done: 3 tiles -> T_audio=4166400 (86.80s @ 48kHz) [VAE-Decode Batch0] Decode: 946.1 ms [Debug] vae_audio: [2, 4166400] first4: 0.000734 0.001445 0.001222 0.001748 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 450 ms (192.79x realtime), 32 threads [MP3] Wrote ggml-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 10:20:25.977 | DEBUG | acestep.core.generation.handler.init_service_memory_basic:_apply_malloc_mmap_threshold:50 - [memory] Set M_MMAP_THRESHOLD=131072 for immediate OS reclaim of large frees Skipping import of cpp extensions due to incompatible torch version. Please upgrade to torch >= 2.11.0 (found 2.10.0+cu128). 2026-07-12 10:20:28.007 | WARNING | acestep.training.trainer::40 - bitsandbytes not installed. Using standard AdamW. 2026-07-12 10:20:28.012 | INFO | acestep.gpu_config:get_gpu_memory_gb:578 - CUDA GPU detected: NVIDIA RTX PRO 6000 Blackwell Workstation Edition (95.0 GB) 2026-07-12 10:20:28.083 | INFO | acestep.core.generation.handler.init_service_loader:_load_main_model_from_checkpoint:174 - [initialize_service] Attempting to load model with attention implementation: sdpa Unable to import `torchao` Tensor objects. This may affect loading checkpoints serialized with `torchao` 2026-07-12 10:20:30.149 | INFO | acestep.core.generation.handler.generate_music:generate_music:289 - [generate_music] Turbo model detected: overriding guidance_scale 0.0 -> 1.0 (turbo does not use CFG). 2026-07-12 10:20:30.149 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:20:30.149 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:20:30.149 | INFO | acestep.core.generation.handler.generate_music_request:_prepare_reference_and_source_audio:179 - [generate_music] cover task: no src_audio but audio codes provided, proceeding with codes 2026-07-12 10:20:30.153 | INFO | acestep.core.generation.handler.generate_music:_vram_preflight_check:156 - [generate_music] VRAM pre-flight: 87.83 GB free, ~0.94 GB needed (batch=1, duration=88s, mode=turbo). 2026-07-12 10:20:30.157 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:20:30.376 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_precomputed_lm_hints:31 - [generate_music] Decoding audio codes for LM hints for item 0... 2026-07-12 10:20:30.378 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:20:30.378 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:20:30.378 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:20:30.378 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:122 - text_prompt: # Instruction Generate audio semantic tokens based on the given conditions: # Caption An upbeat and anthemic pop-rock track driven by bright, slightly overdriven # Metas - bpm: 83 - timesignature: 4 - keyscale: G major - duration: 88 seconds <|endoftext|> 2026-07-12 10:20:30.378 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:20:30.378 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:124 - lyrics_text: # Languages fr # Lyric # Lyric [Intro - Guitar Riff] [Verse 1] Dans le monde des tutos virtuels G ta toise en nouvelle passion Avec Ggendoline et PumbΓ© Γ  midi La communautΓ©, c'est l'unitΓ© Quel joie, une clΓ© [Chorus] Dans le monde des tutos virtuels GΓ’ndoline et PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner [Guitar Solo] [Verse 2] Dans le monde des tutos virtuels GΓ’ndoline, PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner<|endoftext|> 2026-07-12 10:20:30.378 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:20:30.385 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:20:30.400 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:20:30.400 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:20:30.423 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 10:20:30.431 | INFO | acestep.models.common.dcw_correction:__init__:98 - [DCW] Active β€” mode=double, scaler=0.0500, high_scaler=0.0200, wavelet='haar' 2026-07-12 10:20:30.520 | WARNING | acestep.models.common.dcw_loader:_try_import:39 - DCW is enabled but 'pytorch_wavelets' is not installed. Install with `pip install pytorch_wavelets PyWavelets` to use Differential Correction in Wavelet domain. Falling back to no-op for this generation. 2026-07-12 10:20:30.746 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 10:20:30.747 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:64 - [generate_music] pred_latents: torch.Size([1, 2170, 64]), dtype=torch.bfloat16 2026-07-12 10:20:30.747 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.006858110427856445, 'diffusion_time_cost': 0.3152453899383545, 'diffusion_per_step_time_cost': 0.03940567374229431, 'total_time_cost': 0.32210350036621094, 'offload_time_cost': 0.0} 2026-07-12 10:20:30.761 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 10:20:30.763 | DEBUG | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:136 - [generate_music] Before VAE decode: allocated=5.98GB, max=7.29GB 2026-07-12 10:20:30.763 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:154 - [generate_music] Effective free VRAM before VAE decode: 87.73 GB 2026-07-12 10:20:30.763 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:172 - [generate_music] Using tiled VAE decode to reduce VRAM usage... 2026-07-12 10:20:30.763 | DEBUG | acestep.core.generation.handler.memory_utils:_get_auto_decode_chunk_size:108 - [_get_auto_decode_chunk_size] Effective free VRAM: 87.73 GB 2026-07-12 10:20:30.764 | DEBUG | acestep.core.generation.handler.memory_utils:_should_offload_wav_to_cpu:131 - [_should_offload_wav_to_cpu] Effective free VRAM: 87.73 GB 2026-07-12 10:20:30.764 | INFO | acestep.core.generation.handler.vae_decode:tiled_decode:56 - [tiled_decode] chunk_size=512, offload_wav_to_cpu=False, latents_shape=torch.Size([1, 64, 2170]) 2026-07-12 10:20:31.050 | DEBUG | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:193 - [generate_music] After VAE decode: allocated=6.15GB, max=7.44GB 2026-07-12 10:20:31.215 | INFO | acestep.core.generation.handler.generate_music_payload:_build_generate_music_success_payload:38 - [generate_music] VAE decode completed. Preparing audio tensors... 2026-07-12 10:20:31.218 | INFO | acestep.core.generation.handler.generate_music_payload:_build_generate_music_success_payload:50 - [generate_music] Done! Generated 1 audio tensors. [Request] Parsed ggml-sft/request0.json [Request] seed=42 lm_batch=1 synth_batch=1 [Request] caption: An upbeat and anthemic pop-rock track driven by bright, slig... [Request] lyrics: 490 bytes [Request] bpm=83 dur=88 key=G major ts=4 lang=fr [Request] lm: temp=0.85 cfg=2.0 top_p=0.90 top_k=0 [Request] dit: steps=50 guidance=1.0 shift=1.0 [Request] task_type: text2music [Request] solver: euler (stork_substeps=10) [Request] lm_mode: generate [Request] output_format: mp3 [Request] synth_model: acestep-v15-sft-Q8_0.gguf [Request] vae: vae-BF16.gguf [Request] audio_codes: (present) [Ace-Synth] Batch: 1 request(s) [Registry] Qwen3-Embedding-0.6B-BF16.gguf -> Text-Enc [Registry] Qwen3-Embedding-0.6B-Q8_0.gguf -> Text-Enc [Registry] acestep-5Hz-lm-0.6B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-0.6B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-4B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q5_K_M.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q6_K.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q8_0.gguf -> LM [Registry] acestep-v15-base-BF16.gguf -> DiT [Registry] acestep-v15-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-base-Q6_K.gguf -> DiT [Registry] acestep-v15-base-Q8_0.gguf -> DiT [Registry] acestep-v15-sft-BF16.gguf -> DiT [Registry] acestep-v15-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-BF16.gguf -> DiT [Registry] acestep-v15-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-continuous-BF16.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift1-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift3-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-base-BF16.gguf -> DiT [Registry] acestep-v15-xl-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-base-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sft-BF16.gguf -> DiT [Registry] acestep-v15-xl-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-turbo-BF16.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q8_0.gguf -> DiT [Registry] scragvae-BF16.gguf -> VAE [Registry] vae-BF16.gguf -> VAE [Store] Created (policy=STRICT) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q8_0.gguf: 678 tensors, data at offset 56800 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q8_0.gguf: 678 tensors, data at offset 56800 [Synth-Load] Ready: turbo=no, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=50, guidance=1.0, shift=1.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/Qwen3-Embedding-0.6B-BF16.gguf: 310 tensors, data at offset 5337568 [Load] TextEncoder: 28L, H=1024, Nh=16/8 [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 310 tensors, 1136.5 MB into backend [Store] Load TextEnc: 331 ms [Debug] text_hidden: [70, 1024] first4: 3.705836 2.395382 0.221845 -13.145830 [Debug] lyric_embed: [167, 1024] first4: 0.029175 0.032227 -0.022339 -0.028809 [Store] Unload TextEnc (1136.5 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] CondEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q8_0.gguf: 678 tensors, data at offset 56800 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 140 tensors, 616.6 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L), text_proj, null_cond [Store] Load CondEnc: 311 ms [CondEnc] Lyric sliding mask: 167x167, window=128 [CondEnc] Timbre sliding mask: 1x1, window=128 [Encode] Packed: lyric=167 + timbre=1 + text=70 = 238 tokens [Encode-Text Batch0] 70+167 tokens -> enc_S=238, 7.4 ms [Debug] enc_hidden: [238, 2048] first4: 1.758854 -0.049784 -0.133323 0.058382 [Store] Unload CondEnc (616.6 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] Detokenizer backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q8_0.gguf: 678 tensors, data at offset 56800 [WeightCtx] Loaded 30 tensors, 106.5 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 175 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 106.3 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.124812 1.441172 0.303914 -0.631591 [Store] Unload FSQ-Detok (106.5 MB) [Init-Noise Batch0] Philox noise seed=42, [2170, 64] solver=euler [Debug] noise: [2170, 64] first4: 0.194336 2.156250 -0.171875 0.847656 [Debug] context: [2170, 128] first4: -0.124812 1.441172 0.303914 -0.631591 [Init-Noise] Starting: T=2170, S=1085, enc_S=238, steps=50, batch=1 (cover) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] DiT backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q8_0.gguf: 678 tensors, data at offset 56800 [DiT] Self-attn: Q+K+V fused [DiT] Cross-attn: Q+K+V fused [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 478 tensors, 1600.7 MB into backend [Load] DiT: 24 layers, H=2048, Nh=16/8, D=128 [Store] Load DiT: 1017 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2129 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [12288] first4: 0.154891 -0.116111 -0.086457 0.081949 [Debug] temb: [2048] first4: -0.002756 -0.176432 0.004178 -0.001982 [Debug] temb_t: [2048] first4: -0.001185 0.003330 -0.013113 -0.002073 [Debug] temb_r: [2048] first4: -0.001571 -0.179762 0.017291 0.000091 [Debug] sinusoidal_t: [256] first4: 0.562486 0.789701 0.439822 -0.023583 [Debug] sinusoidal_r: [256] first4: 1.000000 1.000000 1.000000 1.000000 [Debug] temb_lin1_t: [2048] first4: -0.038370 0.029874 0.028025 -0.024771 [Debug] temb_lin1_r: [2048] first4: 0.001863 0.003353 -0.000552 -0.000197 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.087891 -0.914795 0.499023 0.441193 [Debug] proj_in_input: [192, 2170] first4: -0.124812 1.441172 0.303914 -0.631591 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.187973 0.873795 0.324585 -0.581024 [Debug] layer0_sa_input: [2048, 1085] first4: -0.912651 -0.709809 -0.037078 0.294389 [Debug] layer0_q_after_rope: [128, 16] first4: -1.803765 -1.369850 -0.116876 0.380624 [Debug] layer0_k_after_rope: [128, 8] first4: -0.187973 0.873795 0.324585 -0.581024 [Debug] layer0_sa_output: [2048, 1085] first4: -1.653320 0.803711 -0.544922 0.540527 [Debug] layer0_attn_out: [2048, 1085] first4: -1.803765 -1.369850 -0.116876 0.380624 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.803765 -1.369850 -0.116876 0.380624 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.943901 -1.019508 -0.380060 0.396234 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.228189 1.154781 55.886856 -1.578429 [Debug] hidden_after_layer6: [2048, 1085] first4: -16.782993 4.341956 56.227089 -2.335757 [Debug] hidden_after_layer12: [2048, 1085] first4: -9.429165 3.932663 -22.469273 -1.937183 [Debug] hidden_after_layer18: [2048, 1085] first4: -3.689018 19.462234 -37.342949 -11.047457 [Debug] hidden_after_layer23: [2048, 1085] first4: 34.442764 63.225845 58.392715 -10.212424 [Debug] dit_step0_vt: [2170, 64] first4: -0.573853 2.509644 -0.187447 1.474609 [Debug] dit_step0_xt: [2170, 64] first4: 0.205813 2.106057 -0.168126 0.818164 [DiT] Step 1/50 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.630768 2.526855 -0.015015 1.292061 [Debug] dit_step1_xt: [2170, 64] first4: 0.218428 2.055520 -0.167826 0.792323 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.613953 2.482666 0.024719 1.317841 [Debug] dit_step2_xt: [2170, 64] first4: 0.230707 2.005867 -0.168320 0.765966 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.610291 2.425903 0.043091 1.332855 [Debug] dit_step3_xt: [2170, 64] first4: 0.242913 1.957349 -0.169182 0.739309 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.544312 2.335327 0.020508 1.322464 [Debug] dit_step4_xt: [2170, 64] first4: 0.253799 1.910642 -0.169592 0.712860 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.496094 2.267334 -0.001587 1.308929 [Debug] dit_step5_xt: [2170, 64] first4: 0.263721 1.865295 -0.169560 0.686681 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.470154 2.208374 -0.028130 1.298279 [Debug] dit_step6_xt: [2170, 64] first4: 0.273124 1.821128 -0.168998 0.660716 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.439331 2.149414 -0.046345 1.292648 [Debug] dit_step7_xt: [2170, 64] first4: 0.281911 1.778140 -0.168071 0.634863 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.380432 2.053345 -0.063217 1.270966 [Debug] dit_step8_xt: [2170, 64] first4: 0.289520 1.737073 -0.166807 0.609443 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.319519 1.967896 -0.068665 1.255386 [Debug] dit_step9_xt: [2170, 64] first4: 0.295910 1.697715 -0.165433 0.584336 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.256104 1.890137 -0.054184 1.251617 [Debug] dit_step10_xt: [2170, 64] first4: 0.301032 1.659912 -0.164350 0.559303 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.224487 1.823486 -0.048401 1.239807 [Debug] dit_step11_xt: [2170, 64] first4: 0.305522 1.623442 -0.163382 0.534507 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.204773 1.757446 -0.050980 1.237015 [Debug] dit_step12_xt: [2170, 64] first4: 0.309617 1.588294 -0.162362 0.509767 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.189209 1.681152 -0.068512 1.219315 [Debug] dit_step13_xt: [2170, 64] first4: 0.313402 1.554671 -0.160992 0.485381 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.174416 1.611115 -0.090393 1.202744 [Debug] dit_step14_xt: [2170, 64] first4: 0.316890 1.522448 -0.159184 0.461326 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.181337 1.528732 -0.119080 1.184258 [Debug] dit_step15_xt: [2170, 64] first4: 0.320517 1.491874 -0.156802 0.437640 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.184448 1.443542 -0.142670 1.175484 [Debug] dit_step16_xt: [2170, 64] first4: 0.324206 1.463003 -0.153949 0.414131 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.187683 1.358007 -0.166260 1.160290 [Debug] dit_step17_xt: [2170, 64] first4: 0.327959 1.435843 -0.150624 0.390925 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.205963 1.278725 -0.203186 1.135723 [Debug] dit_step18_xt: [2170, 64] first4: 0.332078 1.410268 -0.146560 0.368211 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.231659 1.204056 -0.221802 1.122818 [Debug] dit_step19_xt: [2170, 64] first4: 0.336712 1.386187 -0.142124 0.345754 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.248840 1.115814 -0.224060 1.105453 [Debug] dit_step20_xt: [2170, 64] first4: 0.341688 1.363871 -0.137643 0.323645 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.269928 1.022278 -0.246826 1.075836 [Debug] dit_step21_xt: [2170, 64] first4: 0.347087 1.343425 -0.132706 0.302128 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.292130 0.915527 -0.252441 1.050385 [Debug] dit_step22_xt: [2170, 64] first4: 0.352930 1.325115 -0.127657 0.281121 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.314835 0.825684 -0.254333 1.036804 [Debug] dit_step23_xt: [2170, 64] first4: 0.359226 1.308601 -0.122571 0.260385 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.341980 0.727478 -0.259399 1.002594 [Debug] dit_step24_xt: [2170, 64] first4: 0.366066 1.294051 -0.117383 0.240333 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.363785 0.604797 -0.271362 0.979736 [Debug] dit_step25_xt: [2170, 64] first4: 0.373342 1.281955 -0.111955 0.220738 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.386658 0.506104 -0.274658 0.970520 [Debug] dit_step26_xt: [2170, 64] first4: 0.381075 1.271833 -0.106462 0.201328 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.411407 0.389404 -0.252808 0.940857 [Debug] dit_step27_xt: [2170, 64] first4: 0.389303 1.264045 -0.101406 0.182510 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.434036 0.271103 -0.260681 0.917999 [Debug] dit_step28_xt: [2170, 64] first4: 0.397984 1.258623 -0.096193 0.164150 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.461563 0.156174 -0.265228 0.887085 [Debug] dit_step29_xt: [2170, 64] first4: 0.407215 1.255500 -0.090888 0.146409 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.478027 0.036282 -0.248230 0.875977 [Debug] dit_step30_xt: [2170, 64] first4: 0.416775 1.254774 -0.085923 0.128889 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.495369 -0.076294 -0.249451 0.853180 [Debug] dit_step31_xt: [2170, 64] first4: 0.426683 1.256300 -0.080934 0.111826 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.512451 -0.189911 -0.232727 0.832031 [Debug] dit_step32_xt: [2170, 64] first4: 0.436932 1.260098 -0.076280 0.095185 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.520683 -0.287262 -0.220444 0.815735 [Debug] dit_step33_xt: [2170, 64] first4: 0.447346 1.265844 -0.071871 0.078870 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.532341 -0.395935 -0.208969 0.791321 [Debug] dit_step34_xt: [2170, 64] first4: 0.457992 1.273762 -0.067692 0.063044 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.553373 -0.494385 -0.179688 0.761047 [Debug] dit_step35_xt: [2170, 64] first4: 0.469060 1.283650 -0.064098 0.047823 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.567871 -0.595657 -0.156029 0.728577 [Debug] dit_step36_xt: [2170, 64] first4: 0.480417 1.295563 -0.060977 0.033251 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.555824 -0.683998 -0.125435 0.684204 [Debug] dit_step37_xt: [2170, 64] first4: 0.491534 1.309243 -0.058469 0.019567 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.561726 -0.774433 -0.113373 0.651855 [Debug] dit_step38_xt: [2170, 64] first4: 0.502768 1.324732 -0.056201 0.006530 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.559219 -0.858842 -0.075325 0.602417 [Debug] dit_step39_xt: [2170, 64] first4: 0.513953 1.341909 -0.054695 -0.005518 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.554796 -0.931717 -0.045250 0.562897 [Debug] dit_step40_xt: [2170, 64] first4: 0.525049 1.360543 -0.053790 -0.016776 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.543407 -0.995499 0.003399 0.512360 [Debug] dit_step41_xt: [2170, 64] first4: 0.535917 1.380453 -0.053858 -0.027023 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.531044 -1.059387 0.045123 0.466324 [Debug] dit_step42_xt: [2170, 64] first4: 0.546538 1.401641 -0.054760 -0.036350 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.517281 -1.113323 0.086792 0.412628 [Debug] dit_step43_xt: [2170, 64] first4: 0.556883 1.423907 -0.056496 -0.044602 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.497894 -1.166992 0.133804 0.354736 [Debug] dit_step44_xt: [2170, 64] first4: 0.566841 1.447247 -0.059172 -0.051697 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.480843 -1.208649 0.165695 0.298584 [Debug] dit_step45_xt: [2170, 64] first4: 0.576458 1.471420 -0.062486 -0.057669 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.447495 -1.246643 0.205666 0.240643 [Debug] dit_step46_xt: [2170, 64] first4: 0.585408 1.496353 -0.066599 -0.062482 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.428513 -1.281120 0.222218 0.204704 [Debug] dit_step47_xt: [2170, 64] first4: 0.593978 1.521975 -0.071044 -0.066576 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.370972 -1.317520 0.233276 0.168076 [Debug] dit_step48_xt: [2170, 64] first4: 0.601397 1.548326 -0.075709 -0.069937 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.399719 -1.340591 0.222412 0.162598 [Debug] dit_x0: [2170, 64] first4: 0.609392 1.575138 -0.080157 -0.073189 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 1513.4 ms (1513.4 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.609392 1.575138 -0.080157 -0.073189 [Store] Unload DiT (1600.7 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: Vulkan0 (CPU threads: 16) [VAE] Backend: Vulkan0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 666 ms [VAE] Tiled decode: 3 tiles (chunk=1024, overlap=64, stride=896) [VAE] Graph: 479 nodes, T_latent=960 [VAE] Upsample factor: 1920.00 (expected ~1920) [VAE] Graph: 479 nodes, T_latent=1024 [VAE] Graph: 479 nodes, T_latent=442 [VAE] Tiled decode done: 3 tiles -> T_audio=4166400 (86.80s @ 48kHz) [VAE-Decode Batch0] Decode: 943.5 ms [Debug] vae_audio: [2, 4166400] first4: -0.001571 -0.001066 -0.001317 -0.000902 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 403 ms (215.26x realtime), 32 threads [MP3] Wrote ggml-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 10:20:37.805 | INFO | acestep.core.generation.handler.init_service_loader:_load_main_model_from_checkpoint:174 - [initialize_service] Attempting to load model with attention implementation: sdpa 2026-07-12 10:20:39.828 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:20:39.828 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:20:39.828 | INFO | acestep.core.generation.handler.generate_music_request:_prepare_reference_and_source_audio:179 - [generate_music] cover task: no src_audio but audio codes provided, proceeding with codes 2026-07-12 10:20:39.831 | INFO | acestep.core.generation.handler.generate_music:_vram_preflight_check:156 - [generate_music] VRAM pre-flight: 81.69 GB free, ~0.94 GB needed (batch=1, duration=88s, mode=turbo). 2026-07-12 10:20:39.835 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:20:39.840 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_precomputed_lm_hints:31 - [generate_music] Decoding audio codes for LM hints for item 0... 2026-07-12 10:20:39.841 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:20:39.841 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:20:39.841 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:20:39.841 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:122 - text_prompt: # Instruction Generate audio semantic tokens based on the given conditions: # Caption An upbeat and anthemic pop-rock track driven by bright, slightly overdriven # Metas - bpm: 83 - timesignature: 4 - keyscale: G major - duration: 88 seconds <|endoftext|> 2026-07-12 10:20:39.841 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:20:39.841 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:124 - lyrics_text: # Languages fr # Lyric # Lyric [Intro - Guitar Riff] [Verse 1] Dans le monde des tutos virtuels G ta toise en nouvelle passion Avec Ggendoline et PumbΓ© Γ  midi La communautΓ©, c'est l'unitΓ© Quel joie, une clΓ© [Chorus] Dans le monde des tutos virtuels GΓ’ndoline et PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner [Guitar Solo] [Verse 2] Dans le monde des tutos virtuels GΓ’ndoline, PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner<|endoftext|> 2026-07-12 10:20:39.841 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:20:39.845 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:20:39.853 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:20:39.853 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:20:39.860 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... [Request] Loaded request0.json [Turbo] steps=8, shift=3.0 | acestep-v15-turbo-Q8_0.gguf [GGML] Running acestep-v15-turbo-Q8_0.gguf... [GGML] Done, 47 dump files [Python] Initializing acestep-v15-turbo... [Python] Generating (acestep-v15-turbo, 8 steps)... Using precomputed LM hints Using precomputed LM hints [Python] Wrote python-turbo/output.wav: 4166400 samples (86.80s @ 48kHz stereo) [Python] Done, 40 dump files [Turbo] Cosine similarities GGML vs Python stage GGML vs Python text_hidden 0.999817 lyric_embed 1.000000 enc_hidden 0.999829 detok_output 0.999984 context 0.999990 noise 1.000000 temb_t 0.999998 hidden_after_proj_in 0.999985 enc_after_cond_emb 0.999821 layer0_sa_output 0.999939 hidden_after_layer0 0.999870 hidden_after_layer6 0.999895 hidden_after_layer12 0.999049 hidden_after_layer18 0.996339 hidden_after_layer23 0.993108 dit_step0_vt 0.973571 dit_step0_xt 0.999942 dit_step1_vt 0.974901 dit_step1_xt 0.999810 dit_step2_vt 0.976260 dit_step2_xt 0.999472 dit_step3_vt 0.978766 dit_step3_xt 0.998743 dit_step4_vt 0.977913 dit_step4_xt 0.997082 dit_step5_vt 0.975598 dit_step5_xt 0.993419 dit_step6_vt 0.972741 dit_step6_xt 0.986596 dit_step7_vt 0.966413 dit_x0 0.976686 vae_audio 0.896810 vae_audio (STFT cosine) 0.973765 [Turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999942 0.139248 0.006967 -0.002321 0.973049 -0.002342 0.972003 dit_step1_xt 0.999810 0.271278 0.012128 -0.005311 0.942916 -0.005313 0.941730 dit_step2_xt 0.999472 0.460709 0.019040 -0.009339 0.909613 -0.009311 0.908527 dit_step3_xt 0.998743 0.647982 0.027970 -0.014694 0.874467 -0.014577 0.873624 dit_step4_xt 0.997082 0.961244 0.040561 -0.021703 0.842862 -0.021660 0.841995 dit_step5_xt 0.993419 1.431425 0.059125 -0.031735 0.826410 -0.032109 0.824593 dit_step6_xt 0.986596 2.159874 0.087188 -0.045914 0.857474 -0.046482 0.855546 [SFT] steps=50, shift=1.0 | acestep-v15-sft-Q8_0.gguf [GGML] Running acestep-v15-sft-Q8_0.gguf... [GGML] Done, 131 dump files [Python] Initializing acestep-v15-sft... [Python] Generating (acestep-v15-sft, 50 steps)... Using precomputed LM hints Using precomputed LM hints 0%| | 0/50 [00:00 Text-Enc [Registry] Qwen3-Embedding-0.6B-Q8_0.gguf -> Text-Enc [Registry] acestep-5Hz-lm-0.6B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-0.6B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-4B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q5_K_M.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q6_K.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q8_0.gguf -> LM [Registry] acestep-v15-base-BF16.gguf -> DiT [Registry] acestep-v15-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-base-Q6_K.gguf -> DiT [Registry] acestep-v15-base-Q8_0.gguf -> DiT [Registry] acestep-v15-sft-BF16.gguf -> DiT [Registry] acestep-v15-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-BF16.gguf -> DiT [Registry] acestep-v15-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-continuous-BF16.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift1-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift3-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-base-BF16.gguf -> DiT [Registry] acestep-v15-xl-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-base-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sft-BF16.gguf -> DiT [Registry] acestep-v15-xl-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-turbo-BF16.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q8_0.gguf -> DiT [Registry] scragvae-BF16.gguf -> VAE [Registry] vae-BF16.gguf -> VAE [Store] Created (policy=STRICT) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q8_0.gguf: 830 tensors, data at offset 69088 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q8_0.gguf: 830 tensors, data at offset 69088 [Synth-Load] Ready: turbo=yes, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=8, guidance=1.0, shift=3.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/Qwen3-Embedding-0.6B-BF16.gguf: 310 tensors, data at offset 5337568 [Load] TextEncoder: 28L, H=1024, Nh=16/8 [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 310 tensors, 1136.5 MB into backend [Store] Load TextEnc: 498 ms [Debug] text_hidden: [70, 1024] first4: 3.705836 2.395382 0.221845 -13.145830 [Debug] lyric_embed: [167, 1024] first4: 0.029175 0.032227 -0.022339 -0.028809 [Store] Unload TextEnc (1136.5 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] CondEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q8_0.gguf: 830 tensors, data at offset 69088 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 141 tensors, 616.6 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L, CLS), text_proj, null_cond [Store] Load CondEnc: 313 ms [CondEnc] Lyric sliding mask: 167x167, window=128 [CondEnc] Timbre sliding mask: 2x2, window=128 (CLS) [Encode] Packed: lyric=167 + timbre=1 + text=70 = 238 tokens [Encode-Text Batch0] 70+167 tokens -> enc_S=238, 7.3 ms [Debug] enc_hidden: [238, 2048] first4: 0.686671 -0.019193 -0.051889 0.022865 [Store] Unload CondEnc (616.6 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] Detokenizer backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q8_0.gguf: 830 tensors, data at offset 69088 [WeightCtx] Loaded 30 tensors, 106.5 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 174 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 106.3 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.124812 1.441172 0.303914 -0.631591 [Store] Unload FSQ-Detok (106.5 MB) [Init-Noise Batch0] Philox noise seed=42, [2170, 64] solver=euler [Debug] noise: [2170, 64] first4: 0.194336 2.156250 -0.171875 0.847656 [Debug] context: [2170, 128] first4: -0.124812 1.441172 0.303914 -0.631591 [Init-Noise] Starting: T=2170, S=1085, enc_S=238, steps=8, batch=1 (cover) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] DiT backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q8_0.gguf: 830 tensors, data at offset 69088 [DiT] Self-attn: Q+K+V fused [DiT] Cross-attn: Q+K+V fused [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 630 tensors, 4230.2 MB into backend [Load] DiT: 32 layers, H=2560, Nh=32/8, D=128 [Store] Load DiT: 2347 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2825 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [15360] first4: 0.016348 0.109485 0.066159 -0.074161 [Debug] temb: [2560] first4: 0.026323 -0.023037 -0.016484 0.047127 [Debug] temb_t: [2560] first4: 0.014564 -0.022307 -0.020828 0.048741 [Debug] temb_r: [2560] first4: 0.011759 -0.000730 0.004344 -0.001614 [Debug] sinusoidal_t: [256] first4: 0.562486 0.789701 0.439822 -0.023583 [Debug] sinusoidal_r: [256] first4: 1.000000 1.000000 1.000000 1.000000 [Debug] temb_lin1_t: [2560] first4: 0.064043 -0.018896 0.138192 -0.113571 [Debug] temb_lin1_r: [2560] first4: -0.009850 0.016688 -0.018317 0.012020 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.342407 -0.579834 0.230591 -0.188965 [Debug] proj_in_input: [192, 2170] first4: -0.124812 1.441172 0.303914 -0.631591 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.031097 0.652405 0.101044 0.874352 [Debug] layer0_sa_input: [2560, 1085] first4: 0.040291 -0.045382 0.094641 -0.134234 [Debug] layer0_q_after_rope: [128, 32] first4: -0.256279 -2.170833 0.185178 0.018836 [Debug] layer0_k_after_rope: [128, 8] first4: 0.594727 2.179688 0.738281 0.157959 [Debug] layer0_sa_output: [2560, 1085] first4: 1.869141 4.457031 -0.947754 5.628906 [Debug] layer0_attn_out: [4096, 1085] first4: 0.490657 -2.306927 -0.464014 -2.894142 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.490657 -2.306927 -0.464014 -2.894142 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.333797 -2.040814 -0.313013 -3.795997 [Debug] hidden_after_layer0: [2560, 1085] first4: 1.681171 -2.476156 240.881805 -3.767022 [Debug] hidden_after_layer6: [2560, 1085] first4: -5.999249 -2.457250 320.352478 -12.569094 [Debug] hidden_after_layer12: [2560, 1085] first4: -11.112516 29.593269 260.558990 29.256384 [Debug] hidden_after_layer18: [2560, 1085] first4: 41.531723 11.562994 274.958557 2.650103 [Debug] hidden_after_layer31: [2560, 1085] first4: -94.936852 -233.434540 -144.318893 239.793304 [Debug] dit_step0_vt: [2170, 64] first4: -1.132568 1.507210 0.488708 1.207520 [Debug] dit_step0_xt: [2170, 64] first4: 0.245816 2.087740 -0.194089 0.792769 [DiT] Step 1/8 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.975220 1.307678 0.660606 1.104370 [Debug] dit_step1_xt: [2170, 64] first4: 0.299010 2.016412 -0.230122 0.732531 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: -0.410004 1.433228 0.849564 0.895142 [Debug] dit_step2_xt: [2170, 64] first4: 0.326344 1.920864 -0.286760 0.672854 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: -0.251953 1.431351 0.903862 0.937393 [Debug] dit_step3_xt: [2170, 64] first4: 0.347340 1.801585 -0.362081 0.594738 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: -1.106201 1.822098 1.516876 0.852051 [Debug] dit_step4_xt: [2170, 64] first4: 0.465861 1.606360 -0.524604 0.503447 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: 0.175018 2.587036 0.726379 0.506226 [Debug] dit_step5_xt: [2170, 64] first4: 0.440859 1.236784 -0.628372 0.431129 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: 0.884033 2.291153 0.515198 0.586304 [Debug] dit_step6_xt: [2170, 64] first4: 0.264052 0.778553 -0.731412 0.313869 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: -1.638672 2.251587 1.183777 0.539673 [Debug] dit_x0: [2170, 64] first4: 0.755654 0.103077 -1.086545 0.151967 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 636.9 ms (636.9 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.755654 0.103077 -1.086545 0.151967 [Store] Unload DiT (4230.2 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: Vulkan0 (CPU threads: 16) [VAE] Backend: Vulkan0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 699 ms [VAE] Tiled decode: 3 tiles (chunk=1024, overlap=64, stride=896) [VAE] Graph: 479 nodes, T_latent=960 [VAE] Upsample factor: 1920.00 (expected ~1920) [VAE] Graph: 479 nodes, T_latent=1024 [VAE] Graph: 479 nodes, T_latent=442 [VAE] Tiled decode done: 3 tiles -> T_audio=4166400 (86.80s @ 48kHz) [VAE-Decode Batch0] Decode: 972.3 ms [Debug] vae_audio: [2, 4166400] first4: 0.002766 0.004573 0.004261 0.004056 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 456 ms (190.55x realtime), 32 threads [MP3] Wrote ggml-xl-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 10:20:49.541 | INFO | acestep.core.generation.handler.init_service_loader:_load_main_model_from_checkpoint:174 - [initialize_service] Attempting to load model with attention implementation: sdpa [Python] Wrote python-sft/output.wav: 4166400 samples (86.80s @ 48kHz stereo) [Python] Done, 166 dump files [SFT] Cosine similarities GGML vs Python stage GGML vs Python text_hidden 0.999817 lyric_embed 1.000000 enc_hidden 0.999829 detok_output 0.999984 context 0.999990 noise 1.000000 temb_t 0.999994 hidden_after_proj_in 0.999985 enc_after_cond_emb 0.999822 layer0_sa_output 0.999931 hidden_after_layer0 0.999873 hidden_after_layer6 0.999811 hidden_after_layer12 0.999421 hidden_after_layer18 0.998641 hidden_after_layer23 0.998938 dit_step0_vt 0.998856 dit_step0_xt 1.000000 dit_step5_vt 0.999344 dit_step5_xt 0.999978 dit_step10_vt 0.998888 dit_step10_xt 0.999927 dit_step15_vt 0.997760 dit_step15_xt 0.999817 dit_step20_vt 0.995043 dit_step20_xt 0.999521 dit_step25_vt 0.991659 dit_step25_xt 0.998920 dit_step30_vt 0.987430 dit_step30_xt 0.997860 dit_step35_vt 0.982919 dit_step35_xt 0.996447 dit_step40_vt 0.979898 dit_step40_xt 0.994972 dit_step45_vt 0.983258 dit_step45_xt 0.993894 dit_step49_vt 0.988895 dit_x0 0.993518 vae_audio 0.969461 vae_audio (STFT cosine) 0.988335 [SFT] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 1.000000 0.020522 0.001472 -0.001641 0.978491 -0.001619 0.978872 dit_step5_xt 0.999978 0.077852 0.004543 -0.008269 0.884942 -0.008480 0.883811 dit_step10_xt 0.999927 0.132113 0.007408 -0.014846 0.801091 -0.015075 0.802324 dit_step15_xt 0.999817 0.187068 0.010675 -0.021371 0.730535 -0.021933 0.731304 dit_step20_xt 0.999521 0.302029 0.015662 -0.027925 0.677527 -0.028882 0.679066 dit_step25_xt 0.998920 0.551316 0.021978 -0.034552 0.646504 -0.035643 0.650780 dit_step30_xt 0.997860 0.794938 0.030169 -0.041216 0.640813 -0.042577 0.647107 dit_step35_xt 0.996447 1.055773 0.039785 -0.047879 0.661242 -0.049471 0.669256 dit_step40_xt 0.994972 1.314899 0.050486 -0.054558 0.705483 -0.056532 0.715783 dit_step45_xt 0.993894 1.543376 0.060600 -0.061333 0.769182 -0.063533 0.780575 [Xl-turbo] steps=8, shift=3.0 | acestep-v15-xl-turbo-Q8_0.gguf [GGML] Running acestep-v15-xl-turbo-Q8_0.gguf... [GGML] Done, 47 dump files [Python] Initializing acestep-v15-xl-turbo... Loading checkpoint shards: 0%| | 0/4 [00:00 1.0 (turbo does not use CFG). 2026-07-12 10:20:53.799 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:20:53.799 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:20:53.799 | INFO | acestep.core.generation.handler.generate_music_request:_prepare_reference_and_source_audio:179 - [generate_music] cover task: no src_audio but audio codes provided, proceeding with codes 2026-07-12 10:20:53.801 | INFO | acestep.core.generation.handler.generate_music:_vram_preflight_check:156 - [generate_music] VRAM pre-flight: 76.75 GB free, ~1.23 GB needed (batch=1, duration=88s, mode=xl_turbo). 2026-07-12 10:20:53.806 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:20:53.811 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_precomputed_lm_hints:31 - [generate_music] Decoding audio codes for LM hints for item 0... 2026-07-12 10:20:53.813 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:20:53.813 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:20:53.813 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:20:53.813 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:122 - text_prompt: # Instruction Generate audio semantic tokens based on the given conditions: # Caption An upbeat and anthemic pop-rock track driven by bright, slightly overdriven # Metas - bpm: 83 - timesignature: 4 - keyscale: G major - duration: 88 seconds <|endoftext|> 2026-07-12 10:20:53.813 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:20:53.813 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:124 - lyrics_text: # Languages fr # Lyric # Lyric [Intro - Guitar Riff] [Verse 1] Dans le monde des tutos virtuels G ta toise en nouvelle passion Avec Ggendoline et PumbΓ© Γ  midi La communautΓ©, c'est l'unitΓ© Quel joie, une clΓ© [Chorus] Dans le monde des tutos virtuels GΓ’ndoline et PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner [Guitar Solo] [Verse 2] Dans le monde des tutos virtuels GΓ’ndoline, PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner<|endoftext|> 2026-07-12 10:20:53.813 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:20:53.816 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:20:53.824 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:20:53.824 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:20:53.831 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 10:20:53.837 | INFO | acestep.models.common.dcw_correction:__init__:98 - [DCW] Active β€” mode=double, scaler=0.0500, high_scaler=0.0200, wavelet='haar' 2026-07-12 10:20:54.394 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 10:20:54.395 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:64 - [generate_music] pred_latents: torch.Size([1, 2170, 64]), dtype=torch.bfloat16 2026-07-12 10:20:54.395 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.006383657455444336, 'diffusion_time_cost': 0.5564494132995605, 'diffusion_per_step_time_cost': 0.06955617666244507, 'total_time_cost': 0.5628330707550049, 'offload_time_cost': 0.0} 2026-07-12 10:20:54.418 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 10:20:54.419 | DEBUG | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:136 - [generate_music] Before VAE decode: allocated=17.03GB, max=18.34GB 2026-07-12 10:20:54.420 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:154 - [generate_music] Effective free VRAM before VAE decode: 76.66 GB 2026-07-12 10:20:54.420 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:172 - [generate_music] Using tiled VAE decode to reduce VRAM usage... 2026-07-12 10:20:54.420 | DEBUG | acestep.core.generation.handler.memory_utils:_get_auto_decode_chunk_size:108 - [_get_auto_decode_chunk_size] Effective free VRAM: 76.66 GB 2026-07-12 10:20:54.420 | DEBUG | acestep.core.generation.handler.memory_utils:_should_offload_wav_to_cpu:131 - [_should_offload_wav_to_cpu] Effective free VRAM: 76.66 GB 2026-07-12 10:20:54.420 | INFO | acestep.core.generation.handler.vae_decode:tiled_decode:56 - [tiled_decode] chunk_size=512, offload_wav_to_cpu=False, latents_shape=torch.Size([1, 64, 2170]) 2026-07-12 10:20:54.693 | DEBUG | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:193 - [generate_music] After VAE decode: allocated=17.21GB, max=18.50GB 2026-07-12 10:20:54.900 | INFO | acestep.core.generation.handler.generate_music_payload:_build_generate_music_success_payload:38 - [generate_music] VAE decode completed. Preparing audio tensors... 2026-07-12 10:20:54.903 | INFO | acestep.core.generation.handler.generate_music_payload:_build_generate_music_success_payload:50 - [generate_music] Done! Generated 1 audio tensors. [Request] Parsed ggml-xl-sft/request0.json [Request] seed=42 lm_batch=1 synth_batch=1 [Request] caption: An upbeat and anthemic pop-rock track driven by bright, slig... [Request] lyrics: 490 bytes [Request] bpm=83 dur=88 key=G major ts=4 lang=fr [Request] lm: temp=0.85 cfg=2.0 top_p=0.90 top_k=0 [Request] dit: steps=50 guidance=1.0 shift=1.0 [Request] task_type: text2music [Request] solver: euler (stork_substeps=10) [Request] lm_mode: generate [Request] output_format: mp3 [Request] synth_model: acestep-v15-xl-sft-Q8_0.gguf [Request] vae: vae-BF16.gguf [Request] audio_codes: (present) [Ace-Synth] Batch: 1 request(s) [Registry] Qwen3-Embedding-0.6B-BF16.gguf -> Text-Enc [Registry] Qwen3-Embedding-0.6B-Q8_0.gguf -> Text-Enc [Registry] acestep-5Hz-lm-0.6B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-0.6B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-4B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q5_K_M.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q6_K.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q8_0.gguf -> LM [Registry] acestep-v15-base-BF16.gguf -> DiT [Registry] acestep-v15-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-base-Q6_K.gguf -> DiT [Registry] acestep-v15-base-Q8_0.gguf -> DiT [Registry] acestep-v15-sft-BF16.gguf -> DiT [Registry] acestep-v15-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-BF16.gguf -> DiT [Registry] acestep-v15-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-continuous-BF16.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift1-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift3-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-base-BF16.gguf -> DiT [Registry] acestep-v15-xl-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-base-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sft-BF16.gguf -> DiT [Registry] acestep-v15-xl-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-turbo-BF16.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q8_0.gguf -> DiT [Registry] scragvae-BF16.gguf -> VAE [Registry] vae-BF16.gguf -> VAE [Store] Created (policy=STRICT) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q8_0.gguf: 830 tensors, data at offset 69056 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q8_0.gguf: 830 tensors, data at offset 69056 [Synth-Load] Ready: turbo=no, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=50, guidance=1.0, shift=1.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/Qwen3-Embedding-0.6B-BF16.gguf: 310 tensors, data at offset 5337568 [Load] TextEncoder: 28L, H=1024, Nh=16/8 [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 310 tensors, 1136.5 MB into backend [Store] Load TextEnc: 319 ms [Debug] text_hidden: [70, 1024] first4: 3.705836 2.395382 0.221845 -13.145830 [Debug] lyric_embed: [167, 1024] first4: 0.029175 0.032227 -0.022339 -0.028809 [Store] Unload TextEnc (1136.5 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] CondEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q8_0.gguf: 830 tensors, data at offset 69056 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 141 tensors, 616.6 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L, CLS), text_proj, null_cond [Store] Load CondEnc: 338 ms [CondEnc] Lyric sliding mask: 167x167, window=128 [CondEnc] Timbre sliding mask: 2x2, window=128 (CLS) [Encode] Packed: lyric=167 + timbre=1 + text=70 = 238 tokens [Encode-Text Batch0] 70+167 tokens -> enc_S=238, 7.5 ms [Debug] enc_hidden: [238, 2048] first4: 0.686671 -0.019193 -0.051889 0.022865 [Store] Unload CondEnc (616.6 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] Detokenizer backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q8_0.gguf: 830 tensors, data at offset 69056 [WeightCtx] Loaded 30 tensors, 106.5 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 179 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 107.1 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.124812 1.441172 0.303914 -0.631591 [Store] Unload FSQ-Detok (106.5 MB) [Init-Noise Batch0] Philox noise seed=42, [2170, 64] solver=euler [Debug] noise: [2170, 64] first4: 0.194336 2.156250 -0.171875 0.847656 [Debug] context: [2170, 128] first4: -0.124812 1.441172 0.303914 -0.631591 [Init-Noise] Starting: T=2170, S=1085, enc_S=238, steps=50, batch=1 (cover) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] DiT backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q8_0.gguf: 830 tensors, data at offset 69056 [DiT] Self-attn: Q+K+V fused [DiT] Cross-attn: Q+K+V fused [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 630 tensors, 4230.2 MB into backend [Load] DiT: 32 layers, H=2560, Nh=32/8, D=128 [Store] Load DiT: 2412 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2825 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [15360] first4: 0.013019 0.105941 0.074809 -0.031887 [Debug] temb: [2560] first4: 0.009503 -0.007658 -0.006166 0.004559 [Debug] temb_t: [2560] first4: 0.007870 -0.007938 -0.006686 0.003566 [Debug] temb_r: [2560] first4: 0.001633 0.000279 0.000520 0.000994 [Debug] sinusoidal_t: [256] first4: 0.562486 0.789701 0.439822 -0.023583 [Debug] sinusoidal_r: [256] first4: 1.000000 1.000000 1.000000 1.000000 [Debug] temb_lin1_t: [2560] first4: 0.033182 -0.081874 0.089398 -0.065044 [Debug] temb_lin1_r: [2560] first4: 0.006798 -0.005077 -0.002197 -0.011158 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.331421 -0.495850 0.102783 -0.213135 [Debug] proj_in_input: [192, 2170] first4: -0.124812 1.441172 0.303914 -0.631591 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.032883 0.642090 0.113403 0.870438 [Debug] layer0_sa_input: [2560, 1085] first4: 0.037142 -0.019126 0.099905 -0.097168 [Debug] layer0_q_after_rope: [128, 32] first4: -0.283096 -2.257718 0.025087 -0.190857 [Debug] layer0_k_after_rope: [128, 8] first4: 0.639648 2.990234 0.689941 1.805664 [Debug] layer0_sa_output: [2560, 1085] first4: 2.169922 4.503906 -1.257812 8.242188 [Debug] layer0_attn_out: [4096, 1085] first4: 0.473449 -2.221286 -0.833027 -4.311768 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.473449 -2.221286 -0.833027 -4.311768 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.287414 -1.938571 -0.766499 -4.365265 [Debug] hidden_after_layer0: [2560, 1085] first4: -1.726679 -1.262200 250.903458 -4.068712 [Debug] hidden_after_layer6: [2560, 1085] first4: -8.013782 3.541740 316.336548 -21.316015 [Debug] hidden_after_layer12: [2560, 1085] first4: -23.302528 39.728611 230.845215 18.088783 [Debug] hidden_after_layer18: [2560, 1085] first4: 9.482368 -12.917768 237.357864 19.552364 [Debug] hidden_after_layer31: [2560, 1085] first4: 63.308372 -105.382675 11.681061 84.250259 [Debug] dit_step0_vt: [2170, 64] first4: -0.310181 2.216675 -0.000244 1.695740 [Debug] dit_step0_xt: [2170, 64] first4: 0.200540 2.111917 -0.171870 0.813742 [DiT] Step 1/50 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.339233 2.159302 0.077637 1.558441 [Debug] dit_step1_xt: [2170, 64] first4: 0.207324 2.068730 -0.173423 0.782573 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.321899 2.170959 0.064941 1.564240 [Debug] dit_step2_xt: [2170, 64] first4: 0.213762 2.025311 -0.174722 0.751288 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.274170 2.144287 0.062866 1.570480 [Debug] dit_step3_xt: [2170, 64] first4: 0.219246 1.982426 -0.175979 0.719878 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.264893 2.057068 0.034546 1.582016 [Debug] dit_step4_xt: [2170, 64] first4: 0.224543 1.941284 -0.176670 0.688238 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.249756 2.039795 0.054932 1.594238 [Debug] dit_step5_xt: [2170, 64] first4: 0.229539 1.900488 -0.177769 0.656353 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.246094 1.999146 0.081177 1.609955 [Debug] dit_step6_xt: [2170, 64] first4: 0.234460 1.860505 -0.179392 0.624154 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.244385 1.961243 0.120850 1.614410 [Debug] dit_step7_xt: [2170, 64] first4: 0.239348 1.821281 -0.181809 0.591866 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.255249 1.925018 0.148193 1.619293 [Debug] dit_step8_xt: [2170, 64] first4: 0.244453 1.782780 -0.184773 0.559480 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.279907 1.870453 0.125244 1.622040 [Debug] dit_step9_xt: [2170, 64] first4: 0.250051 1.745371 -0.187278 0.527039 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.280396 1.835907 0.106689 1.613987 [Debug] dit_step10_xt: [2170, 64] first4: 0.255659 1.708653 -0.189412 0.494759 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.278320 1.791229 0.098755 1.598408 [Debug] dit_step11_xt: [2170, 64] first4: 0.261226 1.672828 -0.191387 0.462791 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.272461 1.747963 0.089355 1.586517 [Debug] dit_step12_xt: [2170, 64] first4: 0.266675 1.637869 -0.193174 0.431061 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.261353 1.696368 0.081665 1.573288 [Debug] dit_step13_xt: [2170, 64] first4: 0.271902 1.603942 -0.194807 0.399595 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.259277 1.650734 0.062378 1.570023 [Debug] dit_step14_xt: [2170, 64] first4: 0.277087 1.570927 -0.196055 0.368195 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.251465 1.606903 0.044434 1.558716 [Debug] dit_step15_xt: [2170, 64] first4: 0.282117 1.538789 -0.196943 0.337020 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.238647 1.558899 0.034668 1.561035 [Debug] dit_step16_xt: [2170, 64] first4: 0.286890 1.507611 -0.197637 0.305800 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.210815 1.511475 0.017212 1.558777 [Debug] dit_step17_xt: [2170, 64] first4: 0.291106 1.477382 -0.197981 0.274624 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.180298 1.453918 0.004639 1.540375 [Debug] dit_step18_xt: [2170, 64] first4: 0.294712 1.448303 -0.198074 0.243817 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.148560 1.403564 -0.013062 1.537415 [Debug] dit_step19_xt: [2170, 64] first4: 0.297683 1.420232 -0.197812 0.213068 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.123169 1.350220 -0.015625 1.532684 [Debug] dit_step20_xt: [2170, 64] first4: 0.300146 1.393228 -0.197500 0.182415 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.092346 1.301941 -0.035522 1.523407 [Debug] dit_step21_xt: [2170, 64] first4: 0.301993 1.367189 -0.196790 0.151947 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.065735 1.260315 -0.043457 1.503601 [Debug] dit_step22_xt: [2170, 64] first4: 0.303308 1.341982 -0.195920 0.121874 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.049805 1.200256 -0.058960 1.493622 [Debug] dit_step23_xt: [2170, 64] first4: 0.304304 1.317977 -0.194741 0.092002 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.042908 1.145630 -0.065186 1.477264 [Debug] dit_step24_xt: [2170, 64] first4: 0.305162 1.295065 -0.193437 0.062457 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.033936 1.094421 -0.057373 1.471313 [Debug] dit_step25_xt: [2170, 64] first4: 0.305841 1.273176 -0.192290 0.033031 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.034180 1.027466 -0.069580 1.446655 [Debug] dit_step26_xt: [2170, 64] first4: 0.306525 1.252627 -0.190898 0.004097 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.040527 0.967285 -0.071960 1.419159 [Debug] dit_step27_xt: [2170, 64] first4: 0.307335 1.233281 -0.189459 -0.024286 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.052368 0.912079 -0.078613 1.395020 [Debug] dit_step28_xt: [2170, 64] first4: 0.308383 1.215040 -0.187887 -0.052186 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.055542 0.845398 -0.089172 1.378723 [Debug] dit_step29_xt: [2170, 64] first4: 0.309493 1.198132 -0.186103 -0.079761 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.067444 0.770523 -0.101929 1.356812 [Debug] dit_step30_xt: [2170, 64] first4: 0.310842 1.182721 -0.184065 -0.106897 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.079346 0.688187 -0.104980 1.335571 [Debug] dit_step31_xt: [2170, 64] first4: 0.312429 1.168958 -0.181965 -0.133608 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.108704 0.597094 -0.110718 1.309021 [Debug] dit_step32_xt: [2170, 64] first4: 0.314603 1.157016 -0.179751 -0.159789 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.123169 0.516153 -0.134399 1.289551 [Debug] dit_step33_xt: [2170, 64] first4: 0.317067 1.146693 -0.177063 -0.185580 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.142822 0.427513 -0.130676 1.278625 [Debug] dit_step34_xt: [2170, 64] first4: 0.319923 1.138142 -0.174449 -0.211152 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.164795 0.318176 -0.132294 1.228577 [Debug] dit_step35_xt: [2170, 64] first4: 0.323219 1.131779 -0.171804 -0.235724 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.166748 0.222260 -0.114159 1.201660 [Debug] dit_step36_xt: [2170, 64] first4: 0.326554 1.127334 -0.169520 -0.259757 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.187809 0.124695 -0.115940 1.168396 [Debug] dit_step37_xt: [2170, 64] first4: 0.330310 1.124840 -0.167202 -0.283125 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.184547 0.025909 -0.090958 1.128540 [Debug] dit_step38_xt: [2170, 64] first4: 0.334001 1.124322 -0.165382 -0.305696 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.211184 -0.083160 -0.071136 1.102051 [Debug] dit_step39_xt: [2170, 64] first4: 0.338225 1.125985 -0.163960 -0.327737 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.239487 -0.184479 -0.047112 1.050598 [Debug] dit_step40_xt: [2170, 64] first4: 0.343015 1.129674 -0.163017 -0.348749 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.268890 -0.292612 -0.031952 1.013489 [Debug] dit_step41_xt: [2170, 64] first4: 0.348392 1.135527 -0.162378 -0.369018 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.293640 -0.372345 0.003418 0.968567 [Debug] dit_step42_xt: [2170, 64] first4: 0.354265 1.142973 -0.162447 -0.388390 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.315552 -0.478958 0.033287 0.918823 [Debug] dit_step43_xt: [2170, 64] first4: 0.360576 1.152553 -0.163113 -0.406766 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.341064 -0.570740 0.082809 0.864075 [Debug] dit_step44_xt: [2170, 64] first4: 0.367397 1.163967 -0.164769 -0.424048 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.374786 -0.649384 0.112274 0.799805 [Debug] dit_step45_xt: [2170, 64] first4: 0.374893 1.176955 -0.167014 -0.440044 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.426941 -0.727646 0.155151 0.747986 [Debug] dit_step46_xt: [2170, 64] first4: 0.383432 1.191508 -0.170117 -0.455004 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.468348 -0.801224 0.217224 0.711731 [Debug] dit_step47_xt: [2170, 64] first4: 0.392799 1.207533 -0.174462 -0.469238 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.529755 -0.859985 0.265442 0.693359 [Debug] dit_step48_xt: [2170, 64] first4: 0.403394 1.224732 -0.179771 -0.483105 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.524687 -0.846359 0.191711 0.705200 [Debug] dit_x0: [2170, 64] first4: 0.413888 1.241659 -0.183605 -0.497209 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 3719.3 ms (3719.3 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.413888 1.241659 -0.183605 -0.497209 [Store] Unload DiT (4230.2 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: Vulkan0 (CPU threads: 16) [VAE] Backend: Vulkan0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 673 ms [VAE] Tiled decode: 3 tiles (chunk=1024, overlap=64, stride=896) [VAE] Graph: 479 nodes, T_latent=960 [VAE] Upsample factor: 1920.00 (expected ~1920) [VAE] Graph: 479 nodes, T_latent=1024 [VAE] Graph: 479 nodes, T_latent=442 [VAE] Tiled decode done: 3 tiles -> T_audio=4166400 (86.80s @ 48kHz) [VAE-Decode Batch0] Decode: 952.9 ms [Debug] vae_audio: [2, 4166400] first4: -0.001412 -0.000683 -0.000808 -0.000324 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 415 ms (208.95x realtime), 32 threads [MP3] Wrote ggml-xl-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 10:21:05.248 | INFO | acestep.core.generation.handler.init_service_loader:_load_main_model_from_checkpoint:174 - [initialize_service] Attempting to load model with attention implementation: sdpa [Python] Generating (acestep-v15-xl-turbo, 8 steps)... Using precomputed LM hints Using precomputed LM hints [Python] Wrote python-xl-turbo/output.wav: 4166400 samples (86.80s @ 48kHz stereo) [Python] Done, 40 dump files [Xl-turbo] Cosine similarities GGML vs Python stage GGML vs Python text_hidden 0.999817 lyric_embed 1.000000 enc_hidden 0.999556 detok_output 0.999984 context 0.999990 noise 1.000000 temb_t 0.999996 hidden_after_proj_in 0.999977 enc_after_cond_emb 0.999674 layer0_sa_output 0.989298 hidden_after_layer0 0.999106 hidden_after_layer6 0.999672 hidden_after_layer12 0.995154 hidden_after_layer18 0.995599 hidden_after_layer31 0.986885 dit_step0_vt 0.902680 dit_step0_xt 0.999824 dit_step1_vt 0.916574 dit_step1_xt 0.999431 dit_step2_vt 0.907165 dit_step2_xt 0.998390 dit_step3_vt 0.902560 dit_step3_xt 0.995863 dit_step4_vt 0.898055 dit_step4_xt 0.989806 dit_step5_vt 0.887659 dit_step5_xt 0.975506 dit_step6_vt 0.876946 dit_step6_xt 0.947365 dit_step7_vt 0.855403 dit_x0 0.902570 vae_audio 0.691889 vae_audio (STFT cosine) 0.899977 [Xl-turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999824 0.151871 0.013153 -0.002921 0.978399 -0.002721 0.976534 dit_step1_xt 0.999431 0.326770 0.023265 -0.005826 0.952801 -0.005567 0.949004 dit_step2_xt 0.998390 0.542094 0.037340 -0.009578 0.924377 -0.009330 0.918931 dit_step3_xt 0.995863 0.836863 0.056942 -0.014948 0.893928 -0.014418 0.887241 dit_step4_xt 0.989806 1.287795 0.085603 -0.022681 0.864911 -0.021453 0.857796 dit_step5_xt 0.975506 1.962434 0.130236 -0.033775 0.848339 -0.032240 0.840033 dit_step6_xt 0.947365 2.966949 0.197932 -0.049909 0.870893 -0.049194 0.863964 [Xl-sft] steps=50, shift=1.0 | acestep-v15-xl-sft-Q8_0.gguf [GGML] Running acestep-v15-xl-sft-Q8_0.gguf... [GGML] Done, 131 dump files [Python] Initializing acestep-v15-xl-sft... Loading checkpoint shards: 0%| | 0/4 [00:00 2026-07-12 10:21:09.250 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:21:09.250 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:124 - lyrics_text: # Languages fr # Lyric # Lyric [Intro - Guitar Riff] [Verse 1] Dans le monde des tutos virtuels G ta toise en nouvelle passion Avec Ggendoline et PumbΓ© Γ  midi La communautΓ©, c'est l'unitΓ© Quel joie, une clΓ© [Chorus] Dans le monde des tutos virtuels GΓ’ndoline et PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner [Guitar Solo] [Verse 2] Dans le monde des tutos virtuels GΓ’ndoline, PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner<|endoftext|> 2026-07-12 10:21:09.250 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:21:09.253 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:21:09.261 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:21:09.262 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:21:09.268 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... [Python] Generating (acestep-v15-xl-sft, 50 steps)... Using precomputed LM hints Using precomputed LM hints 0%| | 0/50 [00:00