[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: CPU (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: 232 ms [Debug] text_hidden: [70, 1024] first4: 3.704526 2.436253 0.222853 -13.131872 [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: CPU (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: 348 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, 155.7 ms [Debug] enc_hidden: [238, 2048] first4: 1.758412 -0.049824 -0.132638 0.057941 [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: CPU (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: 184 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 496.4 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.126218 1.441045 0.305219 -0.629688 [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: -0.126218 1.441045 0.305219 -0.629688 [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: CPU (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: 996 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: 0.259485 -0.161550 -0.096885 0.051766 [Debug] temb: [2048] first4: 0.000214 -0.132557 -0.035428 0.064847 [Debug] temb_t: [2048] first4: 0.001194 0.026823 -0.052744 0.063762 [Debug] temb_r: [2048] first4: -0.000980 -0.159380 0.017316 0.001084 [Debug] sinusoidal_t: [256] first4: 0.562379 0.789627 0.439928 -0.023645 [Debug] sinusoidal_r: [256] first4: 1.000000 1.000000 1.000000 1.000000 [Debug] temb_lin1_t: [2048] first4: -0.049228 -0.051913 -0.015026 -0.038076 [Debug] temb_lin1_r: [2048] first4: -0.013066 -0.018835 -0.015731 0.008462 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.038249 -0.957445 0.537078 0.447006 [Debug] proj_in_input: [192, 2170] first4: -0.126218 1.441045 0.305219 -0.629688 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.174778 0.815771 0.331865 -0.559953 [Debug] layer0_sa_input: [2048, 1085] first4: -0.718529 -0.757126 -0.047071 0.261381 [Debug] layer0_q_after_rope: [128, 16] first4: -1.544202 -1.032205 0.191246 0.456493 [Debug] layer0_k_after_rope: [128, 8] first4: -0.174778 0.815771 0.331865 -0.559953 [Debug] layer0_sa_output: [2048, 1085] first4: -1.514885 0.168405 -0.354086 0.508003 [Debug] layer0_attn_out: [2048, 1085] first4: -1.544202 -1.032205 0.191246 0.456493 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.544202 -1.032205 0.191246 0.456493 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.596852 -0.819318 -0.283829 0.491399 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.076782 0.569024 52.384930 -0.865368 [Debug] hidden_after_layer6: [2048, 1085] first4: -21.564232 0.242397 33.815739 -4.188777 [Debug] hidden_after_layer12: [2048, 1085] first4: -14.881989 -18.805410 69.848145 28.630205 [Debug] hidden_after_layer18: [2048, 1085] first4: -26.210476 15.271601 58.086185 21.408840 [Debug] hidden_after_layer23: [2048, 1085] first4: -4.709541 41.288780 192.847153 148.677338 [Debug] dit_step0_vt: [2170, 64] first4: 0.016164 1.116295 0.320523 2.375249 [Debug] dit_step0_xt: [2170, 64] first4: 0.193601 2.105509 -0.186444 0.739690 [DiT] Step 1/8 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.193502 1.360758 -0.151318 1.857501 [Debug] dit_step1_xt: [2170, 64] first4: 0.204156 2.031286 -0.178191 0.638372 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: -0.016619 1.241479 0.111612 2.392810 [Debug] dit_step2_xt: [2170, 64] first4: 0.205264 1.948521 -0.185631 0.478851 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: 0.250226 1.084855 0.270988 2.656159 [Debug] dit_step3_xt: [2170, 64] first4: 0.184412 1.858116 -0.208214 0.257505 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: 0.281901 1.035304 0.111452 2.713949 [Debug] dit_step4_xt: [2170, 64] first4: 0.154208 1.747191 -0.220155 -0.033275 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: 0.264211 0.928700 -0.288185 2.749583 [Debug] dit_step5_xt: [2170, 64] first4: 0.116463 1.614520 -0.178986 -0.426073 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: 0.173657 0.643419 -0.835215 2.794378 [Debug] dit_step6_xt: [2170, 64] first4: 0.081732 1.485836 -0.011943 -0.984949 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: -0.009575 0.263735 -1.486802 3.090469 [Debug] dit_x0: [2170, 64] first4: 0.084605 1.406715 0.434098 -1.912089 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 29518.6 ms (29518.6 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.084605 1.406715 0.434098 -1.912089 [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: CPU (CPU threads: 16) [VAE] Backend: CPU, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 715 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: 31983.0 ms [Debug] vae_audio: [2, 4166400] first4: 0.000661 0.001372 0.001156 0.001681 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 491 ms (176.92x realtime), 32 threads [MP3] Wrote ggml-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 09:24:33.996 | 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 09:24:36.207 | WARNING | acestep.training.trainer::40 - bitsandbytes not installed. Using standard AdamW. 2026-07-12 09:24:36.212 | 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 09:24:36.333 | 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 09:24:37.992 | 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 09:24:37.992 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 09:24:37.992 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 09:24:37.992 | 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 09:24:37.995 | 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 09:24:38.001 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 09:24:38.211 | 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 09:24:38.213 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 09:24:38.213 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 09:24:38.213 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 09:24:38.213 | 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 09:24:38.213 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 09:24:38.213 | 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 09:24:38.213 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 09:24:38.220 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 09:24:38.235 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 09:24:38.235 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 09:24:38.270 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 09:24:38.278 | 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 09:24:38.363 | 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 09:24:38.588 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 09:24:38.589 | 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 09:24:38.589 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.006975412368774414, 'diffusion_time_cost': 0.31067824363708496, 'diffusion_per_step_time_cost': 0.03883478045463562, 'total_time_cost': 0.3176536560058594, 'offload_time_cost': 0.0} 2026-07-12 09:24:38.603 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 09:24:38.605 | 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 09:24:38.605 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:154 - [generate_music] Effective free VRAM before VAE decode: 87.72 GB 2026-07-12 09:24:38.605 | 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 09:24:38.605 | DEBUG | acestep.core.generation.handler.memory_utils:_get_auto_decode_chunk_size:108 - [_get_auto_decode_chunk_size] Effective free VRAM: 87.72 GB 2026-07-12 09:24:38.605 | DEBUG | acestep.core.generation.handler.memory_utils:_should_offload_wav_to_cpu:131 - [_should_offload_wav_to_cpu] Effective free VRAM: 87.72 GB 2026-07-12 09:24:38.605 | 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 09:24:38.881 | 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 09:24:39.047 | 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 09:24:39.051 | 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: CPU (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: 223 ms [Debug] text_hidden: [70, 1024] first4: 3.704526 2.436253 0.222853 -13.131872 [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: CPU (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: 333 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, 136.0 ms [Debug] enc_hidden: [238, 2048] first4: 1.758412 -0.049824 -0.132638 0.057941 [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: CPU (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: 181 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 497.5 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.126218 1.441045 0.305219 -0.629688 [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: -0.126218 1.441045 0.305219 -0.629688 [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: CPU (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: 1097 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: 0.155643 -0.115659 -0.086888 0.081457 [Debug] temb: [2048] first4: -0.002814 -0.176488 0.004230 -0.002015 [Debug] temb_t: [2048] first4: -0.001218 0.003344 -0.013065 -0.002081 [Debug] temb_r: [2048] first4: -0.001596 -0.179832 0.017296 0.000066 [Debug] sinusoidal_t: [256] first4: 0.562379 0.789627 0.439928 -0.023645 [Debug] sinusoidal_r: [256] first4: 1.000000 1.000000 1.000000 1.000000 [Debug] temb_lin1_t: [2048] first4: -0.038611 0.029679 0.028318 -0.024159 [Debug] temb_lin1_r: [2048] first4: 0.001863 0.003353 -0.000552 -0.000197 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.088787 -0.914308 0.499170 0.440692 [Debug] proj_in_input: [192, 2170] first4: -0.126218 1.441045 0.305219 -0.629688 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.193547 0.871285 0.327645 -0.583599 [Debug] layer0_sa_input: [2048, 1085] first4: -0.912407 -0.708866 -0.037435 0.293606 [Debug] layer0_q_after_rope: [128, 16] first4: -1.802147 -1.365389 -0.120474 0.380596 [Debug] layer0_k_after_rope: [128, 8] first4: -0.193547 0.871285 0.327645 -0.583599 [Debug] layer0_sa_output: [2048, 1085] first4: -1.649812 0.797531 -0.549257 0.536640 [Debug] layer0_attn_out: [2048, 1085] first4: -1.802147 -1.365389 -0.120474 0.380596 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.802147 -1.365389 -0.120474 0.380596 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.959073 -1.027594 -0.397269 0.384385 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.420540 1.180536 57.404362 -1.597696 [Debug] hidden_after_layer6: [2048, 1085] first4: -17.385201 4.430685 58.665813 -2.951168 [Debug] hidden_after_layer12: [2048, 1085] first4: -11.284355 4.942255 -19.330605 -2.780930 [Debug] hidden_after_layer18: [2048, 1085] first4: -3.720545 19.587526 -35.580078 -11.202410 [Debug] hidden_after_layer23: [2048, 1085] first4: 34.085365 65.913452 61.195549 -7.649443 [Debug] dit_step0_vt: [2170, 64] first4: -0.573803 2.549292 -0.174947 1.459109 [Debug] dit_step0_xt: [2170, 64] first4: 0.205812 2.105264 -0.168376 0.818474 [DiT] Step 1/50 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.645159 2.545997 0.004436 1.296897 [Debug] dit_step1_xt: [2170, 64] first4: 0.218715 2.054344 -0.168465 0.792536 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.639208 2.496887 0.038706 1.310340 [Debug] dit_step2_xt: [2170, 64] first4: 0.231499 2.004406 -0.169239 0.766329 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.626352 2.447198 0.032850 1.319771 [Debug] dit_step3_xt: [2170, 64] first4: 0.244026 1.955463 -0.169896 0.739934 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.555068 2.353463 0.029818 1.309741 [Debug] dit_step4_xt: [2170, 64] first4: 0.255128 1.908393 -0.170492 0.713739 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.519853 2.275097 0.001329 1.286966 [Debug] dit_step5_xt: [2170, 64] first4: 0.265525 1.862891 -0.170519 0.688000 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.476843 2.217056 -0.025233 1.296081 [Debug] dit_step6_xt: [2170, 64] first4: 0.275062 1.818550 -0.170014 0.662078 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.447000 2.156301 -0.046777 1.265486 [Debug] dit_step7_xt: [2170, 64] first4: 0.284002 1.775424 -0.169079 0.636769 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.387715 2.055564 -0.052714 1.242817 [Debug] dit_step8_xt: [2170, 64] first4: 0.291756 1.734313 -0.168024 0.611912 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.327880 1.962937 -0.053383 1.228035 [Debug] dit_step9_xt: [2170, 64] first4: 0.298314 1.695054 -0.166957 0.587351 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.267809 1.876237 -0.038295 1.226301 [Debug] dit_step10_xt: [2170, 64] first4: 0.303670 1.657529 -0.166191 0.562825 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.235644 1.812230 -0.046852 1.229588 [Debug] dit_step11_xt: [2170, 64] first4: 0.308383 1.621285 -0.165254 0.538234 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.205502 1.748946 -0.060904 1.203737 [Debug] dit_step12_xt: [2170, 64] first4: 0.312493 1.586306 -0.164036 0.514159 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.195648 1.658799 -0.073467 1.206374 [Debug] dit_step13_xt: [2170, 64] first4: 0.316406 1.553130 -0.162566 0.490031 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.180158 1.597013 -0.096607 1.176015 [Debug] dit_step14_xt: [2170, 64] first4: 0.320009 1.521190 -0.160634 0.466511 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.196391 1.504632 -0.125093 1.145564 [Debug] dit_step15_xt: [2170, 64] first4: 0.323937 1.491097 -0.158132 0.443600 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.180535 1.428227 -0.145792 1.147970 [Debug] dit_step16_xt: [2170, 64] first4: 0.327547 1.462532 -0.155217 0.420640 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.205499 1.352937 -0.169683 1.128811 [Debug] dit_step17_xt: [2170, 64] first4: 0.331657 1.435474 -0.151823 0.398064 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.207817 1.261377 -0.211675 1.106355 [Debug] dit_step18_xt: [2170, 64] first4: 0.335814 1.410246 -0.147589 0.375937 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.227330 1.176153 -0.222775 1.067168 [Debug] dit_step19_xt: [2170, 64] first4: 0.340360 1.386723 -0.143134 0.354594 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.247702 1.081392 -0.240563 1.060618 [Debug] dit_step20_xt: [2170, 64] first4: 0.345314 1.365095 -0.138323 0.333381 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.245698 1.009554 -0.256118 1.045029 [Debug] dit_step21_xt: [2170, 64] first4: 0.350228 1.344904 -0.133200 0.312481 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.296614 0.916811 -0.256166 1.038846 [Debug] dit_step22_xt: [2170, 64] first4: 0.356161 1.326568 -0.128077 0.291704 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.310686 0.801235 -0.280401 0.987383 [Debug] dit_step23_xt: [2170, 64] first4: 0.362374 1.310543 -0.122469 0.271956 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.326664 0.711247 -0.280467 0.956102 [Debug] dit_step24_xt: [2170, 64] first4: 0.368908 1.296318 -0.116860 0.252834 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.363719 0.575427 -0.281427 0.942674 [Debug] dit_step25_xt: [2170, 64] first4: 0.376182 1.284810 -0.111231 0.233981 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.400494 0.475536 -0.261690 0.905569 [Debug] dit_step26_xt: [2170, 64] first4: 0.384192 1.275299 -0.105997 0.215869 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.429878 0.359205 -0.284482 0.880842 [Debug] dit_step27_xt: [2170, 64] first4: 0.392789 1.268115 -0.100308 0.198252 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.442096 0.237171 -0.275339 0.894205 [Debug] dit_step28_xt: [2170, 64] first4: 0.401631 1.263371 -0.094801 0.180368 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.463313 0.132073 -0.252260 0.865775 [Debug] dit_step29_xt: [2170, 64] first4: 0.410898 1.260730 -0.089756 0.163053 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.489181 0.021042 -0.239360 0.841795 [Debug] dit_step30_xt: [2170, 64] first4: 0.420681 1.260309 -0.084968 0.146217 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.508189 -0.107647 -0.243637 0.805043 [Debug] dit_step31_xt: [2170, 64] first4: 0.430845 1.262462 -0.080096 0.130116 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.523606 -0.218630 -0.205325 0.797583 [Debug] dit_step32_xt: [2170, 64] first4: 0.441317 1.266835 -0.075989 0.114164 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.529760 -0.341635 -0.202587 0.767422 [Debug] dit_step33_xt: [2170, 64] first4: 0.451912 1.273667 -0.071937 0.098816 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.547979 -0.425165 -0.169826 0.775434 [Debug] dit_step34_xt: [2170, 64] first4: 0.462872 1.282171 -0.068541 0.083307 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.546766 -0.507606 -0.162904 0.727184 [Debug] dit_step35_xt: [2170, 64] first4: 0.473807 1.292323 -0.065283 0.068764 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.551306 -0.650787 -0.128632 0.676417 [Debug] dit_step36_xt: [2170, 64] first4: 0.484833 1.305339 -0.062710 0.055235 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.555256 -0.737650 -0.121827 0.652337 [Debug] dit_step37_xt: [2170, 64] first4: 0.495938 1.320092 -0.060274 0.042189 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.570568 -0.805595 -0.076084 0.613571 [Debug] dit_step38_xt: [2170, 64] first4: 0.507350 1.336203 -0.058752 0.029917 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.558790 -0.888561 -0.070633 0.573183 [Debug] dit_step39_xt: [2170, 64] first4: 0.518526 1.353975 -0.057339 0.018454 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.536172 -0.971758 -0.026069 0.549169 [Debug] dit_step40_xt: [2170, 64] first4: 0.529249 1.373410 -0.056818 0.007470 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.543736 -1.047134 0.043716 0.502513 [Debug] dit_step41_xt: [2170, 64] first4: 0.540124 1.394353 -0.057692 -0.002580 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.544472 -1.110510 0.044906 0.429289 [Debug] dit_step42_xt: [2170, 64] first4: 0.551013 1.416563 -0.058590 -0.011166 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.488421 -1.169407 0.116450 0.388561 [Debug] dit_step43_xt: [2170, 64] first4: 0.560782 1.439951 -0.060919 -0.018937 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.486064 -1.184953 0.149836 0.355803 [Debug] dit_step44_xt: [2170, 64] first4: 0.570503 1.463650 -0.063916 -0.026053 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.492709 -1.222530 0.187035 0.291861 [Debug] dit_step45_xt: [2170, 64] first4: 0.580357 1.488101 -0.067657 -0.031890 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.436641 -1.282196 0.200670 0.239391 [Debug] dit_step46_xt: [2170, 64] first4: 0.589090 1.513745 -0.071670 -0.036678 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.410028 -1.297645 0.240038 0.204558 [Debug] dit_step47_xt: [2170, 64] first4: 0.597290 1.539698 -0.076471 -0.040769 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.360111 -1.325131 0.244100 0.168746 [Debug] dit_step48_xt: [2170, 64] first4: 0.604493 1.566200 -0.081353 -0.044144 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.397660 -1.362908 0.240166 0.131544 [Debug] dit_x0: [2170, 64] first4: 0.612446 1.593458 -0.086156 -0.046775 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 184916.6 ms (184916.6 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.612446 1.593458 -0.086156 -0.046775 [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: CPU (CPU threads: 16) [VAE] Backend: CPU, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 717 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: 33736.0 ms [Debug] vae_audio: [2, 4166400] first4: -0.001513 -0.000973 -0.001231 -0.000831 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 435 ms (199.74x realtime), 32 threads [MP3] Wrote ggml-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 09:28:22.461 | 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 09:28:23.595 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 09:28:23.595 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 09:28:23.596 | 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 09:28:23.598 | INFO | acestep.core.generation.handler.generate_music:_vram_preflight_check:156 - [generate_music] VRAM pre-flight: 81.68 GB free, ~0.94 GB needed (batch=1, duration=88s, mode=turbo). 2026-07-12 09:28:23.604 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 09:28:23.609 | 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 09:28:23.612 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 09:28:23.612 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 09:28:23.612 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 09:28:23.612 | 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 09:28:23.612 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 09:28:23.612 | 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 09:28:23.612 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 09:28:23.615 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 09:28:23.624 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 09:28:23.624 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 09:28:23.630 | 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 0.999810 lyric_embed 1.000000 enc_hidden 0.999811 detok_output 0.999983 context 0.999990 noise 1.000000 temb_t 0.999997 hidden_after_proj_in 0.999985 enc_after_cond_emb 0.999790 layer0_sa_output 0.999925 hidden_after_layer0 0.999955 hidden_after_layer6 0.999890 hidden_after_layer12 0.999341 hidden_after_layer18 0.996784 hidden_after_layer23 0.993886 dit_step0_vt 0.976465 dit_step0_xt 0.999948 dit_step1_vt 0.978674 dit_step1_xt 0.999831 dit_step2_vt 0.980743 dit_step2_xt 0.999545 dit_step3_vt 0.982345 dit_step3_xt 0.998910 dit_step4_vt 0.981068 dit_step4_xt 0.997490 dit_step5_vt 0.978689 dit_step5_xt 0.994358 dit_step6_vt 0.975875 dit_step6_xt 0.988462 dit_step7_vt 0.970037 dit_x0 0.979746 vae_audio 0.902395 vae_audio (STFT cosine) 0.975871 [Turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999948 0.136893 0.006566 -0.002309 0.972910 -0.002342 0.972003 dit_step1_xt 0.999831 0.268167 0.011368 -0.005308 0.942667 -0.005313 0.941730 dit_step2_xt 0.999545 0.451272 0.017690 -0.009356 0.909159 -0.009311 0.908527 dit_step3_xt 0.998910 0.623028 0.026157 -0.014727 0.873777 -0.014577 0.873624 dit_step4_xt 0.997490 0.757370 0.037950 -0.021793 0.841943 -0.021660 0.841995 dit_step5_xt 0.994358 1.235104 0.055394 -0.031907 0.825216 -0.032109 0.824593 dit_step6_xt 0.988462 2.109199 0.081805 -0.046226 0.856172 -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: CPU (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: 237 ms [Debug] text_hidden: [70, 1024] first4: 3.704526 2.436253 0.222853 -13.131872 [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: CPU (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: 363 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, 135.5 ms [Debug] enc_hidden: [238, 2048] first4: 0.686937 -0.019156 -0.051555 0.022736 [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: CPU (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: 184 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 501.7 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.126218 1.441045 0.305219 -0.629688 [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: -0.126218 1.441045 0.305219 -0.629688 [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: CPU (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: 2949 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: 0.014266 0.109556 0.062634 -0.073523 [Debug] temb: [2560] first4: 0.026310 -0.023038 -0.016457 0.047174 [Debug] temb_t: [2560] first4: 0.014577 -0.022289 -0.020778 0.048747 [Debug] temb_r: [2560] first4: 0.011734 -0.000750 0.004320 -0.001573 [Debug] sinusoidal_t: [256] first4: 0.562379 0.789627 0.439928 -0.023645 [Debug] sinusoidal_r: [256] first4: 1.000000 1.000000 1.000000 1.000000 [Debug] temb_lin1_t: [2560] first4: 0.063855 -0.019018 0.138265 -0.113400 [Debug] temb_lin1_r: [2560] first4: -0.009850 0.016687 -0.018316 0.012019 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.343503 -0.578708 0.229538 -0.188917 [Debug] proj_in_input: [192, 2170] first4: -0.126218 1.441045 0.305219 -0.629688 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.033626 0.652631 0.104875 0.873484 [Debug] layer0_sa_input: [2560, 1085] first4: 0.038285 -0.044974 0.091055 -0.133569 [Debug] layer0_q_after_rope: [128, 32] first4: -0.247461 -2.091876 0.384566 0.186698 [Debug] layer0_k_after_rope: [128, 8] first4: 0.047066 2.194061 0.717729 0.543003 [Debug] layer0_sa_output: [2560, 1085] first4: 1.657912 4.301758 -0.556244 5.407282 [Debug] layer0_attn_out: [4096, 1085] first4: 0.475920 -2.241682 -0.178413 -2.789976 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.475920 -2.241682 -0.178413 -2.789976 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.345974 -1.993592 -0.006429 -3.749107 [Debug] hidden_after_layer0: [2560, 1085] first4: 1.804472 -2.444469 242.960587 -3.724934 [Debug] hidden_after_layer6: [2560, 1085] first4: -6.034281 -2.066730 322.203918 -12.083220 [Debug] hidden_after_layer12: [2560, 1085] first4: -11.093650 29.880548 266.669617 28.304935 [Debug] hidden_after_layer18: [2560, 1085] first4: 41.129612 14.160549 282.558502 4.635794 [Debug] hidden_after_layer31: [2560, 1085] first4: -70.477203 -232.991196 -138.892960 200.803070 [Debug] dit_step0_vt: [2170, 64] first4: -1.145389 1.486296 0.420557 1.194904 [Debug] dit_step0_xt: [2170, 64] first4: 0.246399 2.088691 -0.190991 0.793342 [DiT] Step 1/8 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.708103 2.530723 0.802439 1.126692 [Debug] dit_step1_xt: [2170, 64] first4: 0.285023 1.950652 -0.234761 0.731886 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: -1.404448 2.183197 1.092750 0.875680 [Debug] dit_step2_xt: [2170, 64] first4: 0.378653 1.805105 -0.307611 0.673508 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: -1.356161 1.855755 1.487510 0.992716 [Debug] dit_step3_xt: [2170, 64] first4: 0.491666 1.650459 -0.431570 0.590781 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: 0.349834 2.668602 1.047644 0.649922 [Debug] dit_step4_xt: [2170, 64] first4: 0.454184 1.364537 -0.543817 0.521147 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: -0.139283 1.865305 0.744239 0.163181 [Debug] dit_step5_xt: [2170, 64] first4: 0.474082 1.098065 -0.650137 0.497835 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: -1.304907 2.014624 0.412953 0.671463 [Debug] dit_step6_xt: [2170, 64] first4: 0.735063 0.695140 -0.732728 0.363543 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: -1.277802 1.659054 0.798507 0.635665 [Debug] dit_x0: [2170, 64] first4: 1.118404 0.197424 -0.972280 0.172843 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 80732.2 ms (80732.2 ms/sample) [Debug] dit_output: [2170, 64] first4: 1.118404 0.197424 -0.972280 0.172843 [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: CPU (CPU threads: 16) [VAE] Backend: CPU, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 717 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: 34323.0 ms [Debug] vae_audio: [2, 4166400] first4: 0.002099 0.004610 0.004590 0.003980 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 486 ms (178.55x realtime), 32 threads [MP3] Wrote ggml-xl-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 09:30:27.716 | 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 0.999810 lyric_embed 1.000000 enc_hidden 0.999811 detok_output 0.999983 context 0.999990 noise 1.000000 temb_t 0.999990 hidden_after_proj_in 0.999986 enc_after_cond_emb 0.999791 layer0_sa_output 0.999911 hidden_after_layer0 0.999958 hidden_after_layer6 0.999829 hidden_after_layer12 0.999444 hidden_after_layer18 0.998705 hidden_after_layer23 0.998956 dit_step0_vt 0.998899 dit_step0_xt 1.000000 dit_step5_vt 0.999464 dit_step5_xt 0.999980 dit_step10_vt 0.999081 dit_step10_xt 0.999934 dit_step15_vt 0.998074 dit_step15_xt 0.999840 dit_step20_vt 0.995430 dit_step20_xt 0.999586 dit_step25_vt 0.991297 dit_step25_xt 0.999024 dit_step30_vt 0.986621 dit_step30_xt 0.997969 dit_step35_vt 0.981969 dit_step35_xt 0.996518 dit_step40_vt 0.978996 dit_step40_xt 0.994982 dit_step45_vt 0.982749 dit_step45_xt 0.993857 dit_step49_vt 0.988611 dit_x0 0.993465 vae_audio 0.965188 vae_audio (STFT cosine) 0.985681 [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 0.019781 0.001462 -0.001653 0.978470 -0.001619 0.978872 dit_step5_xt 0.999980 0.076343 0.004361 -0.008355 0.884882 -0.008480 0.883811 dit_step10_xt 0.999934 0.123622 0.006953 -0.015055 0.801047 -0.015075 0.802324 dit_step15_xt 0.999840 0.168954 0.009869 -0.021725 0.730609 -0.021933 0.731304 dit_step20_xt 0.999586 0.229020 0.014558 -0.028448 0.677822 -0.028882 0.679066 dit_step25_xt 0.999024 0.435407 0.020817 -0.035260 0.647194 -0.035643 0.650780 dit_step30_xt 0.997969 0.670583 0.029070 -0.042114 0.641973 -0.042577 0.647107 dit_step35_xt 0.996518 0.890802 0.038817 -0.048974 0.662868 -0.049471 0.669256 dit_step40_xt 0.994982 1.132983 0.049646 -0.055864 0.707534 -0.056532 0.715783 dit_step45_xt 0.993857 1.346886 0.059860 -0.062859 0.771631 -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 09:30:31.000 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 09:30:31.000 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 09:30:31.000 | 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 09:30:31.003 | 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 09:30:31.007 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 09:30:31.012 | 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 09:30:31.014 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 09:30:31.014 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 09:30:31.014 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 09:30:31.014 | 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 09:30:31.014 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 09:30:31.014 | 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 09:30:31.014 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 09:30:31.017 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 09:30:31.025 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 09:30:31.025 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 09:30:31.032 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 09:30:31.038 | 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 09:30:31.584 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 09:30:31.584 | 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 09:30:31.584 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.006397247314453125, 'diffusion_time_cost': 0.5449504852294922, 'diffusion_per_step_time_cost': 0.06811881065368652, 'total_time_cost': 0.5513477325439453, 'offload_time_cost': 0.0} 2026-07-12 09:30:31.607 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 09:30:31.610 | 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 09:30:31.610 | 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 09:30:31.610 | 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 09:30:31.610 | 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 09:30:31.610 | 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 09:30:31.610 | 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 09:30:31.883 | 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 09:30:32.087 | 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 09:30:32.090 | 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: CPU (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: 221 ms [Debug] text_hidden: [70, 1024] first4: 3.704526 2.436253 0.222853 -13.131872 [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: CPU (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: 337 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, 138.8 ms [Debug] enc_hidden: [238, 2048] first4: 0.686937 -0.019156 -0.051555 0.022736 [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: CPU (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: 69 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 498.1 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.126218 1.441045 0.305219 -0.629688 [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: -0.126218 1.441045 0.305219 -0.629688 [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: CPU (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: 2756 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: 0.012917 0.105451 0.074363 -0.032316 [Debug] temb: [2560] first4: 0.009484 -0.007646 -0.006176 0.004539 [Debug] temb_t: [2560] first4: 0.007823 -0.007880 -0.006709 0.003563 [Debug] temb_r: [2560] first4: 0.001661 0.000234 0.000533 0.000975 [Debug] sinusoidal_t: [256] first4: 0.562379 0.789627 0.439928 -0.023645 [Debug] sinusoidal_r: [256] first4: 1.000000 1.000000 1.000000 1.000000 [Debug] temb_lin1_t: [2560] first4: 0.033000 -0.082023 0.089453 -0.064865 [Debug] temb_lin1_r: [2560] first4: 0.006798 -0.005077 -0.002197 -0.011157 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.332896 -0.494266 0.100738 -0.212418 [Debug] proj_in_input: [192, 2170] first4: -0.126218 1.441045 0.305219 -0.629688 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.035997 0.642421 0.117086 0.870434 [Debug] layer0_sa_input: [2560, 1085] first4: 0.037157 -0.019220 0.099356 -0.097384 [Debug] layer0_q_after_rope: [128, 32] first4: -0.275659 -2.152586 0.308523 -0.009736 [Debug] layer0_k_after_rope: [128, 8] first4: 0.698652 2.944157 1.163719 2.030455 [Debug] layer0_sa_output: [2560, 1085] first4: 1.889837 4.370225 -0.862801 7.825058 [Debug] layer0_attn_out: [4096, 1085] first4: 0.457090 -2.168639 -0.540006 -4.110604 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.457090 -2.168639 -0.540006 -4.110604 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.293262 -1.898011 -0.457177 -4.191905 [Debug] hidden_after_layer0: [2560, 1085] first4: -1.706833 -1.232268 247.762390 -3.927608 [Debug] hidden_after_layer6: [2560, 1085] first4: -7.879635 3.664357 315.311981 -21.930168 [Debug] hidden_after_layer12: [2560, 1085] first4: -24.024397 39.492359 230.458252 16.594635 [Debug] hidden_after_layer18: [2560, 1085] first4: 9.399558 -13.303671 233.932190 19.979710 [Debug] hidden_after_layer31: [2560, 1085] first4: 62.853333 -105.289902 7.430450 86.143692 [Debug] dit_step0_vt: [2170, 64] first4: -0.326653 2.216254 0.017260 1.696667 [Debug] dit_step0_xt: [2170, 64] first4: 0.200869 2.111925 -0.172220 0.813723 [DiT] Step 1/50 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.357049 2.153441 0.066694 1.552394 [Debug] dit_step1_xt: [2170, 64] first4: 0.208010 2.068856 -0.173554 0.782675 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.336572 2.184491 0.052536 1.571164 [Debug] dit_step2_xt: [2170, 64] first4: 0.214741 2.025166 -0.174605 0.751252 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.272128 2.132073 0.038226 1.565696 [Debug] dit_step3_xt: [2170, 64] first4: 0.220184 1.982525 -0.175369 0.719938 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.274122 2.061871 0.005652 1.584626 [Debug] dit_step4_xt: [2170, 64] first4: 0.225666 1.941287 -0.175482 0.688245 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.262427 2.049347 0.026175 1.579012 [Debug] dit_step5_xt: [2170, 64] first4: 0.230915 1.900301 -0.176006 0.656665 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.255523 2.020804 0.052468 1.609167 [Debug] dit_step6_xt: [2170, 64] first4: 0.236025 1.859885 -0.177055 0.624482 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.244373 1.992708 0.102450 1.602613 [Debug] dit_step7_xt: [2170, 64] first4: 0.240913 1.820030 -0.179104 0.592430 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.251009 1.933441 0.129493 1.600110 [Debug] dit_step8_xt: [2170, 64] first4: 0.245933 1.781361 -0.181694 0.560427 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.277932 1.882045 0.132553 1.605368 [Debug] dit_step9_xt: [2170, 64] first4: 0.251492 1.743721 -0.184345 0.528320 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.298166 1.836462 0.118155 1.596339 [Debug] dit_step10_xt: [2170, 64] first4: 0.257455 1.706991 -0.186708 0.496393 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.300156 1.780471 0.094778 1.596560 [Debug] dit_step11_xt: [2170, 64] first4: 0.263458 1.671382 -0.188604 0.464462 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.292173 1.732649 0.099070 1.574346 [Debug] dit_step12_xt: [2170, 64] first4: 0.269302 1.636729 -0.190585 0.432975 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.291424 1.689848 0.078722 1.553606 [Debug] dit_step13_xt: [2170, 64] first4: 0.275130 1.602932 -0.192160 0.401903 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.290577 1.625613 0.052287 1.557594 [Debug] dit_step14_xt: [2170, 64] first4: 0.280942 1.570420 -0.193205 0.370751 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.268354 1.576627 0.039685 1.545158 [Debug] dit_step15_xt: [2170, 64] first4: 0.286309 1.538887 -0.193999 0.339848 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.251683 1.541795 0.009132 1.534993 [Debug] dit_step16_xt: [2170, 64] first4: 0.291342 1.508051 -0.194182 0.309148 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.209424 1.490457 0.001103 1.537034 [Debug] dit_step17_xt: [2170, 64] first4: 0.295531 1.478242 -0.194204 0.278407 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.184602 1.426140 -0.020950 1.541193 [Debug] dit_step18_xt: [2170, 64] first4: 0.299223 1.449719 -0.193785 0.247583 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.161899 1.388255 -0.020779 1.529410 [Debug] dit_step19_xt: [2170, 64] first4: 0.302461 1.421954 -0.193369 0.216995 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.133211 1.333215 -0.042859 1.519591 [Debug] dit_step20_xt: [2170, 64] first4: 0.305125 1.395290 -0.192512 0.186604 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.097237 1.290747 -0.056441 1.497018 [Debug] dit_step21_xt: [2170, 64] first4: 0.307070 1.369475 -0.191383 0.156663 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.077736 1.234835 -0.066734 1.493803 [Debug] dit_step22_xt: [2170, 64] first4: 0.308625 1.344778 -0.190049 0.126787 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.079994 1.199941 -0.080699 1.521542 [Debug] dit_step23_xt: [2170, 64] first4: 0.310224 1.320780 -0.188435 0.096356 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.035827 1.119311 -0.094776 1.457574 [Debug] dit_step24_xt: [2170, 64] first4: 0.310941 1.298393 -0.186539 0.067205 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.028416 1.061734 -0.089782 1.429418 [Debug] dit_step25_xt: [2170, 64] first4: 0.311509 1.277159 -0.184743 0.038616 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.038133 1.015656 -0.109317 1.419507 [Debug] dit_step26_xt: [2170, 64] first4: 0.312272 1.256846 -0.182557 0.010226 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.044063 0.955100 -0.129720 1.389229 [Debug] dit_step27_xt: [2170, 64] first4: 0.313153 1.237744 -0.179963 -0.017558 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.028482 0.892159 -0.122321 1.363067 [Debug] dit_step28_xt: [2170, 64] first4: 0.313723 1.219901 -0.177516 -0.044820 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.052923 0.805194 -0.151617 1.347045 [Debug] dit_step29_xt: [2170, 64] first4: 0.314781 1.203797 -0.174484 -0.071761 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.063515 0.731531 -0.161896 1.310028 [Debug] dit_step30_xt: [2170, 64] first4: 0.316052 1.189166 -0.171246 -0.097961 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.076297 0.656095 -0.212420 1.277217 [Debug] dit_step31_xt: [2170, 64] first4: 0.317578 1.176044 -0.166998 -0.123505 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.078876 0.575717 -0.180815 1.258439 [Debug] dit_step32_xt: [2170, 64] first4: 0.319155 1.164530 -0.163381 -0.148674 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.098380 0.472289 -0.210451 1.255917 [Debug] dit_step33_xt: [2170, 64] first4: 0.321123 1.155084 -0.159172 -0.173793 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.127017 0.401035 -0.208230 1.221735 [Debug] dit_step34_xt: [2170, 64] first4: 0.323663 1.147064 -0.155008 -0.198227 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.144828 0.290011 -0.172863 1.182503 [Debug] dit_step35_xt: [2170, 64] first4: 0.326560 1.141263 -0.151550 -0.221877 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.145453 0.195820 -0.171144 1.162920 [Debug] dit_step36_xt: [2170, 64] first4: 0.329469 1.137347 -0.148128 -0.245136 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.155062 0.087109 -0.168531 1.113162 [Debug] dit_step37_xt: [2170, 64] first4: 0.332570 1.135605 -0.144757 -0.267399 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.153062 -0.004280 -0.157478 1.059040 [Debug] dit_step38_xt: [2170, 64] first4: 0.335631 1.135690 -0.141607 -0.288580 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.186323 -0.136399 -0.172178 1.027746 [Debug] dit_step39_xt: [2170, 64] first4: 0.339357 1.138418 -0.138164 -0.309135 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.214459 -0.217776 -0.149508 0.981482 [Debug] dit_step40_xt: [2170, 64] first4: 0.343647 1.142774 -0.135174 -0.328764 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.231071 -0.307675 -0.138062 0.944334 [Debug] dit_step41_xt: [2170, 64] first4: 0.348268 1.148927 -0.132412 -0.347651 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.266101 -0.415712 -0.097381 0.904513 [Debug] dit_step42_xt: [2170, 64] first4: 0.353590 1.157242 -0.130465 -0.365741 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.308052 -0.488827 -0.056500 0.884347 [Debug] dit_step43_xt: [2170, 64] first4: 0.359751 1.167018 -0.129335 -0.383428 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.319844 -0.583172 -0.025037 0.834639 [Debug] dit_step44_xt: [2170, 64] first4: 0.366148 1.178682 -0.128834 -0.400121 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.353472 -0.678945 0.044614 0.756830 [Debug] dit_step45_xt: [2170, 64] first4: 0.373217 1.192261 -0.129726 -0.415258 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.411942 -0.748114 0.104478 0.717099 [Debug] dit_step46_xt: [2170, 64] first4: 0.381456 1.207223 -0.131816 -0.429600 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.454500 -0.826544 0.151991 0.658872 [Debug] dit_step47_xt: [2170, 64] first4: 0.390546 1.223754 -0.134856 -0.442777 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.497933 -0.882459 0.231968 0.658596 [Debug] dit_step48_xt: [2170, 64] first4: 0.400505 1.241403 -0.139495 -0.455949 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.507496 -0.867640 0.150937 0.676870 [Debug] dit_x0: [2170, 64] first4: 0.410655 1.258756 -0.142514 -0.469486 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 507531.6 ms (507531.6 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.410655 1.258756 -0.142514 -0.469486 [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: CPU (CPU threads: 16) [VAE] Backend: CPU, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 719 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: 34254.6 ms [Debug] vae_audio: [2, 4166400] first4: -0.001393 -0.000677 -0.000800 -0.000314 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 447 ms (194.18x realtime), 32 threads [MP3] Wrote ggml-xl-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 09:39:40.286 | 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 0.999810 lyric_embed 1.000000 enc_hidden 0.999541 detok_output 0.999983 context 0.999990 noise 1.000000 temb_t 0.999994 hidden_after_proj_in 0.999977 enc_after_cond_emb 0.999637 layer0_sa_output 0.999594 hidden_after_layer0 0.999411 hidden_after_layer6 0.999788 hidden_after_layer12 0.997050 hidden_after_layer18 0.996193 hidden_after_layer31 0.988188 dit_step0_vt 0.911428 dit_step0_xt 0.999840 dit_step1_vt 0.927547 dit_step1_xt 0.999524 dit_step2_vt 0.928120 dit_step2_xt 0.998737 dit_step3_vt 0.925437 dit_step3_xt 0.996922 dit_step4_vt 0.920268 dit_step4_xt 0.992410 dit_step5_vt 0.911450 dit_step5_xt 0.981547 dit_step6_vt 0.902658 dit_step6_xt 0.959947 dit_step7_vt 0.881509 dit_x0 0.924533 vae_audio 0.729928 vae_audio (STFT cosine) 0.922131 [Xl-turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999840 0.159269 0.012524 -0.003006 0.978372 -0.002721 0.976534 dit_step1_xt 0.999524 0.313860 0.021359 -0.005985 0.952124 -0.005567 0.949004 dit_step2_xt 0.998737 0.543794 0.033286 -0.009745 0.923393 -0.009330 0.918931 dit_step3_xt 0.996922 0.807820 0.049381 -0.015423 0.892348 -0.014418 0.887241 dit_step4_xt 0.992410 1.176894 0.074291 -0.023323 0.863002 -0.021453 0.857796 dit_step5_xt 0.981547 1.768017 0.113404 -0.034920 0.846333 -0.032240 0.840033 dit_step6_xt 0.959947 2.512638 0.172654 -0.051927 0.870061 -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 09:39:43.517 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 09:39:43.517 | 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 09:39:43.517 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 09:39:43.520 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 09:39:43.528 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 09:39:43.528 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 09:39:43.535 | 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