[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-BF16.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-BF16.gguf: 678 tensors, data at offset 57024 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-BF16.gguf: 678 tensors, data at offset 57024 [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: 308 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-BF16.gguf: 678 tensors, data at offset 57024 [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, 1160.5 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L), text_proj, null_cond [Store] Load CondEnc: 134 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, 98.6 ms [Debug] enc_hidden: [238, 2048] first4: 1.758276 -0.049599 -0.132827 0.058514 [Store] Unload CondEnc (1160.5 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] Detokenizer backend: CPU (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-BF16.gguf: 678 tensors, data at offset 57024 [WeightCtx] Loaded 30 tensors, 200.3 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 27 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 910.6 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.124172 1.435321 0.310164 -0.624504 [Store] Unload FSQ-Detok (200.3 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: -0.124172 1.435321 0.310164 -0.624504 [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-BF16.gguf: 678 tensors, data at offset 57024 [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, 3007.9 MB into backend [Load] DiT: 24 layers, H=2048, Nh=16/8, D=128 [Store] Load DiT: 292 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: 0.260222 -0.161617 -0.097078 0.052346 [Debug] temb: [2048] first4: 0.000077 -0.132559 -0.035432 0.064735 [Debug] temb_t: [2048] first4: 0.001069 0.026790 -0.052756 0.063697 [Debug] temb_r: [2048] first4: -0.000991 -0.159349 0.017324 0.001038 [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: -0.049513 -0.051899 -0.014138 -0.038434 [Debug] temb_lin1_r: [2048] first4: -0.013266 -0.018319 -0.016375 0.008532 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.039860 -0.969668 0.533155 0.446510 [Debug] proj_in_input: [192, 2170] first4: -0.124172 1.435321 0.310164 -0.624504 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.168718 0.814694 0.326832 -0.562843 [Debug] layer0_sa_input: [2048, 1085] first4: -0.719531 -0.764448 -0.047718 0.261790 [Debug] layer0_q_after_rope: [128, 16] first4: -1.541144 -1.045369 0.187089 0.455729 [Debug] layer0_k_after_rope: [128, 8] first4: -0.168718 0.814694 0.326832 -0.562843 [Debug] layer0_sa_output: [2048, 1085] first4: -1.500230 0.170585 -0.354305 0.512672 [Debug] layer0_attn_out: [2048, 1085] first4: -1.541144 -1.045369 0.187089 0.455729 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.541144 -1.045369 0.187089 0.455729 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.598600 -0.822307 -0.294828 0.493505 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.101245 0.570783 52.416466 -0.907659 [Debug] hidden_after_layer6: [2048, 1085] first4: -21.190231 0.074572 33.690300 -4.495945 [Debug] hidden_after_layer12: [2048, 1085] first4: -15.056320 -17.914547 72.452255 28.845539 [Debug] hidden_after_layer18: [2048, 1085] first4: -27.582886 15.800710 60.610252 21.090904 [Debug] hidden_after_layer23: [2048, 1085] first4: -12.761587 45.375771 198.019775 145.498611 [Debug] dit_step0_vt: [2170, 64] first4: 0.022098 1.114333 0.345665 2.377955 [Debug] dit_step0_xt: [2170, 64] first4: 0.193331 2.105598 -0.187587 0.739567 [DiT] Step 1/8 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.235107 1.306051 -0.132166 1.882218 [Debug] dit_step1_xt: [2170, 64] first4: 0.206156 2.034359 -0.180378 0.636901 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: -0.023448 1.209205 0.106523 2.388911 [Debug] dit_step2_xt: [2170, 64] first4: 0.207719 1.953746 -0.187480 0.477640 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: 0.245050 1.089456 0.259572 2.637362 [Debug] dit_step3_xt: [2170, 64] first4: 0.187298 1.862958 -0.209111 0.257860 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: 0.290814 1.012087 0.108243 2.713056 [Debug] dit_step4_xt: [2170, 64] first4: 0.156139 1.754520 -0.220708 -0.032824 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: 0.266973 0.927089 -0.300413 2.760701 [Debug] dit_step5_xt: [2170, 64] first4: 0.118000 1.622078 -0.177792 -0.427210 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: 0.172337 0.637162 -0.831558 2.825220 [Debug] dit_step6_xt: [2170, 64] first4: 0.083533 1.494646 -0.011480 -0.992254 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: -0.007689 0.185485 -1.474427 3.111979 [Debug] dit_x0: [2170, 64] first4: 0.085840 1.439000 0.430848 -1.925848 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 17315.2 ms (17315.2 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.085840 1.439000 0.430848 -1.925848 [Store] Unload DiT (3007.9 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: 703 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: 27037.8 ms [Debug] vae_audio: [2, 4166400] first4: 0.000701 0.001430 0.001203 0.001720 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 480 ms (180.95x realtime), 32 threads [MP3] Wrote ggml-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 08:17:50.366 | 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 08:17:52.068 | WARNING | acestep.training.trainer::40 - bitsandbytes not installed. Using standard AdamW. 2026-07-12 08:17:52.070 | 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 08:17:52.150 | 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 08:17:53.577 | 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 08:17:53.577 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 08:17:53.577 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 08:17:53.577 | 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 08:17:53.580 | INFO | acestep.core.generation.handler.generate_music:_vram_preflight_check:156 - [generate_music] VRAM pre-flight: 88.44 GB free, ~0.94 GB needed (batch=1, duration=88s, mode=turbo). 2026-07-12 08:17:53.586 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 08:17:53.789 | 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 08:17:53.792 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 08:17:53.792 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 08:17:53.792 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 08:17:53.792 | 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 08:17:53.792 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 08:17:53.792 | 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 08:17:53.792 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 08:17:53.799 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 08:17:53.814 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 08:17:53.814 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 08:17:53.839 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 08:17:53.846 | 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 08:17:53.931 | 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 08:17:54.155 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 08:17:54.156 | 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 08:17:54.156 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.006940364837646484, 'diffusion_time_cost': 0.3087425231933594, 'diffusion_per_step_time_cost': 0.03859281539916992, 'total_time_cost': 0.31568288803100586, 'offload_time_cost': 0.0} 2026-07-12 08:17:54.170 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 08:17:54.172 | 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 08:17:54.172 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:154 - [generate_music] Effective free VRAM before VAE decode: 88.34 GB 2026-07-12 08:17:54.172 | 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 08:17:54.172 | DEBUG | acestep.core.generation.handler.memory_utils:_get_auto_decode_chunk_size:108 - [_get_auto_decode_chunk_size] Effective free VRAM: 88.34 GB 2026-07-12 08:17:54.172 | DEBUG | acestep.core.generation.handler.memory_utils:_should_offload_wav_to_cpu:131 - [_should_offload_wav_to_cpu] Effective free VRAM: 88.34 GB 2026-07-12 08:17:54.172 | 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 08:17:54.447 | 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 08:17:54.610 | 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 08:17:54.613 | 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-BF16.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-BF16.gguf: 678 tensors, data at offset 56960 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-BF16.gguf: 678 tensors, data at offset 56960 [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: 256 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-BF16.gguf: 678 tensors, data at offset 56960 [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, 1160.5 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L), text_proj, null_cond [Store] Load CondEnc: 958 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, 104.1 ms [Debug] enc_hidden: [238, 2048] first4: 1.758276 -0.049599 -0.132827 0.058514 [Store] Unload CondEnc (1160.5 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] Detokenizer backend: CPU (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-BF16.gguf: 678 tensors, data at offset 56960 [WeightCtx] Loaded 30 tensors, 200.3 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 130 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 899.1 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.124172 1.435321 0.310164 -0.624504 [Store] Unload FSQ-Detok (200.3 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: -0.124172 1.435321 0.310164 -0.624504 [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-BF16.gguf: 678 tensors, data at offset 56960 [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, 3007.9 MB into backend [Load] DiT: 24 layers, H=2048, Nh=16/8, D=128 [Store] Load DiT: 1900 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: 0.155061 -0.116337 -0.087159 0.082613 [Debug] temb: [2048] first4: -0.002789 -0.176915 0.004231 -0.001984 [Debug] temb_t: [2048] first4: -0.001270 0.003305 -0.013137 -0.002086 [Debug] temb_r: [2048] first4: -0.001519 -0.180220 0.017368 0.000102 [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: -0.039439 0.029574 0.028407 -0.024894 [Debug] temb_lin1_r: [2048] first4: 0.002274 0.002999 -0.001040 -0.000388 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.090499 -0.926464 0.495247 0.440201 [Debug] proj_in_input: [192, 2170] first4: -0.124172 1.435321 0.310164 -0.624504 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.188757 0.875783 0.328665 -0.577830 [Debug] layer0_sa_input: [2048, 1085] first4: -0.918423 -0.718803 -0.038139 0.295172 [Debug] layer0_q_after_rope: [128, 16] first4: -1.808892 -1.387187 -0.135358 0.377958 [Debug] layer0_k_after_rope: [128, 8] first4: -0.188757 0.875783 0.328665 -0.577830 [Debug] layer0_sa_output: [2048, 1085] first4: -1.649195 0.808645 -0.556233 0.543696 [Debug] layer0_attn_out: [2048, 1085] first4: -1.808892 -1.387187 -0.135358 0.377958 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.808892 -1.387187 -0.135358 0.377958 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.971036 -1.049833 -0.413624 0.383357 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.557844 1.194601 58.143230 -1.642145 [Debug] hidden_after_layer6: [2048, 1085] first4: -17.473732 4.451920 57.714855 -2.524643 [Debug] hidden_after_layer12: [2048, 1085] first4: -9.789162 4.848970 -20.656143 -2.014215 [Debug] hidden_after_layer18: [2048, 1085] first4: -3.175152 18.849306 -35.539726 -11.244928 [Debug] hidden_after_layer23: [2048, 1085] first4: 33.378967 64.624985 59.612682 -9.689672 [Debug] dit_step0_vt: [2170, 64] first4: -0.581620 2.525226 -0.201663 1.465749 [Debug] dit_step0_xt: [2170, 64] first4: 0.205968 2.105746 -0.167842 0.818341 [DiT] Step 1/50 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.648619 2.536328 -0.028497 1.299685 [Debug] dit_step1_xt: [2170, 64] first4: 0.218941 2.055019 -0.167272 0.792348 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.646851 2.491205 0.005344 1.318794 [Debug] dit_step2_xt: [2170, 64] first4: 0.231878 2.005195 -0.167379 0.765972 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.624331 2.433946 0.026509 1.326102 [Debug] dit_step3_xt: [2170, 64] first4: 0.244364 1.956516 -0.167909 0.739450 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.568358 2.344923 0.006659 1.321841 [Debug] dit_step4_xt: [2170, 64] first4: 0.255732 1.909617 -0.168042 0.713013 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.540510 2.290946 -0.021219 1.309145 [Debug] dit_step5_xt: [2170, 64] first4: 0.266542 1.863798 -0.167618 0.686830 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.507919 2.227365 -0.053542 1.298645 [Debug] dit_step6_xt: [2170, 64] first4: 0.276700 1.819251 -0.166547 0.660857 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.463547 2.160049 -0.074018 1.283235 [Debug] dit_step7_xt: [2170, 64] first4: 0.285971 1.776050 -0.165066 0.635192 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.402033 2.064819 -0.087984 1.262099 [Debug] dit_step8_xt: [2170, 64] first4: 0.294012 1.734754 -0.163307 0.609950 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.338877 1.972970 -0.093641 1.244855 [Debug] dit_step9_xt: [2170, 64] first4: 0.300789 1.695294 -0.161434 0.585053 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.271688 1.889462 -0.077603 1.235510 [Debug] dit_step10_xt: [2170, 64] first4: 0.306223 1.657505 -0.159882 0.560343 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.232332 1.813437 -0.075055 1.234986 [Debug] dit_step11_xt: [2170, 64] first4: 0.310870 1.621236 -0.158381 0.535643 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.206181 1.742622 -0.087321 1.219010 [Debug] dit_step12_xt: [2170, 64] first4: 0.314993 1.586384 -0.156634 0.511263 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.193771 1.671303 -0.106323 1.200053 [Debug] dit_step13_xt: [2170, 64] first4: 0.318869 1.552958 -0.154508 0.487262 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.187309 1.591692 -0.131943 1.185596 [Debug] dit_step14_xt: [2170, 64] first4: 0.322615 1.521124 -0.151869 0.463550 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.185428 1.508886 -0.170669 1.165323 [Debug] dit_step15_xt: [2170, 64] first4: 0.326323 1.490946 -0.148456 0.440244 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.193209 1.436094 -0.196551 1.146308 [Debug] dit_step16_xt: [2170, 64] first4: 0.330188 1.462224 -0.144525 0.417318 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.205434 1.352405 -0.222775 1.127202 [Debug] dit_step17_xt: [2170, 64] first4: 0.334296 1.435176 -0.140069 0.394774 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.221665 1.275156 -0.251119 1.109041 [Debug] dit_step18_xt: [2170, 64] first4: 0.338730 1.409673 -0.135047 0.372593 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.234985 1.194670 -0.275361 1.084906 [Debug] dit_step19_xt: [2170, 64] first4: 0.343429 1.385780 -0.129540 0.350895 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.255574 1.104878 -0.290603 1.065150 [Debug] dit_step20_xt: [2170, 64] first4: 0.348541 1.363682 -0.123727 0.329592 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.273636 1.014015 -0.296465 1.041767 [Debug] dit_step21_xt: [2170, 64] first4: 0.354013 1.343402 -0.117798 0.308756 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.297965 0.911248 -0.309497 1.011165 [Debug] dit_step22_xt: [2170, 64] first4: 0.359973 1.325177 -0.111608 0.288533 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.325241 0.811368 -0.305610 0.988175 [Debug] dit_step23_xt: [2170, 64] first4: 0.366478 1.308950 -0.105496 0.268769 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.343456 0.703388 -0.311911 0.958141 [Debug] dit_step24_xt: [2170, 64] first4: 0.373347 1.294882 -0.099258 0.249607 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.368849 0.595716 -0.309047 0.938895 [Debug] dit_step25_xt: [2170, 64] first4: 0.380724 1.282968 -0.093077 0.230829 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.394844 0.475150 -0.305969 0.920393 [Debug] dit_step26_xt: [2170, 64] first4: 0.388621 1.273465 -0.086957 0.212421 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.417248 0.362810 -0.300485 0.898984 [Debug] dit_step27_xt: [2170, 64] first4: 0.396966 1.266208 -0.080948 0.194441 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.445670 0.238852 -0.291926 0.876388 [Debug] dit_step28_xt: [2170, 64] first4: 0.405879 1.261431 -0.075109 0.176913 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.458475 0.129535 -0.285232 0.856115 [Debug] dit_step29_xt: [2170, 64] first4: 0.415048 1.258841 -0.069405 0.159791 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.484824 0.007390 -0.269935 0.846297 [Debug] dit_step30_xt: [2170, 64] first4: 0.424745 1.258693 -0.064006 0.142865 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.501825 -0.106487 -0.258887 0.816058 [Debug] dit_step31_xt: [2170, 64] first4: 0.434781 1.260823 -0.058828 0.126544 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.516942 -0.222644 -0.242260 0.796909 [Debug] dit_step32_xt: [2170, 64] first4: 0.445120 1.265275 -0.053983 0.110606 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.527673 -0.332222 -0.229080 0.774601 [Debug] dit_step33_xt: [2170, 64] first4: 0.455674 1.271920 -0.049401 0.095114 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.537143 -0.437428 -0.205124 0.753138 [Debug] dit_step34_xt: [2170, 64] first4: 0.466417 1.280668 -0.045299 0.080051 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.541242 -0.533983 -0.179314 0.731150 [Debug] dit_step35_xt: [2170, 64] first4: 0.477241 1.291348 -0.041713 0.065428 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.553446 -0.634943 -0.159333 0.701408 [Debug] dit_step36_xt: [2170, 64] first4: 0.488310 1.304047 -0.038526 0.051400 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.557128 -0.716115 -0.136339 0.671350 [Debug] dit_step37_xt: [2170, 64] first4: 0.499453 1.318369 -0.035799 0.037973 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.552832 -0.801394 -0.100619 0.628371 [Debug] dit_step38_xt: [2170, 64] first4: 0.510509 1.334397 -0.033787 0.025406 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.544394 -0.885736 -0.068400 0.593040 [Debug] dit_step39_xt: [2170, 64] first4: 0.521397 1.352112 -0.032419 0.013545 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.535777 -0.946764 -0.041117 0.554355 [Debug] dit_step40_xt: [2170, 64] first4: 0.532113 1.371047 -0.031596 0.002458 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.525664 -1.016544 -0.002547 0.503790 [Debug] dit_step41_xt: [2170, 64] first4: 0.542626 1.391378 -0.031546 -0.007618 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.509587 -1.080180 0.041550 0.453302 [Debug] dit_step42_xt: [2170, 64] first4: 0.552818 1.412982 -0.032377 -0.016684 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.497659 -1.130402 0.068390 0.402836 [Debug] dit_step43_xt: [2170, 64] first4: 0.562771 1.435590 -0.033744 -0.024741 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.478302 -1.182549 0.115562 0.348239 [Debug] dit_step44_xt: [2170, 64] first4: 0.572337 1.459241 -0.036056 -0.031706 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.461027 -1.221343 0.152739 0.285886 [Debug] dit_step45_xt: [2170, 64] first4: 0.581558 1.483668 -0.039110 -0.037423 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.439156 -1.261086 0.181517 0.241726 [Debug] dit_step46_xt: [2170, 64] first4: 0.590341 1.508889 -0.042741 -0.042258 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.409888 -1.287494 0.193848 0.193589 [Debug] dit_step47_xt: [2170, 64] first4: 0.598539 1.534639 -0.046618 -0.046130 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.367127 -1.309679 0.211649 0.159172 [Debug] dit_step48_xt: [2170, 64] first4: 0.605881 1.560833 -0.050851 -0.049313 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.402161 -1.345310 0.192536 0.148982 [Debug] dit_x0: [2170, 64] first4: 0.613924 1.587739 -0.054701 -0.052293 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 115820.1 ms (115820.1 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.613924 1.587739 -0.054701 -0.052293 [Store] Unload DiT (3007.9 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: 27531.1 ms [Debug] vae_audio: [2, 4166400] first4: -0.001522 -0.000982 -0.001252 -0.000844 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 429 ms (202.18x realtime), 32 threads [MP3] Wrote ggml-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 08:20:24.428 | 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 08:20:26.937 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 08:20:26.937 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 08:20:26.937 | 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 08:20:26.941 | INFO | acestep.core.generation.handler.generate_music:_vram_preflight_check:156 - [generate_music] VRAM pre-flight: 82.30 GB free, ~0.94 GB needed (batch=1, duration=88s, mode=turbo). 2026-07-12 08:20:26.945 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 08:20:26.949 | 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 08:20:26.951 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 08:20:26.951 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 08:20:26.951 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 08:20:26.951 | 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 08:20:26.951 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 08:20:26.951 | 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 08:20:26.951 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 08:20:26.955 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 08:20:26.964 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 08:20:26.964 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 08:20:26.970 | 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-BF16.gguf [GGML] Running acestep-v15-turbo-BF16.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 0.999836 detok_output 0.999996 context 0.999997 noise 1.000000 temb_t 0.999999 hidden_after_proj_in 0.999988 enc_after_cond_emb 0.999829 layer0_sa_output 0.999960 hidden_after_layer0 0.999982 hidden_after_layer6 0.999923 hidden_after_layer12 0.999334 hidden_after_layer18 0.996714 hidden_after_layer23 0.993821 dit_step0_vt 0.975849 dit_step0_xt 0.999947 dit_step1_vt 0.979487 dit_step1_xt 0.999833 dit_step2_vt 0.981791 dit_step2_xt 0.999554 dit_step3_vt 0.982387 dit_step3_xt 0.998920 dit_step4_vt 0.980796 dit_step4_xt 0.997488 dit_step5_vt 0.978042 dit_step5_xt 0.994277 dit_step6_vt 0.974833 dit_step6_xt 0.988160 dit_step7_vt 0.969098 dit_x0 0.979123 vae_audio 0.901816 vae_audio (STFT cosine) 0.975801 [Turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999947 0.136835 0.006598 -0.002311 0.972945 -0.002342 0.972003 dit_step1_xt 0.999833 0.266113 0.011271 -0.005310 0.942680 -0.005313 0.941730 dit_step2_xt 0.999554 0.451848 0.017463 -0.009351 0.909200 -0.009311 0.908527 dit_step3_xt 0.998920 0.642987 0.025943 -0.014719 0.873841 -0.014577 0.873624 dit_step4_xt 0.997488 0.780718 0.037845 -0.021766 0.842029 -0.021660 0.841995 dit_step5_xt 0.994277 1.239408 0.055620 -0.031842 0.825348 -0.032109 0.824593 dit_step6_xt 0.988160 2.070782 0.082599 -0.046121 0.856209 -0.046482 0.855546 [SFT] steps=50, shift=1.0 | acestep-v15-sft-BF16.gguf [GGML] Running acestep-v15-sft-BF16.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-BF16.gguf: 830 tensors, data at offset 69344 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-BF16.gguf: 830 tensors, data at offset 69344 [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: 257 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-BF16.gguf: 830 tensors, data at offset 69344 [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, 1160.5 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L, CLS), text_proj, null_cond [Store] Load CondEnc: 928 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, 100.9 ms [Debug] enc_hidden: [238, 2048] first4: 0.686898 -0.019098 -0.051494 0.022655 [Store] Unload CondEnc (1160.5 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] Detokenizer backend: CPU (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-BF16.gguf: 830 tensors, data at offset 69344 [WeightCtx] Loaded 30 tensors, 200.3 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 112 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 905.1 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.124172 1.435321 0.310164 -0.624504 [Store] Unload FSQ-Detok (200.3 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: -0.124172 1.435321 0.310164 -0.624504 [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-BF16.gguf: 830 tensors, data at offset 69344 [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, 7955.6 MB into backend [Load] DiT: 32 layers, H=2560, Nh=32/8, D=128 [Store] Load DiT: 4936 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: 0.016426 0.108683 0.065360 -0.074070 [Debug] temb: [2560] first4: 0.026340 -0.023021 -0.016529 0.047201 [Debug] temb_t: [2560] first4: 0.014587 -0.022342 -0.020864 0.048816 [Debug] temb_r: [2560] first4: 0.011753 -0.000679 0.004335 -0.001615 [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: 0.063947 -0.019714 0.138310 -0.113208 [Debug] temb_lin1_r: [2560] first4: -0.009879 0.015926 -0.017671 0.012016 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.337150 -0.577257 0.230419 -0.190830 [Debug] proj_in_input: [192, 2170] first4: -0.124172 1.435321 0.310164 -0.624504 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.030978 0.651376 0.102890 0.871310 [Debug] layer0_sa_input: [2560, 1085] first4: 0.039979 -0.045431 0.093871 -0.134719 [Debug] layer0_q_after_rope: [128, 32] first4: -0.240838 -2.074609 0.340630 0.212767 [Debug] layer0_k_after_rope: [128, 8] first4: 0.020321 2.181354 0.699668 0.516729 [Debug] layer0_sa_output: [2560, 1085] first4: 1.632464 4.282640 -0.558102 5.413632 [Debug] layer0_attn_out: [4096, 1085] first4: 0.466511 -2.236335 -0.177157 -2.789362 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.466511 -2.236335 -0.177157 -2.789362 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.335877 -1.990901 -0.007252 -3.742178 [Debug] hidden_after_layer0: [2560, 1085] first4: 1.722611 -2.430920 241.825974 -3.714182 [Debug] hidden_after_layer6: [2560, 1085] first4: -6.174104 -2.155552 321.130463 -12.388517 [Debug] hidden_after_layer12: [2560, 1085] first4: -10.868447 29.452194 264.720306 28.747551 [Debug] hidden_after_layer18: [2560, 1085] first4: 42.068722 14.974079 278.738159 5.103224 [Debug] hidden_after_layer31: [2560, 1085] first4: -79.931030 -237.501785 -167.839706 174.702591 [Debug] dit_step0_vt: [2170, 64] first4: -1.215084 1.553296 0.396014 1.184635 [Debug] dit_step0_xt: [2170, 64] first4: 0.249567 2.085646 -0.189876 0.793809 [DiT] Step 1/8 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.664746 2.489604 0.842081 1.059206 [Debug] dit_step1_xt: [2170, 64] first4: 0.285826 1.949849 -0.235807 0.736034 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: -1.137094 2.213027 0.933505 1.018314 [Debug] dit_step2_xt: [2170, 64] first4: 0.361632 1.802314 -0.298041 0.668147 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: -1.359937 2.364017 1.099361 0.496145 [Debug] dit_step3_xt: [2170, 64] first4: 0.474960 1.605313 -0.389654 0.626801 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: -0.960422 2.206104 1.129881 0.204636 [Debug] dit_step4_xt: [2170, 64] first4: 0.577863 1.368944 -0.510713 0.604876 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: -1.175987 2.258690 0.890145 0.801434 [Debug] dit_step5_xt: [2170, 64] first4: 0.745861 1.046274 -0.637877 0.490385 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: -0.970684 2.009483 0.629019 0.824656 [Debug] dit_step6_xt: [2170, 64] first4: 0.939998 0.644378 -0.763680 0.325454 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: -1.036474 1.570236 0.639234 0.604436 [Debug] dit_x0: [2170, 64] first4: 1.250940 0.173307 -0.955451 0.144123 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 51135.0 ms (51135.0 ms/sample) [Debug] dit_output: [2170, 64] first4: 1.250940 0.173307 -0.955451 0.144123 [Store] Unload DiT (7955.6 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: 713 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: 27407.1 ms [Debug] vae_audio: [2, 4166400] first4: 0.001675 0.004377 0.004577 0.003906 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 515 ms (168.62x realtime), 32 threads [MP3] Wrote ggml-xl-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 08:21:57.253 | 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 0.999836 detok_output 0.999996 context 0.999997 noise 1.000000 temb_t 0.999998 hidden_after_proj_in 0.999988 enc_after_cond_emb 0.999829 layer0_sa_output 0.999957 hidden_after_layer0 0.999983 hidden_after_layer6 0.999852 hidden_after_layer12 0.999464 hidden_after_layer18 0.998728 hidden_after_layer23 0.998993 dit_step0_vt 0.998957 dit_step0_xt 1.000000 dit_step5_vt 0.999521 dit_step5_xt 0.999981 dit_step10_vt 0.999207 dit_step10_xt 0.999939 dit_step15_vt 0.998363 dit_step15_xt 0.999859 dit_step20_vt 0.996599 dit_step20_xt 0.999648 dit_step25_vt 0.993454 dit_step25_xt 0.999211 dit_step30_vt 0.989315 dit_step30_xt 0.998389 dit_step35_vt 0.984870 dit_step35_xt 0.997223 dit_step40_vt 0.982111 dit_step40_xt 0.995952 dit_step45_vt 0.985428 dit_step45_xt 0.995002 dit_step49_vt 0.990488 dit_x0 0.994665 vae_audio 0.971593 vae_audio (STFT cosine) 0.988878 [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 0.019908 0.001449 -0.001655 0.978462 -0.001619 0.978872 dit_step5_xt 0.999981 0.078142 0.004202 -0.008401 0.884789 -0.008480 0.883811 dit_step10_xt 0.999939 0.116878 0.006627 -0.015154 0.800968 -0.015075 0.802324 dit_step15_xt 0.999859 0.163477 0.009242 -0.021882 0.730558 -0.021933 0.731304 dit_step20_xt 0.999648 0.223157 0.013524 -0.028646 0.677786 -0.028882 0.679066 dit_step25_xt 0.999211 0.329346 0.019215 -0.035480 0.647080 -0.035643 0.650780 dit_step30_xt 0.998389 0.564806 0.026849 -0.042358 0.641704 -0.042577 0.647107 dit_step35_xt 0.997223 0.825694 0.036007 -0.049246 0.662387 -0.049471 0.669256 dit_step40_xt 0.995952 1.086012 0.046274 -0.056164 0.706796 -0.056532 0.715783 dit_step45_xt 0.995002 1.316992 0.055950 -0.063188 0.770647 -0.063533 0.780575 [Xl-turbo] steps=8, shift=3.0 | acestep-v15-xl-turbo-BF16.gguf [GGML] Running acestep-v15-xl-turbo-BF16.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 08:22:02.832 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 08:22:02.832 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 08:22:02.833 | 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 08:22:02.836 | INFO | acestep.core.generation.handler.generate_music:_vram_preflight_check:156 - [generate_music] VRAM pre-flight: 77.36 GB free, ~1.23 GB needed (batch=1, duration=88s, mode=xl_turbo). 2026-07-12 08:22:02.841 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 08:22:02.846 | 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 08:22:02.847 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 08:22:02.847 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 08:22:02.847 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 08:22:02.847 | 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 08:22:02.847 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 08:22:02.847 | 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 08:22:02.847 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 08:22:02.851 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 08:22:02.858 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 08:22:02.859 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 08:22:02.865 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 08:22:02.872 | 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 08:22:03.418 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 08:22:03.419 | 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 08:22:03.419 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.006819009780883789, 'diffusion_time_cost': 0.545757532119751, 'diffusion_per_step_time_cost': 0.06821969151496887, 'total_time_cost': 0.5525765419006348, 'offload_time_cost': 0.0} 2026-07-12 08:22:03.441 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 08:22:03.443 | 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 08:22:03.443 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:154 - [generate_music] Effective free VRAM before VAE decode: 77.27 GB 2026-07-12 08:22:03.443 | 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 08:22:03.443 | DEBUG | acestep.core.generation.handler.memory_utils:_get_auto_decode_chunk_size:108 - [_get_auto_decode_chunk_size] Effective free VRAM: 77.27 GB 2026-07-12 08:22:03.443 | DEBUG | acestep.core.generation.handler.memory_utils:_should_offload_wav_to_cpu:131 - [_should_offload_wav_to_cpu] Effective free VRAM: 77.27 GB 2026-07-12 08:22:03.443 | 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 08:22:03.714 | 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 08:22:03.919 | 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 08:22:03.922 | 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-BF16.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-BF16.gguf: 830 tensors, data at offset 69312 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-BF16.gguf: 830 tensors, data at offset 69312 [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: 268 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-BF16.gguf: 830 tensors, data at offset 69312 [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, 1160.5 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L, CLS), text_proj, null_cond [Store] Load CondEnc: 804 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, 96.4 ms [Debug] enc_hidden: [238, 2048] first4: 0.686898 -0.019098 -0.051494 0.022655 [Store] Unload CondEnc (1160.5 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] Detokenizer backend: CPU (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-BF16.gguf: 830 tensors, data at offset 69312 [WeightCtx] Loaded 30 tensors, 200.3 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 111 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 911.3 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.124172 1.435321 0.310164 -0.624504 [Store] Unload FSQ-Detok (200.3 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: -0.124172 1.435321 0.310164 -0.624504 [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-BF16.gguf: 830 tensors, data at offset 69312 [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, 7955.6 MB into backend [Load] DiT: 32 layers, H=2560, Nh=32/8, D=128 [Store] Load DiT: 5273 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: 0.012579 0.105540 0.073639 -0.031893 [Debug] temb: [2560] first4: 0.009506 -0.007716 -0.006105 0.004584 [Debug] temb_t: [2560] first4: 0.007969 -0.007948 -0.006674 0.003606 [Debug] temb_r: [2560] first4: 0.001538 0.000232 0.000569 0.000978 [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: 0.032161 -0.082097 0.088242 -0.064953 [Debug] temb_lin1_r: [2560] first4: 0.006956 -0.005390 -0.002806 -0.011722 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.326588 -0.492784 0.101537 -0.214441 [Debug] proj_in_input: [192, 2170] first4: -0.124172 1.435321 0.310164 -0.624504 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.035580 0.640036 0.108195 0.870601 [Debug] layer0_sa_input: [2560, 1085] first4: 0.036394 -0.019050 0.098659 -0.097736 [Debug] layer0_q_after_rope: [128, 32] first4: -0.273283 -2.149641 0.304081 -0.031929 [Debug] layer0_k_after_rope: [128, 8] first4: 0.727837 2.921756 1.168267 2.043445 [Debug] layer0_sa_output: [2560, 1085] first4: 1.948995 4.407659 -0.829639 7.829544 [Debug] layer0_attn_out: [4096, 1085] first4: 0.457383 -2.189255 -0.516980 -4.126828 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.457383 -2.189255 -0.516980 -4.126828 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.293133 -1.919286 -0.432752 -4.233349 [Debug] hidden_after_layer0: [2560, 1085] first4: -1.838778 -1.227677 248.406464 -3.998169 [Debug] hidden_after_layer6: [2560, 1085] first4: -7.859870 3.671378 316.644318 -21.743362 [Debug] hidden_after_layer12: [2560, 1085] first4: -23.728960 39.843269 233.167023 16.567801 [Debug] hidden_after_layer18: [2560, 1085] first4: 9.443115 -13.155821 234.980499 19.720192 [Debug] hidden_after_layer31: [2560, 1085] first4: 63.239422 -107.273117 7.537132 89.375351 [Debug] dit_step0_vt: [2170, 64] first4: -0.321423 2.209531 0.011965 1.702935 [Debug] dit_step0_xt: [2170, 64] first4: 0.200764 2.112059 -0.172114 0.813598 [DiT] Step 1/50 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.348141 2.148278 0.068977 1.556879 [Debug] dit_step1_xt: [2170, 64] first4: 0.207727 2.069094 -0.173494 0.782460 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.325172 2.168985 0.049205 1.567360 [Debug] dit_step2_xt: [2170, 64] first4: 0.214231 2.025714 -0.174478 0.751113 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.280391 2.135005 0.045635 1.567819 [Debug] dit_step3_xt: [2170, 64] first4: 0.219838 1.983014 -0.175391 0.719756 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.269533 2.069282 0.020692 1.584703 [Debug] dit_step4_xt: [2170, 64] first4: 0.225229 1.941628 -0.175804 0.688062 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.260456 2.048395 0.028222 1.590406 [Debug] dit_step5_xt: [2170, 64] first4: 0.230438 1.900661 -0.176369 0.656254 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.251416 2.018544 0.065080 1.605371 [Debug] dit_step6_xt: [2170, 64] first4: 0.235467 1.860290 -0.177671 0.624147 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.249728 1.984772 0.109279 1.615452 [Debug] dit_step7_xt: [2170, 64] first4: 0.240461 1.820594 -0.179856 0.591838 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.263806 1.937999 0.139662 1.608950 [Debug] dit_step8_xt: [2170, 64] first4: 0.245737 1.781834 -0.182649 0.559659 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.283624 1.876493 0.128396 1.603702 [Debug] dit_step9_xt: [2170, 64] first4: 0.251410 1.744304 -0.185217 0.527585 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.291585 1.833047 0.119359 1.598635 [Debug] dit_step10_xt: [2170, 64] first4: 0.257241 1.707643 -0.187604 0.495612 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.291233 1.787727 0.104628 1.586840 [Debug] dit_step11_xt: [2170, 64] first4: 0.263066 1.671889 -0.189697 0.463875 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.284824 1.737530 0.092142 1.573861 [Debug] dit_step12_xt: [2170, 64] first4: 0.268763 1.637138 -0.191540 0.432398 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.278503 1.690122 0.076445 1.557655 [Debug] dit_step13_xt: [2170, 64] first4: 0.274333 1.603336 -0.193069 0.401245 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.271321 1.633191 0.055675 1.556922 [Debug] dit_step14_xt: [2170, 64] first4: 0.279759 1.570672 -0.194182 0.370106 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.263431 1.585763 0.039728 1.552130 [Debug] dit_step15_xt: [2170, 64] first4: 0.285028 1.538957 -0.194977 0.339064 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.237744 1.540442 0.016469 1.547803 [Debug] dit_step16_xt: [2170, 64] first4: 0.289783 1.508148 -0.195306 0.308108 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.215681 1.491821 0.000529 1.540475 [Debug] dit_step17_xt: [2170, 64] first4: 0.294096 1.478311 -0.195317 0.277298 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.184768 1.438839 -0.011662 1.531355 [Debug] dit_step18_xt: [2170, 64] first4: 0.297792 1.449535 -0.195084 0.246671 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.153401 1.393398 -0.024278 1.522411 [Debug] dit_step19_xt: [2170, 64] first4: 0.300860 1.421667 -0.194598 0.216223 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.123694 1.342377 -0.032353 1.512923 [Debug] dit_step20_xt: [2170, 64] first4: 0.303333 1.394819 -0.193951 0.185965 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.089509 1.290682 -0.041349 1.500790 [Debug] dit_step21_xt: [2170, 64] first4: 0.305124 1.369006 -0.193124 0.155949 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.067453 1.249073 -0.052924 1.489589 [Debug] dit_step22_xt: [2170, 64] first4: 0.306473 1.344024 -0.192065 0.126157 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.054616 1.191301 -0.064074 1.473606 [Debug] dit_step23_xt: [2170, 64] first4: 0.307565 1.320198 -0.190784 0.096685 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.041831 1.137526 -0.071955 1.454951 [Debug] dit_step24_xt: [2170, 64] first4: 0.308402 1.297448 -0.189345 0.067586 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.028231 1.084667 -0.081016 1.437627 [Debug] dit_step25_xt: [2170, 64] first4: 0.308966 1.275754 -0.187725 0.038833 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.028279 1.018891 -0.090530 1.417422 [Debug] dit_step26_xt: [2170, 64] first4: 0.309532 1.255376 -0.185914 0.010485 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.031917 0.966425 -0.105896 1.394005 [Debug] dit_step27_xt: [2170, 64] first4: 0.310170 1.236048 -0.183796 -0.017395 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.038651 0.897976 -0.119445 1.364133 [Debug] dit_step28_xt: [2170, 64] first4: 0.310943 1.218088 -0.181407 -0.044678 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.046172 0.823057 -0.134399 1.340340 [Debug] dit_step29_xt: [2170, 64] first4: 0.311867 1.201627 -0.178719 -0.071485 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.064703 0.751143 -0.146440 1.314089 [Debug] dit_step30_xt: [2170, 64] first4: 0.313161 1.186604 -0.175790 -0.097767 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.073931 0.668391 -0.158181 1.284853 [Debug] dit_step31_xt: [2170, 64] first4: 0.314639 1.173237 -0.172627 -0.123464 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.092794 0.586252 -0.167332 1.263310 [Debug] dit_step32_xt: [2170, 64] first4: 0.316495 1.161512 -0.169280 -0.148730 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.109193 0.504978 -0.177490 1.234948 [Debug] dit_step33_xt: [2170, 64] first4: 0.318679 1.151412 -0.165730 -0.173429 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.126408 0.414742 -0.178863 1.207128 [Debug] dit_step34_xt: [2170, 64] first4: 0.321207 1.143117 -0.162153 -0.197571 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.151068 0.304825 -0.180338 1.177581 [Debug] dit_step35_xt: [2170, 64] first4: 0.324228 1.137021 -0.158546 -0.221123 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.166092 0.212327 -0.180307 1.138870 [Debug] dit_step36_xt: [2170, 64] first4: 0.327550 1.132774 -0.154940 -0.243900 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.162245 0.119998 -0.173204 1.109470 [Debug] dit_step37_xt: [2170, 64] first4: 0.330795 1.130374 -0.151476 -0.266090 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.166531 0.015829 -0.159653 1.058953 [Debug] dit_step38_xt: [2170, 64] first4: 0.334126 1.130058 -0.148283 -0.287269 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.182369 -0.091567 -0.147169 1.028989 [Debug] dit_step39_xt: [2170, 64] first4: 0.337773 1.131889 -0.145340 -0.307849 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.209654 -0.199956 -0.130185 0.987970 [Debug] dit_step40_xt: [2170, 64] first4: 0.341966 1.135888 -0.142736 -0.327608 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.238116 -0.303977 -0.099801 0.940686 [Debug] dit_step41_xt: [2170, 64] first4: 0.346729 1.141968 -0.140740 -0.346422 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.252346 -0.391197 -0.075406 0.896112 [Debug] dit_step42_xt: [2170, 64] first4: 0.351776 1.149792 -0.139232 -0.364344 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.283108 -0.488520 -0.036481 0.847143 [Debug] dit_step43_xt: [2170, 64] first4: 0.357438 1.159562 -0.138502 -0.381287 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.311389 -0.581431 0.013386 0.797240 [Debug] dit_step44_xt: [2170, 64] first4: 0.363665 1.171191 -0.138770 -0.397232 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.344210 -0.650828 0.056550 0.745599 [Debug] dit_step45_xt: [2170, 64] first4: 0.370550 1.184207 -0.139901 -0.412144 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.390258 -0.726730 0.105948 0.703666 [Debug] dit_step46_xt: [2170, 64] first4: 0.378355 1.198742 -0.142020 -0.426217 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.438016 -0.804332 0.164782 0.663591 [Debug] dit_step47_xt: [2170, 64] first4: 0.387115 1.214828 -0.145315 -0.439489 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.494025 -0.864318 0.241254 0.653116 [Debug] dit_step48_xt: [2170, 64] first4: 0.396996 1.232115 -0.150141 -0.452551 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.485023 -0.850604 0.167278 0.666200 [Debug] dit_x0: [2170, 64] first4: 0.406696 1.249127 -0.153486 -0.465875 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 325146.0 ms (325146.0 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.406696 1.249127 -0.153486 -0.465875 [Store] Unload DiT (7955.6 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: 711 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: 27721.1 ms [Debug] vae_audio: [2, 4166400] first4: -0.001383 -0.000662 -0.000767 -0.000288 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 460 ms (188.66x realtime), 32 threads [MP3] Wrote ggml-xl-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 08:28:06.557 | 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 0.999574 detok_output 0.999996 context 0.999997 noise 1.000000 temb_t 0.999995 hidden_after_proj_in 0.999980 enc_after_cond_emb 0.999693 layer0_sa_output 0.999651 hidden_after_layer0 0.999433 hidden_after_layer6 0.999823 hidden_after_layer12 0.997282 hidden_after_layer18 0.996205 hidden_after_layer31 0.988315 dit_step0_vt 0.912461 dit_step0_xt 0.999842 dit_step1_vt 0.931192 dit_step1_xt 0.999541 dit_step2_vt 0.933803 dit_step2_xt 0.998823 dit_step3_vt 0.929596 dit_step3_xt 0.997086 dit_step4_vt 0.923362 dit_step4_xt 0.992719 dit_step5_vt 0.914847 dit_step5_xt 0.982235 dit_step6_vt 0.907437 dit_step6_xt 0.961513 dit_step7_vt 0.887285 dit_x0 0.927893 vae_audio 0.734169 vae_audio (STFT cosine) 0.923323 [Xl-turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999842 0.159259 0.012448 -0.002991 0.978362 -0.002721 0.976534 dit_step1_xt 0.999541 0.371954 0.021009 -0.005948 0.952133 -0.005567 0.949004 dit_step2_xt 0.998823 0.545209 0.032087 -0.009744 0.923230 -0.009330 0.918931 dit_step3_xt 0.997086 0.809188 0.047817 -0.015394 0.891898 -0.014418 0.887241 dit_step4_xt 0.992719 1.177347 0.072342 -0.023152 0.862755 -0.021453 0.857796 dit_step5_xt 0.982235 1.745782 0.110621 -0.034669 0.846415 -0.032240 0.840033 dit_step6_xt 0.961513 2.505169 0.168418 -0.051763 0.870650 -0.049194 0.863964 [Xl-sft] steps=50, shift=1.0 | acestep-v15-xl-sft-BF16.gguf [GGML] Running acestep-v15-xl-sft-BF16.gguf... [GGML] Done, 131 dump files [Python] Initializing acestep-v15-xl-sft... Loading checkpoint shards: 0%| | 0/4 [00:00 2026-07-12 08:28:12.410 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 08:28:12.410 | 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 08:28:12.410 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 08:28:12.413 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 08:28:12.421 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 08:28:12.421 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 08:28:12.427 | 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