[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: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/Qwen3-Embedding-0.6B-BF16.gguf: 310 tensors, data at offset 5337568 [Load] TextEncoder: 28L, H=1024, Nh=16/8 [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 310 tensors, 1136.5 MB into backend [Store] Load TextEnc: 420 ms [Debug] text_hidden: [70, 1024] first4: 3.705836 2.395382 0.221845 -13.145830 [Debug] lyric_embed: [167, 1024] first4: 0.029175 0.032227 -0.022339 -0.028809 [Store] Unload TextEnc (1136.5 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] CondEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-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: 797 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, 1001.4 ms [Debug] enc_hidden: [238, 2048] first4: 1.758251 -0.049636 -0.132758 0.058457 [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: Vulkan0 (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: 131 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 2848.2 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.125156 1.434936 0.308101 -0.624459 [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: -0.125156 1.434936 0.308101 -0.624459 [Init-Noise] Starting: T=2170, S=1085, enc_S=238, steps=8, batch=1 (cover) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] DiT backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-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: 1681 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: 0.260062 -0.161562 -0.097030 0.052313 [Debug] temb: [2048] first4: 0.000069 -0.132499 -0.035430 0.064753 [Debug] temb_t: [2048] first4: 0.001065 0.026818 -0.052754 0.063717 [Debug] temb_r: [2048] first4: -0.000996 -0.159317 0.017323 0.001036 [Debug] sinusoidal_t: [256] first4: 0.562486 0.789701 0.439822 -0.023583 [Debug] sinusoidal_r: [256] first4: 1.000000 1.000000 1.000000 1.000000 [Debug] temb_lin1_t: [2048] first4: -0.049318 -0.051829 -0.014251 -0.038444 [Debug] temb_lin1_r: [2048] first4: -0.013266 -0.018319 -0.016375 0.008532 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.039551 -0.968811 0.536133 0.446747 [Debug] proj_in_input: [192, 2170] first4: -0.125156 1.434936 0.308101 -0.624459 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.168520 0.815133 0.327705 -0.561921 [Debug] layer0_sa_input: [2048, 1085] first4: -0.719109 -0.763732 -0.047328 0.261808 [Debug] layer0_q_after_rope: [128, 16] first4: -1.541005 -1.043474 0.190447 0.456097 [Debug] layer0_k_after_rope: [128, 8] first4: -0.168520 0.815133 0.327705 -0.561921 [Debug] layer0_sa_output: [2048, 1085] first4: -1.501962 0.168340 -0.354029 0.513214 [Debug] layer0_attn_out: [2048, 1085] first4: -1.541005 -1.043474 0.190447 0.456097 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.541005 -1.043474 0.190447 0.456097 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.597807 -0.819423 -0.287355 0.493105 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.086174 0.568512 52.464794 -0.905845 [Debug] hidden_after_layer6: [2048, 1085] first4: -21.310276 0.154845 33.490002 -4.489417 [Debug] hidden_after_layer12: [2048, 1085] first4: -15.196171 -17.842777 72.516197 28.803352 [Debug] hidden_after_layer18: [2048, 1085] first4: -27.642179 15.526653 60.733097 20.945080 [Debug] hidden_after_layer23: [2048, 1085] first4: -11.903603 46.295460 196.521896 143.028198 [Debug] dit_step0_vt: [2170, 64] first4: 0.025665 1.110815 0.342369 2.378593 [Debug] dit_step0_xt: [2170, 64] first4: 0.193169 2.105758 -0.187437 0.739538 [DiT] Step 1/8 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.229782 1.295525 -0.131083 1.890993 [Debug] dit_step1_xt: [2170, 64] first4: 0.205703 2.035094 -0.180287 0.636393 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: -0.016106 1.209785 0.106354 2.391113 [Debug] dit_step2_xt: [2170, 64] first4: 0.206777 1.954441 -0.187378 0.476986 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: 0.249237 1.091537 0.260468 2.648376 [Debug] dit_step3_xt: [2170, 64] first4: 0.186007 1.863480 -0.209083 0.256288 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: 0.291870 1.000519 0.107147 2.705139 [Debug] dit_step4_xt: [2170, 64] first4: 0.154735 1.756281 -0.220563 -0.033549 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: 0.266418 0.923019 -0.294495 2.760838 [Debug] dit_step5_xt: [2170, 64] first4: 0.116675 1.624421 -0.178493 -0.427954 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: 0.176162 0.640289 -0.822159 2.826843 [Debug] dit_step6_xt: [2170, 64] first4: 0.081443 1.496364 -0.014061 -0.993323 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: -0.004242 0.179077 -1.457825 3.122925 [Debug] dit_x0: [2170, 64] first4: 0.082716 1.442640 0.423287 -1.930200 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 1264.5 ms (1264.5 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.082716 1.442640 0.423287 -1.930200 [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: Vulkan0 (CPU threads: 16) [VAE] Backend: Vulkan0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 669 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: 2169.1 ms [Debug] vae_audio: [2, 4166400] first4: 0.000702 0.001420 0.001181 0.001716 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 451 ms (192.42x realtime), 32 threads [MP3] Wrote ggml-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 10:15:03.143 | DEBUG | acestep.core.generation.handler.init_service_memory_basic:_apply_malloc_mmap_threshold:50 - [memory] Set M_MMAP_THRESHOLD=131072 for immediate OS reclaim of large frees Skipping import of cpp extensions due to incompatible torch version. Please upgrade to torch >= 2.11.0 (found 2.10.0+cu128). 2026-07-12 10:15:05.102 | WARNING | acestep.training.trainer::40 - bitsandbytes not installed. Using standard AdamW. 2026-07-12 10:15:05.107 | INFO | acestep.gpu_config:get_gpu_memory_gb:578 - CUDA GPU detected: NVIDIA RTX PRO 6000 Blackwell Workstation Edition (95.0 GB) 2026-07-12 10:15:05.175 | INFO | acestep.core.generation.handler.init_service_loader:_load_main_model_from_checkpoint:174 - [initialize_service] Attempting to load model with attention implementation: sdpa Unable to import `torchao` Tensor objects. This may affect loading checkpoints serialized with `torchao` 2026-07-12 10:15:07.813 | INFO | acestep.core.generation.handler.generate_music:generate_music:289 - [generate_music] Turbo model detected: overriding guidance_scale 0.0 -> 1.0 (turbo does not use CFG). 2026-07-12 10:15:07.813 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:15:07.813 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:15:07.814 | INFO | acestep.core.generation.handler.generate_music_request:_prepare_reference_and_source_audio:179 - [generate_music] cover task: no src_audio but audio codes provided, proceeding with codes 2026-07-12 10:15:07.817 | INFO | acestep.core.generation.handler.generate_music:_vram_preflight_check:156 - [generate_music] VRAM pre-flight: 87.83 GB free, ~0.94 GB needed (batch=1, duration=88s, mode=turbo). 2026-07-12 10:15:07.823 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:15:08.028 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_precomputed_lm_hints:31 - [generate_music] Decoding audio codes for LM hints for item 0... 2026-07-12 10:15:08.030 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:15:08.030 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:15:08.030 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:15:08.030 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:122 - text_prompt: # Instruction Generate audio semantic tokens based on the given conditions: # Caption An upbeat and anthemic pop-rock track driven by bright, slightly overdriven # Metas - bpm: 83 - timesignature: 4 - keyscale: G major - duration: 88 seconds <|endoftext|> 2026-07-12 10:15:08.030 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:15:08.030 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:124 - lyrics_text: # Languages fr # Lyric # Lyric [Intro - Guitar Riff] [Verse 1] Dans le monde des tutos virtuels G ta toise en nouvelle passion Avec Ggendoline et PumbΓ© Γ  midi La communautΓ©, c'est l'unitΓ© Quel joie, une clΓ© [Chorus] Dans le monde des tutos virtuels GΓ’ndoline et PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner [Guitar Solo] [Verse 2] Dans le monde des tutos virtuels GΓ’ndoline, PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner<|endoftext|> 2026-07-12 10:15:08.030 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:15:08.038 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:15:08.052 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:15:08.052 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:15:08.080 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 10:15:08.087 | INFO | acestep.models.common.dcw_correction:__init__:98 - [DCW] Active β€” mode=double, scaler=0.0500, high_scaler=0.0200, wavelet='haar' 2026-07-12 10:15:08.172 | WARNING | acestep.models.common.dcw_loader:_try_import:39 - DCW is enabled but 'pytorch_wavelets' is not installed. Install with `pip install pytorch_wavelets PyWavelets` to use Differential Correction in Wavelet domain. Falling back to no-op for this generation. 2026-07-12 10:15:08.402 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 10:15:08.404 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:64 - [generate_music] pred_latents: torch.Size([1, 2170, 64]), dtype=torch.bfloat16 2026-07-12 10:15:08.404 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.006848335266113281, 'diffusion_time_cost': 0.31514811515808105, 'diffusion_per_step_time_cost': 0.03939351439476013, 'total_time_cost': 0.32199645042419434, 'offload_time_cost': 0.0} 2026-07-12 10:15:08.417 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 10:15:08.420 | DEBUG | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:136 - [generate_music] Before VAE decode: allocated=5.98GB, max=7.29GB 2026-07-12 10:15:08.420 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:154 - [generate_music] Effective free VRAM before VAE decode: 87.73 GB 2026-07-12 10:15:08.420 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:172 - [generate_music] Using tiled VAE decode to reduce VRAM usage... 2026-07-12 10:15:08.420 | DEBUG | acestep.core.generation.handler.memory_utils:_get_auto_decode_chunk_size:108 - [_get_auto_decode_chunk_size] Effective free VRAM: 87.73 GB 2026-07-12 10:15:08.420 | DEBUG | acestep.core.generation.handler.memory_utils:_should_offload_wav_to_cpu:131 - [_should_offload_wav_to_cpu] Effective free VRAM: 87.73 GB 2026-07-12 10:15:08.420 | INFO | acestep.core.generation.handler.vae_decode:tiled_decode:56 - [tiled_decode] chunk_size=512, offload_wav_to_cpu=False, latents_shape=torch.Size([1, 64, 2170]) 2026-07-12 10:15:08.696 | DEBUG | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:193 - [generate_music] After VAE decode: allocated=6.15GB, max=7.44GB 2026-07-12 10:15:08.856 | INFO | acestep.core.generation.handler.generate_music_payload:_build_generate_music_success_payload:38 - [generate_music] VAE decode completed. Preparing audio tensors... 2026-07-12 10:15:08.859 | 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: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/Qwen3-Embedding-0.6B-BF16.gguf: 310 tensors, data at offset 5337568 [Load] TextEncoder: 28L, H=1024, Nh=16/8 [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 310 tensors, 1136.5 MB into backend [Store] Load TextEnc: 296 ms [Debug] text_hidden: [70, 1024] first4: 3.705836 2.395382 0.221845 -13.145830 [Debug] lyric_embed: [167, 1024] first4: 0.029175 0.032227 -0.022339 -0.028809 [Store] Unload TextEnc (1136.5 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] CondEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-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: 775 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, 11.3 ms [Debug] enc_hidden: [238, 2048] first4: 1.758251 -0.049636 -0.132758 0.058457 [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: Vulkan0 (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: 111 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 182.5 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.125156 1.434936 0.308101 -0.624459 [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: -0.125156 1.434936 0.308101 -0.624459 [Init-Noise] Starting: T=2170, S=1085, enc_S=238, steps=50, batch=1 (cover) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] DiT backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-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: 1557 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: 0.154506 -0.116115 -0.086937 0.082315 [Debug] temb: [2048] first4: -0.002757 -0.176460 0.004187 -0.001983 [Debug] temb_t: [2048] first4: -0.001235 0.003330 -0.013135 -0.002086 [Debug] temb_r: [2048] first4: -0.001522 -0.179790 0.017322 0.000103 [Debug] sinusoidal_t: [256] first4: 0.562486 0.789701 0.439822 -0.023583 [Debug] sinusoidal_r: [256] first4: 1.000000 1.000000 1.000000 1.000000 [Debug] temb_lin1_t: [2048] first4: -0.039237 0.029644 0.028280 -0.024901 [Debug] temb_lin1_r: [2048] first4: 0.002274 0.002999 -0.001040 -0.000388 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.089844 -0.924561 0.498047 0.441437 [Debug] proj_in_input: [192, 2170] first4: -0.125156 1.434936 0.308101 -0.624459 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.188605 0.876239 0.329525 -0.576920 [Debug] layer0_sa_input: [2048, 1085] first4: -0.915453 -0.715979 -0.037661 0.294912 [Debug] layer0_q_after_rope: [128, 16] first4: -1.802135 -1.382284 -0.129866 0.380222 [Debug] layer0_k_after_rope: [128, 8] first4: -0.188605 0.876239 0.329525 -0.576920 [Debug] layer0_sa_output: [2048, 1085] first4: -1.645108 0.807896 -0.555700 0.545140 [Debug] layer0_attn_out: [2048, 1085] first4: -1.802135 -1.382284 -0.129866 0.380222 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.802135 -1.382284 -0.129866 0.380222 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.962886 -1.045038 -0.403431 0.386077 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.439390 1.158046 57.794827 -1.609144 [Debug] hidden_after_layer6: [2048, 1085] first4: -17.244553 4.508382 57.686226 -2.312790 [Debug] hidden_after_layer12: [2048, 1085] first4: -10.055096 4.856796 -21.509155 -1.868328 [Debug] hidden_after_layer18: [2048, 1085] first4: -3.366667 18.997288 -36.617020 -10.940506 [Debug] hidden_after_layer23: [2048, 1085] first4: 34.113728 63.109352 57.997849 -9.051547 [Debug] dit_step0_vt: [2170, 64] first4: -0.575928 2.522217 -0.199131 1.470572 [Debug] dit_step0_xt: [2170, 64] first4: 0.205854 2.105806 -0.167892 0.818245 [DiT] Step 1/50 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.644287 2.529297 -0.031433 1.302032 [Debug] dit_step1_xt: [2170, 64] first4: 0.218740 2.055220 -0.167264 0.792204 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.639160 2.487183 -0.000061 1.316925 [Debug] dit_step2_xt: [2170, 64] first4: 0.231523 2.005476 -0.167262 0.765866 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.619324 2.424561 0.023773 1.327393 [Debug] dit_step3_xt: [2170, 64] first4: 0.243910 1.956985 -0.167738 0.739318 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.572021 2.348633 0.009583 1.321716 [Debug] dit_step4_xt: [2170, 64] first4: 0.255350 1.910012 -0.167930 0.712883 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.527100 2.278442 -0.017960 1.306213 [Debug] dit_step5_xt: [2170, 64] first4: 0.265892 1.864443 -0.167570 0.686759 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.499329 2.223877 -0.051697 1.299225 [Debug] dit_step6_xt: [2170, 64] first4: 0.275879 1.819966 -0.166536 0.660775 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.462708 2.158325 -0.074280 1.285736 [Debug] dit_step7_xt: [2170, 64] first4: 0.285133 1.776799 -0.165051 0.635060 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.394653 2.068115 -0.084400 1.269592 [Debug] dit_step8_xt: [2170, 64] first4: 0.293026 1.735437 -0.163363 0.609668 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.334473 1.976746 -0.086823 1.252640 [Debug] dit_step9_xt: [2170, 64] first4: 0.299716 1.695902 -0.161626 0.584615 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.274414 1.890320 -0.071671 1.243958 [Debug] dit_step10_xt: [2170, 64] first4: 0.305204 1.658095 -0.160193 0.559736 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.232849 1.816833 -0.063934 1.236954 [Debug] dit_step11_xt: [2170, 64] first4: 0.309861 1.621759 -0.158914 0.534997 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.204773 1.743500 -0.078125 1.225708 [Debug] dit_step12_xt: [2170, 64] first4: 0.313956 1.586889 -0.157352 0.510483 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.192215 1.667847 -0.093628 1.207764 [Debug] dit_step13_xt: [2170, 64] first4: 0.317801 1.553532 -0.155479 0.486328 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.186516 1.591705 -0.120575 1.189987 [Debug] dit_step14_xt: [2170, 64] first4: 0.321531 1.521698 -0.153068 0.462528 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.185950 1.517822 -0.157227 1.175064 [Debug] dit_step15_xt: [2170, 64] first4: 0.325250 1.491341 -0.149923 0.439027 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.190407 1.431000 -0.183105 1.153824 [Debug] dit_step16_xt: [2170, 64] first4: 0.329058 1.462721 -0.146261 0.415950 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.204529 1.353081 -0.211243 1.135815 [Debug] dit_step17_xt: [2170, 64] first4: 0.333149 1.435660 -0.142036 0.393234 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.219879 1.273682 -0.237976 1.124039 [Debug] dit_step18_xt: [2170, 64] first4: 0.337546 1.410186 -0.137277 0.370753 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.234909 1.190720 -0.259460 1.099720 [Debug] dit_step19_xt: [2170, 64] first4: 0.342244 1.386372 -0.132088 0.348759 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.252640 1.108459 -0.269531 1.077709 [Debug] dit_step20_xt: [2170, 64] first4: 0.347297 1.364203 -0.126697 0.327205 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.277634 1.012421 -0.283142 1.046616 [Debug] dit_step21_xt: [2170, 64] first4: 0.352850 1.343954 -0.121034 0.306272 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.301834 0.911804 -0.287781 1.018784 [Debug] dit_step22_xt: [2170, 64] first4: 0.358887 1.325718 -0.115278 0.285897 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.330811 0.810608 -0.301086 0.996017 [Debug] dit_step23_xt: [2170, 64] first4: 0.365503 1.309506 -0.109257 0.265976 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.349976 0.705933 -0.304016 0.970932 [Debug] dit_step24_xt: [2170, 64] first4: 0.372502 1.295388 -0.103176 0.246558 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.373276 0.597473 -0.301147 0.951920 [Debug] dit_step25_xt: [2170, 64] first4: 0.379968 1.283438 -0.097153 0.227519 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.400734 0.478027 -0.291565 0.925476 [Debug] dit_step26_xt: [2170, 64] first4: 0.387982 1.273878 -0.091322 0.209010 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.424156 0.362274 -0.289185 0.905457 [Debug] dit_step27_xt: [2170, 64] first4: 0.396466 1.266632 -0.085538 0.190901 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.446457 0.248230 -0.283020 0.890198 [Debug] dit_step28_xt: [2170, 64] first4: 0.405395 1.261667 -0.079878 0.173097 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.469948 0.127096 -0.266113 0.861603 [Debug] dit_step29_xt: [2170, 64] first4: 0.414794 1.259126 -0.074556 0.155865 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.484722 0.013313 -0.257355 0.847748 [Debug] dit_step30_xt: [2170, 64] first4: 0.424488 1.258859 -0.069409 0.138910 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.504712 -0.100014 -0.247437 0.829651 [Debug] dit_step31_xt: [2170, 64] first4: 0.434582 1.260860 -0.064460 0.122317 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.521059 -0.216339 -0.233765 0.802582 [Debug] dit_step32_xt: [2170, 64] first4: 0.445004 1.265186 -0.059785 0.106265 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.529373 -0.326294 -0.215881 0.785248 [Debug] dit_step33_xt: [2170, 64] first4: 0.455591 1.271712 -0.055467 0.090560 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.541252 -0.421661 -0.194504 0.760132 [Debug] dit_step34_xt: [2170, 64] first4: 0.466416 1.280146 -0.051577 0.075357 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.552841 -0.529602 -0.176086 0.735016 [Debug] dit_step35_xt: [2170, 64] first4: 0.477473 1.290738 -0.048055 0.060657 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.559322 -0.628727 -0.147636 0.698792 [Debug] dit_step36_xt: [2170, 64] first4: 0.488659 1.303312 -0.045103 0.046681 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.556705 -0.713518 -0.131134 0.672607 [Debug] dit_step37_xt: [2170, 64] first4: 0.499793 1.317582 -0.042480 0.033229 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.560322 -0.795479 -0.101242 0.639648 [Debug] dit_step38_xt: [2170, 64] first4: 0.511000 1.333492 -0.040455 0.020436 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.552338 -0.876106 -0.062042 0.600586 [Debug] dit_step39_xt: [2170, 64] first4: 0.522047 1.351014 -0.039214 0.008424 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.541878 -0.944748 -0.033936 0.549805 [Debug] dit_step40_xt: [2170, 64] first4: 0.532884 1.369909 -0.038535 -0.002572 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.533014 -1.015640 0.002983 0.507080 [Debug] dit_step41_xt: [2170, 64] first4: 0.543544 1.390222 -0.038595 -0.012713 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.519150 -1.075264 0.050107 0.456848 [Debug] dit_step42_xt: [2170, 64] first4: 0.553927 1.411727 -0.039597 -0.021850 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.504318 -1.131897 0.087883 0.404297 [Debug] dit_step43_xt: [2170, 64] first4: 0.564014 1.434365 -0.041355 -0.029936 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.490322 -1.176483 0.124756 0.349365 [Debug] dit_step44_xt: [2170, 64] first4: 0.573820 1.457895 -0.043850 -0.036924 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.467514 -1.219910 0.159439 0.295258 [Debug] dit_step45_xt: [2170, 64] first4: 0.583171 1.482293 -0.047039 -0.042829 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.446793 -1.257721 0.195705 0.237518 [Debug] dit_step46_xt: [2170, 64] first4: 0.592106 1.507447 -0.050953 -0.047579 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.414459 -1.286896 0.206978 0.195801 [Debug] dit_step47_xt: [2170, 64] first4: 0.600396 1.533185 -0.055092 -0.051495 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.371643 -1.311859 0.225754 0.152420 [Debug] dit_step48_xt: [2170, 64] first4: 0.607828 1.559423 -0.059608 -0.054543 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.406494 -1.337708 0.207794 0.146149 [Debug] dit_x0: [2170, 64] first4: 0.615958 1.586177 -0.063763 -0.057466 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 4569.6 ms (4569.6 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.615958 1.586177 -0.063763 -0.057466 [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: Vulkan0 (CPU threads: 16) [VAE] Backend: Vulkan0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 694 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: 782.8 ms [Debug] vae_audio: [2, 4166400] first4: -0.001514 -0.000959 -0.001229 -0.000823 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 416 ms (208.43x realtime), 32 threads [MP3] Wrote ggml-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 10:15:19.524 | INFO | acestep.core.generation.handler.init_service_loader:_load_main_model_from_checkpoint:174 - [initialize_service] Attempting to load model with attention implementation: sdpa 2026-07-12 10:15:20.992 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:15:20.992 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:15:20.992 | INFO | acestep.core.generation.handler.generate_music_request:_prepare_reference_and_source_audio:179 - [generate_music] cover task: no src_audio but audio codes provided, proceeding with codes 2026-07-12 10:15:20.995 | INFO | acestep.core.generation.handler.generate_music:_vram_preflight_check:156 - [generate_music] VRAM pre-flight: 81.69 GB free, ~0.94 GB needed (batch=1, duration=88s, mode=turbo). 2026-07-12 10:15:21.000 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:15:21.005 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_precomputed_lm_hints:31 - [generate_music] Decoding audio codes for LM hints for item 0... 2026-07-12 10:15:21.006 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:15:21.006 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:15:21.006 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:15:21.006 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:122 - text_prompt: # Instruction Generate audio semantic tokens based on the given conditions: # Caption An upbeat and anthemic pop-rock track driven by bright, slightly overdriven # Metas - bpm: 83 - timesignature: 4 - keyscale: G major - duration: 88 seconds <|endoftext|> 2026-07-12 10:15:21.006 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:15:21.006 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:124 - lyrics_text: # Languages fr # Lyric # Lyric [Intro - Guitar Riff] [Verse 1] Dans le monde des tutos virtuels G ta toise en nouvelle passion Avec Ggendoline et PumbΓ© Γ  midi La communautΓ©, c'est l'unitΓ© Quel joie, une clΓ© [Chorus] Dans le monde des tutos virtuels GΓ’ndoline et PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner [Guitar Solo] [Verse 2] Dans le monde des tutos virtuels GΓ’ndoline, PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner<|endoftext|> 2026-07-12 10:15:21.006 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:15:21.010 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:15:21.018 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:15:21.018 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:15:21.024 | 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 0.999812 lyric_embed 1.000000 enc_hidden 0.999841 detok_output 0.999997 context 0.999998 noise 1.000000 temb_t 0.999999 hidden_after_proj_in 0.999987 enc_after_cond_emb 0.999830 layer0_sa_output 0.999960 hidden_after_layer0 0.999981 hidden_after_layer6 0.999917 hidden_after_layer12 0.999325 hidden_after_layer18 0.996723 hidden_after_layer23 0.993837 dit_step0_vt 0.975958 dit_step0_xt 0.999947 dit_step1_vt 0.979995 dit_step1_xt 0.999836 dit_step2_vt 0.982104 dit_step2_xt 0.999562 dit_step3_vt 0.982614 dit_step3_xt 0.998939 dit_step4_vt 0.981015 dit_step4_xt 0.997529 dit_step5_vt 0.978424 dit_step5_xt 0.994376 dit_step6_vt 0.975361 dit_step6_xt 0.988376 dit_step7_vt 0.969708 dit_x0 0.979522 vae_audio 0.903905 vae_audio (STFT cosine) 0.976143 [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 0.138063 0.006583 -0.002309 0.972929 -0.002342 0.972003 dit_step1_xt 0.999836 0.266496 0.011191 -0.005307 0.942656 -0.005313 0.941730 dit_step2_xt 0.999562 0.452788 0.017346 -0.009345 0.909152 -0.009311 0.908527 dit_step3_xt 0.998939 0.645615 0.025795 -0.014709 0.873772 -0.014577 0.873624 dit_step4_xt 0.997529 0.781190 0.037650 -0.021757 0.841928 -0.021660 0.841995 dit_step5_xt 0.994376 1.217481 0.055295 -0.031831 0.825212 -0.032109 0.824593 dit_step6_xt 0.988376 2.072577 0.082011 -0.046132 0.856064 -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: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/Qwen3-Embedding-0.6B-BF16.gguf: 310 tensors, data at offset 5337568 [Load] TextEncoder: 28L, H=1024, Nh=16/8 [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 310 tensors, 1136.5 MB into backend [Store] Load TextEnc: 495 ms [Debug] text_hidden: [70, 1024] first4: 3.705836 2.395382 0.221845 -13.145830 [Debug] lyric_embed: [167, 1024] first4: 0.029175 0.032227 -0.022339 -0.028809 [Store] Unload TextEnc (1136.5 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] CondEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-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: 681 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, 812.0 ms [Debug] enc_hidden: [238, 2048] first4: 0.686875 -0.019087 -0.051410 0.022588 [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: Vulkan0 (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: 100 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 175.4 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.125156 1.434936 0.308101 -0.624459 [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: -0.125156 1.434936 0.308101 -0.624459 [Init-Noise] Starting: T=2170, S=1085, enc_S=238, steps=8, batch=1 (cover) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] DiT backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-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: 4937 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: 0.016458 0.108603 0.065523 -0.074042 [Debug] temb: [2560] first4: 0.026321 -0.023010 -0.016508 0.047175 [Debug] temb_t: [2560] first4: 0.014560 -0.022321 -0.020859 0.048784 [Debug] temb_r: [2560] first4: 0.011760 -0.000690 0.004351 -0.001609 [Debug] sinusoidal_t: [256] first4: 0.562486 0.789701 0.439822 -0.023583 [Debug] sinusoidal_r: [256] first4: 1.000000 1.000000 1.000000 1.000000 [Debug] temb_lin1_t: [2560] first4: 0.064010 -0.019467 0.138295 -0.113216 [Debug] temb_lin1_r: [2560] first4: -0.009879 0.015926 -0.017671 0.012016 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.338257 -0.576416 0.228638 -0.191162 [Debug] proj_in_input: [192, 2170] first4: -0.125156 1.434936 0.308101 -0.624459 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.030416 0.652095 0.102835 0.871851 [Debug] layer0_sa_input: [2560, 1085] first4: 0.040103 -0.045358 0.093961 -0.134828 [Debug] layer0_q_after_rope: [128, 32] first4: -0.227198 -2.176497 0.196191 -0.060062 [Debug] layer0_k_after_rope: [128, 8] first4: 0.576306 2.180901 0.746600 0.176882 [Debug] layer0_sa_output: [2560, 1085] first4: 1.863272 4.426610 -0.946739 5.656893 [Debug] layer0_attn_out: [4096, 1085] first4: 0.486700 -2.292673 -0.463547 -2.908695 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.486700 -2.292673 -0.463547 -2.908695 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.353539 -2.045633 -0.288238 -3.871331 [Debug] hidden_after_layer0: [2560, 1085] first4: 1.702963 -2.482954 240.973358 -3.841229 [Debug] hidden_after_layer6: [2560, 1085] first4: -6.179946 -2.164359 320.246765 -12.277622 [Debug] hidden_after_layer12: [2560, 1085] first4: -10.614133 29.579311 261.171844 29.408081 [Debug] hidden_after_layer18: [2560, 1085] first4: 41.203049 13.446496 278.918793 4.406458 [Debug] hidden_after_layer31: [2560, 1085] first4: -83.402046 -233.666306 -161.509064 253.163101 [Debug] dit_step0_vt: [2170, 64] first4: -1.162354 1.468521 0.334213 1.148193 [Debug] dit_step0_xt: [2170, 64] first4: 0.247170 2.089499 -0.187067 0.795466 [DiT] Step 1/8 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.809082 1.296021 0.507019 0.953613 [Debug] dit_step1_xt: [2170, 64] first4: 0.291302 2.018807 -0.214722 0.743450 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: -1.383423 1.854743 1.409149 1.351562 [Debug] dit_step2_xt: [2170, 64] first4: 0.383530 1.895157 -0.308665 0.653346 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: -0.746185 1.601089 1.242027 0.895935 [Debug] dit_step3_xt: [2170, 64] first4: 0.445712 1.761733 -0.412168 0.578685 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: 0.050125 2.847046 1.007507 0.568726 [Debug] dit_step4_xt: [2170, 64] first4: 0.440342 1.456693 -0.520115 0.517750 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: 0.529541 2.280090 1.050507 0.334229 [Debug] dit_step5_xt: [2170, 64] first4: 0.364693 1.130965 -0.670187 0.470003 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: 0.971481 2.036499 0.379242 0.571777 [Debug] dit_step6_xt: [2170, 64] first4: 0.170397 0.723666 -0.746036 0.355648 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: -1.157349 2.776550 1.377594 0.749634 [Debug] dit_x0: [2170, 64] first4: 0.517601 -0.109299 -1.159314 0.130757 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 1893.0 ms (1893.0 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.517601 -0.109299 -1.159314 0.130757 [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: Vulkan0 (CPU threads: 16) [VAE] Backend: Vulkan0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 679 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: 786.7 ms [Debug] vae_audio: [2, 4166400] first4: 0.005116 0.007718 0.007233 0.006485 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 462 ms (187.96x realtime), 32 threads [MP3] Wrote ggml-xl-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 10:15:35.623 | 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 0.999812 lyric_embed 1.000000 enc_hidden 0.999841 detok_output 0.999997 context 0.999998 noise 1.000000 temb_t 0.999998 hidden_after_proj_in 0.999987 enc_after_cond_emb 0.999831 layer0_sa_output 0.999950 hidden_after_layer0 0.999981 hidden_after_layer6 0.999856 hidden_after_layer12 0.999479 hidden_after_layer18 0.998728 hidden_after_layer23 0.998994 dit_step0_vt 0.998955 dit_step0_xt 1.000000 dit_step5_vt 0.999522 dit_step5_xt 0.999981 dit_step10_vt 0.999173 dit_step10_xt 0.999938 dit_step15_vt 0.998250 dit_step15_xt 0.999855 dit_step20_vt 0.996599 dit_step20_xt 0.999639 dit_step25_vt 0.993582 dit_step25_xt 0.999198 dit_step30_vt 0.989631 dit_step30_xt 0.998384 dit_step35_vt 0.985256 dit_step35_xt 0.997238 dit_step40_vt 0.982582 dit_step40_xt 0.995994 dit_step45_vt 0.985747 dit_step45_xt 0.995066 dit_step49_vt 0.990662 dit_x0 0.994738 vae_audio 0.970218 vae_audio (STFT cosine) 0.988082 [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 0.020009 0.001450 -0.001655 0.978450 -0.001619 0.978872 dit_step5_xt 0.999981 0.079072 0.004214 -0.008378 0.884731 -0.008480 0.883811 dit_step10_xt 0.999938 0.120608 0.006690 -0.015100 0.800851 -0.015075 0.802324 dit_step15_xt 0.999855 0.167795 0.009393 -0.021794 0.730401 -0.021933 0.731304 dit_step20_xt 0.999639 0.230612 0.013706 -0.028525 0.677642 -0.028882 0.679066 dit_step25_xt 0.999198 0.330766 0.019402 -0.035328 0.647031 -0.035643 0.650780 dit_step30_xt 0.998384 0.566098 0.026959 -0.042167 0.641832 -0.042577 0.647107 dit_step35_xt 0.997238 0.825428 0.036034 -0.049013 0.662746 -0.049471 0.669256 dit_step40_xt 0.995994 1.083351 0.046195 -0.055889 0.707392 -0.056532 0.715783 dit_step45_xt 0.995066 1.312607 0.055793 -0.062869 0.771423 -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 10:15:38.771 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:15:38.771 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:15:38.772 | INFO | acestep.core.generation.handler.generate_music_request:_prepare_reference_and_source_audio:179 - [generate_music] cover task: no src_audio but audio codes provided, proceeding with codes 2026-07-12 10:15:38.775 | INFO | acestep.core.generation.handler.generate_music:_vram_preflight_check:156 - [generate_music] VRAM pre-flight: 76.75 GB free, ~1.23 GB needed (batch=1, duration=88s, mode=xl_turbo). 2026-07-12 10:15:38.780 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:15:38.785 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_precomputed_lm_hints:31 - [generate_music] Decoding audio codes for LM hints for item 0... 2026-07-12 10:15:38.787 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:15:38.787 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:15:38.787 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:15:38.787 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:122 - text_prompt: # Instruction Generate audio semantic tokens based on the given conditions: # Caption An upbeat and anthemic pop-rock track driven by bright, slightly overdriven # Metas - bpm: 83 - timesignature: 4 - keyscale: G major - duration: 88 seconds <|endoftext|> 2026-07-12 10:15:38.787 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:15:38.787 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:124 - lyrics_text: # Languages fr # Lyric # Lyric [Intro - Guitar Riff] [Verse 1] Dans le monde des tutos virtuels G ta toise en nouvelle passion Avec Ggendoline et PumbΓ© Γ  midi La communautΓ©, c'est l'unitΓ© Quel joie, une clΓ© [Chorus] Dans le monde des tutos virtuels GΓ’ndoline et PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner [Guitar Solo] [Verse 2] Dans le monde des tutos virtuels GΓ’ndoline, PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner<|endoftext|> 2026-07-12 10:15:38.787 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:15:38.791 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:15:38.798 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:15:38.798 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:15:38.805 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 10:15:38.812 | INFO | acestep.models.common.dcw_correction:__init__:98 - [DCW] Active β€” mode=double, scaler=0.0500, high_scaler=0.0200, wavelet='haar' 2026-07-12 10:15:39.367 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 10:15:39.367 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:64 - [generate_music] pred_latents: torch.Size([1, 2170, 64]), dtype=torch.bfloat16 2026-07-12 10:15:39.367 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.007027626037597656, 'diffusion_time_cost': 0.5544929504394531, 'diffusion_per_step_time_cost': 0.06931161880493164, 'total_time_cost': 0.5615205764770508, 'offload_time_cost': 0.0} 2026-07-12 10:15:39.391 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 10:15:39.421 | DEBUG | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:136 - [generate_music] Before VAE decode: allocated=17.03GB, max=18.34GB 2026-07-12 10:15:39.421 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:154 - [generate_music] Effective free VRAM before VAE decode: 76.66 GB 2026-07-12 10:15:39.421 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:172 - [generate_music] Using tiled VAE decode to reduce VRAM usage... 2026-07-12 10:15:39.421 | DEBUG | acestep.core.generation.handler.memory_utils:_get_auto_decode_chunk_size:108 - [_get_auto_decode_chunk_size] Effective free VRAM: 76.66 GB 2026-07-12 10:15:39.421 | DEBUG | acestep.core.generation.handler.memory_utils:_should_offload_wav_to_cpu:131 - [_should_offload_wav_to_cpu] Effective free VRAM: 76.66 GB 2026-07-12 10:15:39.421 | INFO | acestep.core.generation.handler.vae_decode:tiled_decode:56 - [tiled_decode] chunk_size=512, offload_wav_to_cpu=False, latents_shape=torch.Size([1, 64, 2170]) 2026-07-12 10:15:39.696 | DEBUG | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:193 - [generate_music] After VAE decode: allocated=17.21GB, max=18.50GB 2026-07-12 10:15:39.899 | INFO | acestep.core.generation.handler.generate_music_payload:_build_generate_music_success_payload:38 - [generate_music] VAE decode completed. Preparing audio tensors... 2026-07-12 10:15:39.903 | INFO | acestep.core.generation.handler.generate_music_payload:_build_generate_music_success_payload:50 - [generate_music] Done! Generated 1 audio tensors. [Request] Parsed ggml-xl-sft/request0.json [Request] seed=42 lm_batch=1 synth_batch=1 [Request] caption: An upbeat and anthemic pop-rock track driven by bright, slig... [Request] lyrics: 490 bytes [Request] bpm=83 dur=88 key=G major ts=4 lang=fr [Request] lm: temp=0.85 cfg=2.0 top_p=0.90 top_k=0 [Request] dit: steps=50 guidance=1.0 shift=1.0 [Request] task_type: text2music [Request] solver: euler (stork_substeps=10) [Request] lm_mode: generate [Request] output_format: mp3 [Request] synth_model: acestep-v15-xl-sft-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: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/Qwen3-Embedding-0.6B-BF16.gguf: 310 tensors, data at offset 5337568 [Load] TextEncoder: 28L, H=1024, Nh=16/8 [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 310 tensors, 1136.5 MB into backend [Store] Load TextEnc: 322 ms [Debug] text_hidden: [70, 1024] first4: 3.705836 2.395382 0.221845 -13.145830 [Debug] lyric_embed: [167, 1024] first4: 0.029175 0.032227 -0.022339 -0.028809 [Store] Unload TextEnc (1136.5 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] CondEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-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: 712 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, 18.4 ms [Debug] enc_hidden: [238, 2048] first4: 0.686875 -0.019087 -0.051410 0.022588 [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: Vulkan0 (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: 98 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 174.8 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.125156 1.434936 0.308101 -0.624459 [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: -0.125156 1.434936 0.308101 -0.624459 [Init-Noise] Starting: T=2170, S=1085, enc_S=238, steps=50, batch=1 (cover) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] DiT backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-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: 3911 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: 0.012546 0.105322 0.073633 -0.031913 [Debug] temb: [2560] first4: 0.009488 -0.007695 -0.006104 0.004570 [Debug] temb_t: [2560] first4: 0.007949 -0.007936 -0.006669 0.003595 [Debug] temb_r: [2560] first4: 0.001539 0.000241 0.000566 0.000975 [Debug] sinusoidal_t: [256] first4: 0.562486 0.789701 0.439822 -0.023583 [Debug] sinusoidal_r: [256] first4: 1.000000 1.000000 1.000000 1.000000 [Debug] temb_lin1_t: [2560] first4: 0.032232 -0.081851 0.088236 -0.064962 [Debug] temb_lin1_r: [2560] first4: 0.006956 -0.005390 -0.002806 -0.011722 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.327515 -0.491455 0.102783 -0.215088 [Debug] proj_in_input: [192, 2170] first4: -0.125156 1.434936 0.308101 -0.624459 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.035021 0.640747 0.108140 0.871140 [Debug] layer0_sa_input: [2560, 1085] first4: 0.036374 -0.018622 0.098722 -0.097802 [Debug] layer0_q_after_rope: [128, 32] first4: -0.253384 -2.265408 0.031060 -0.232315 [Debug] layer0_k_after_rope: [128, 8] first4: 0.647914 2.979680 0.672841 1.817435 [Debug] layer0_sa_output: [2560, 1085] first4: 2.196742 4.505742 -1.249434 8.231668 [Debug] layer0_attn_out: [4096, 1085] first4: 0.471710 -2.219070 -0.825796 -4.313847 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.471710 -2.219070 -0.825796 -4.313847 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.310233 -1.951335 -0.738785 -4.389479 [Debug] hidden_after_layer0: [2560, 1085] first4: -1.703593 -1.277146 250.086655 -4.100759 [Debug] hidden_after_layer6: [2560, 1085] first4: -7.900845 3.729650 316.167816 -21.358837 [Debug] hidden_after_layer12: [2560, 1085] first4: -23.703049 40.317162 229.394943 17.508188 [Debug] hidden_after_layer18: [2560, 1085] first4: 9.075687 -12.724508 233.976532 19.595821 [Debug] hidden_after_layer31: [2560, 1085] first4: 62.375664 -104.938370 8.656990 90.371292 [Debug] dit_step0_vt: [2170, 64] first4: -0.316406 2.216187 0.003906 1.698120 [Debug] dit_step0_xt: [2170, 64] first4: 0.200664 2.111926 -0.171953 0.813694 [DiT] Step 1/50 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.348145 2.149902 0.082764 1.557556 [Debug] dit_step1_xt: [2170, 64] first4: 0.207627 2.068928 -0.173608 0.782543 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.325317 2.165649 0.059082 1.568512 [Debug] dit_step2_xt: [2170, 64] first4: 0.214133 2.025615 -0.174790 0.751172 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.274780 2.131714 0.048218 1.568039 [Debug] dit_step3_xt: [2170, 64] first4: 0.219629 1.982981 -0.175754 0.719812 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.267090 2.065308 0.020020 1.585907 [Debug] dit_step4_xt: [2170, 64] first4: 0.224971 1.941675 -0.176155 0.688094 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.259766 2.043365 0.032959 1.587677 [Debug] dit_step5_xt: [2170, 64] first4: 0.230166 1.900807 -0.176814 0.656340 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.250000 2.009216 0.067505 1.608887 [Debug] dit_step6_xt: [2170, 64] first4: 0.235166 1.860623 -0.178164 0.624162 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.244141 1.973572 0.112305 1.616028 [Debug] dit_step7_xt: [2170, 64] first4: 0.240049 1.821152 -0.180410 0.591842 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.255615 1.928894 0.141113 1.614807 [Debug] dit_step8_xt: [2170, 64] first4: 0.245161 1.782574 -0.183232 0.559546 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.278076 1.877136 0.125244 1.612930 [Debug] dit_step9_xt: [2170, 64] first4: 0.250723 1.745031 -0.185737 0.527287 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.277954 1.838287 0.119141 1.609753 [Debug] dit_step10_xt: [2170, 64] first4: 0.256282 1.708265 -0.188120 0.495092 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.278198 1.791672 0.110107 1.598598 [Debug] dit_step11_xt: [2170, 64] first4: 0.261846 1.672432 -0.190322 0.463120 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.266479 1.747620 0.101807 1.587952 [Debug] dit_step12_xt: [2170, 64] first4: 0.267175 1.637480 -0.192358 0.431361 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.264893 1.698116 0.091919 1.578461 [Debug] dit_step13_xt: [2170, 64] first4: 0.272473 1.603517 -0.194197 0.399792 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.256958 1.644302 0.072388 1.571503 [Debug] dit_step14_xt: [2170, 64] first4: 0.277612 1.570631 -0.195645 0.368362 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.249390 1.598297 0.056152 1.566498 [Debug] dit_step15_xt: [2170, 64] first4: 0.282600 1.538665 -0.196768 0.337032 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.234741 1.554199 0.032837 1.560608 [Debug] dit_step16_xt: [2170, 64] first4: 0.287295 1.507581 -0.197424 0.305820 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.212036 1.501465 0.016357 1.555359 [Debug] dit_step17_xt: [2170, 64] first4: 0.291536 1.477552 -0.197751 0.274712 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.181580 1.450378 0.004517 1.549530 [Debug] dit_step18_xt: [2170, 64] first4: 0.295167 1.448544 -0.197842 0.243722 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.152039 1.401733 -0.010132 1.535309 [Debug] dit_step19_xt: [2170, 64] first4: 0.298208 1.420510 -0.197639 0.213016 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.120422 1.356445 -0.017090 1.528625 [Debug] dit_step20_xt: [2170, 64] first4: 0.300616 1.393381 -0.197297 0.182443 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.096130 1.300293 -0.026001 1.523956 [Debug] dit_step21_xt: [2170, 64] first4: 0.302539 1.367375 -0.196777 0.151964 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.068542 1.255005 -0.042969 1.512207 [Debug] dit_step22_xt: [2170, 64] first4: 0.303910 1.342275 -0.195918 0.121720 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.053040 1.201843 -0.057007 1.496979 [Debug] dit_step23_xt: [2170, 64] first4: 0.304971 1.318238 -0.194778 0.091780 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.041870 1.148315 -0.056519 1.479950 [Debug] dit_step24_xt: [2170, 64] first4: 0.305808 1.295272 -0.193647 0.062181 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.033752 1.097534 -0.066895 1.468170 [Debug] dit_step25_xt: [2170, 64] first4: 0.306483 1.273321 -0.192310 0.032818 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.030884 1.036011 -0.072388 1.441101 [Debug] dit_step26_xt: [2170, 64] first4: 0.307101 1.252601 -0.190862 0.003996 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.036133 0.975769 -0.071777 1.424591 [Debug] dit_step27_xt: [2170, 64] first4: 0.307823 1.233085 -0.189426 -0.024496 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.047119 0.918304 -0.087708 1.397980 [Debug] dit_step28_xt: [2170, 64] first4: 0.308766 1.214719 -0.187672 -0.052456 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.054688 0.848633 -0.099487 1.377563 [Debug] dit_step29_xt: [2170, 64] first4: 0.309860 1.197747 -0.185682 -0.080007 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.068054 0.774475 -0.111023 1.358246 [Debug] dit_step30_xt: [2170, 64] first4: 0.311221 1.182257 -0.183462 -0.107172 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.085266 0.700592 -0.118347 1.337036 [Debug] dit_step31_xt: [2170, 64] first4: 0.312926 1.168245 -0.181095 -0.133912 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.101135 0.611082 -0.127380 1.309875 [Debug] dit_step32_xt: [2170, 64] first4: 0.314949 1.156024 -0.178547 -0.160110 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.114929 0.532196 -0.135132 1.289673 [Debug] dit_step33_xt: [2170, 64] first4: 0.317247 1.145380 -0.175845 -0.185903 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.127502 0.447815 -0.141602 1.257019 [Debug] dit_step34_xt: [2170, 64] first4: 0.319797 1.136423 -0.173013 -0.211044 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.144318 0.342323 -0.147644 1.233154 [Debug] dit_step35_xt: [2170, 64] first4: 0.322684 1.129577 -0.170060 -0.235707 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.155212 0.248901 -0.134247 1.202942 [Debug] dit_step36_xt: [2170, 64] first4: 0.325788 1.124599 -0.167375 -0.259766 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.153740 0.159088 -0.121895 1.161926 [Debug] dit_step37_xt: [2170, 64] first4: 0.328863 1.121417 -0.164937 -0.283004 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.166504 0.051086 -0.119675 1.121887 [Debug] dit_step38_xt: [2170, 64] first4: 0.332193 1.120395 -0.162543 -0.305442 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.194481 -0.054657 -0.094055 1.093933 [Debug] dit_step39_xt: [2170, 64] first4: 0.336082 1.121488 -0.160662 -0.327321 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.212463 -0.163483 -0.078231 1.048035 [Debug] dit_step40_xt: [2170, 64] first4: 0.340332 1.124758 -0.159098 -0.348281 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.236084 -0.271400 -0.059311 1.004700 [Debug] dit_step41_xt: [2170, 64] first4: 0.345053 1.130186 -0.157912 -0.368375 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.258850 -0.356674 -0.024078 0.963196 [Debug] dit_step42_xt: [2170, 64] first4: 0.350230 1.137319 -0.157430 -0.387639 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.290710 -0.459518 -0.006454 0.908203 [Debug] dit_step43_xt: [2170, 64] first4: 0.356045 1.146510 -0.157301 -0.405803 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.319916 -0.554718 0.051743 0.850403 [Debug] dit_step44_xt: [2170, 64] first4: 0.362443 1.157604 -0.158336 -0.422811 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.354156 -0.631592 0.087448 0.797363 [Debug] dit_step45_xt: [2170, 64] first4: 0.369526 1.170236 -0.160085 -0.438759 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.394440 -0.700943 0.135712 0.742859 [Debug] dit_step46_xt: [2170, 64] first4: 0.377415 1.184255 -0.162799 -0.453616 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.444146 -0.785172 0.203552 0.699951 [Debug] dit_step47_xt: [2170, 64] first4: 0.386298 1.199958 -0.166870 -0.467615 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.498672 -0.845062 0.265076 0.684753 [Debug] dit_step48_xt: [2170, 64] first4: 0.396271 1.216860 -0.172171 -0.481310 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.500023 -0.831848 0.188049 0.699951 [Debug] dit_x0: [2170, 64] first4: 0.406272 1.233497 -0.175932 -0.495309 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 11649.4 ms (11649.4 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.406272 1.233497 -0.175932 -0.495309 [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: Vulkan0 (CPU threads: 16) [VAE] Backend: Vulkan0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 662 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: 761.9 ms [Debug] vae_audio: [2, 4166400] first4: -0.001384 -0.000642 -0.000765 -0.000267 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 412 ms (210.90x realtime), 32 threads [MP3] Wrote ggml-xl-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 10:15:59.875 | 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 0.999812 lyric_embed 1.000000 enc_hidden 0.999572 detok_output 0.999997 context 0.999998 noise 1.000000 temb_t 0.999996 hidden_after_proj_in 0.999980 enc_after_cond_emb 0.999689 layer0_sa_output 0.989369 hidden_after_layer0 0.999252 hidden_after_layer6 0.999717 hidden_after_layer12 0.997136 hidden_after_layer18 0.996172 hidden_after_layer31 0.987842 dit_step0_vt 0.908717 dit_step0_xt 0.999836 dit_step1_vt 0.922692 dit_step1_xt 0.999498 dit_step2_vt 0.926940 dit_step2_xt 0.998700 dit_step3_vt 0.924410 dit_step3_xt 0.996857 dit_step4_vt 0.919379 dit_step4_xt 0.992302 dit_step5_vt 0.908835 dit_step5_xt 0.981241 dit_step6_vt 0.899287 dit_step6_xt 0.959190 dit_step7_vt 0.879336 dit_x0 0.923063 vae_audio 0.728125 vae_audio (STFT cosine) 0.920606 [Xl-turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999836 0.153562 0.012723 -0.002988 0.978400 -0.002721 0.976534 dit_step1_xt 0.999498 0.317477 0.022097 -0.005862 0.952730 -0.005567 0.949004 dit_step2_xt 0.998700 0.547592 0.034056 -0.009543 0.924196 -0.009330 0.918931 dit_step3_xt 0.996857 0.810684 0.050353 -0.015077 0.893459 -0.014418 0.887241 dit_step4_xt 0.992302 1.201432 0.075256 -0.022783 0.864213 -0.021453 0.857796 dit_step5_xt 0.981241 2.447535 0.114975 -0.033996 0.847735 -0.032240 0.840033 dit_step6_xt 0.959190 4.133563 0.175471 -0.050179 0.870877 -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 10:16:04.239 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:16:04.239 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:124 - lyrics_text: # Languages fr # Lyric # Lyric [Intro - Guitar Riff] [Verse 1] Dans le monde des tutos virtuels G ta toise en nouvelle passion Avec Ggendoline et PumbΓ© Γ  midi La communautΓ©, c'est l'unitΓ© Quel joie, une clΓ© [Chorus] Dans le monde des tutos virtuels GΓ’ndoline et PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner [Guitar Solo] [Verse 2] Dans le monde des tutos virtuels GΓ’ndoline, PumbΓ© Γ  midi Une famille Γ  connecter, c'est vrai D'un enfant qui voit toi fusionner<|endoftext|> 2026-07-12 10:16:04.239 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:16:04.242 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:16:04.250 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:16:04.250 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:16:04.256 | 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