[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-Q6_K.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-Q6_K.gguf: 678 tensors, data at offset 56864 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q6_K.gguf: 678 tensors, data at offset 56864 [Synth-Load] Ready: turbo=yes, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=8, guidance=1.0, shift=3.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/Qwen3-Embedding-0.6B-BF16.gguf: 310 tensors, data at offset 5337568 [Load] TextEncoder: 28L, H=1024, Nh=16/8 [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 310 tensors, 1136.5 MB into backend [Store] Load TextEnc: 328 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-Q6_K.gguf: 678 tensors, data at offset 56864 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 140 tensors, 476.3 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L), text_proj, null_cond [Store] Load CondEnc: 303 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, 3922.3 ms [Debug] enc_hidden: [238, 2048] first4: 1.761343 -0.050573 -0.132979 0.058387 [Store] Unload CondEnc (476.3 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-Q6_K.gguf: 678 tensors, data at offset 56864 [WeightCtx] Loaded 30 tensors, 82.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 54 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 401.4 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.141068 1.454457 0.315091 -0.623697 [Store] Unload FSQ-Detok (82.2 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: -0.141068 1.454457 0.315091 -0.623697 [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-Q6_K.gguf: 678 tensors, data at offset 56864 [DiT] Self-attn: Q+K+V fused [DiT] Cross-attn: Q+K+V fused [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 478 tensors, 1237.2 MB into backend [Load] DiT: 24 layers, H=2048, Nh=16/8, D=128 [Store] Load DiT: 919 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: 0.261089 -0.161223 -0.098727 0.051901 [Debug] temb: [2048] first4: 0.000236 -0.132397 -0.035347 0.064653 [Debug] temb_t: [2048] first4: 0.001398 0.026957 -0.052741 0.063660 [Debug] temb_r: [2048] first4: -0.001162 -0.159353 0.017394 0.000993 [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: -0.049071 -0.051112 -0.017769 -0.037193 [Debug] temb_lin1_r: [2048] first4: -0.014408 -0.020609 -0.015729 0.003875 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.037598 -0.957581 0.541260 0.451874 [Debug] proj_in_input: [192, 2170] first4: -0.141068 1.454457 0.315091 -0.623697 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.170334 0.815842 0.309875 -0.571411 [Debug] layer0_sa_input: [2048, 1085] first4: -0.716077 -0.756539 -0.048409 0.263530 [Debug] layer0_q_after_rope: [128, 16] first4: -1.541147 -1.051668 0.182985 0.461973 [Debug] layer0_k_after_rope: [128, 8] first4: -0.170334 0.815842 0.309875 -0.571411 [Debug] layer0_sa_output: [2048, 1085] first4: -1.505859 0.211914 -0.366699 0.520020 [Debug] layer0_attn_out: [2048, 1085] first4: -1.541147 -1.051668 0.182985 0.461973 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.541147 -1.051668 0.182985 0.461973 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.586069 -0.808137 -0.317991 0.502622 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.128009 0.526317 51.478546 -0.854151 [Debug] hidden_after_layer6: [2048, 1085] first4: -20.769688 -0.200171 34.400143 -4.366263 [Debug] hidden_after_layer12: [2048, 1085] first4: -14.763856 -16.890856 77.643944 30.925003 [Debug] hidden_after_layer18: [2048, 1085] first4: -27.960796 12.541347 65.524132 19.350559 [Debug] hidden_after_layer23: [2048, 1085] first4: -16.374619 45.989719 197.664993 135.937531 [Debug] dit_step0_vt: [2170, 64] first4: 0.102573 1.130112 0.337555 2.339844 [Debug] dit_step0_xt: [2170, 64] first4: 0.189674 2.104881 -0.187218 0.741300 [DiT] Step 1/8 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.127975 1.327759 -0.189728 1.916592 [Debug] dit_step1_xt: [2170, 64] first4: 0.196654 2.032458 -0.176870 0.636758 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: 0.050018 1.238385 0.077057 2.381897 [Debug] dit_step2_xt: [2170, 64] first4: 0.193319 1.949899 -0.182007 0.477965 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: 0.281342 1.096268 0.231539 2.627777 [Debug] dit_step3_xt: [2170, 64] first4: 0.169874 1.858543 -0.201302 0.258984 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: 0.325378 1.031242 0.083282 2.714096 [Debug] dit_step4_xt: [2170, 64] first4: 0.135012 1.748053 -0.210225 -0.031812 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: 0.314392 0.919373 -0.284790 2.722832 [Debug] dit_step5_xt: [2170, 64] first4: 0.090099 1.616714 -0.169541 -0.420788 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: 0.223419 0.637863 -0.772179 2.771240 [Debug] dit_step6_xt: [2170, 64] first4: 0.045415 1.489141 -0.015105 -0.975036 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: 0.135040 0.106384 -1.339771 3.027344 [Debug] dit_x0: [2170, 64] first4: 0.004903 1.457226 0.386827 -1.883239 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 291.3 ms (291.3 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.004903 1.457226 0.386827 -1.883239 [Store] Unload DiT (1237.2 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: Vulkan0 (CPU threads: 16) [VAE] Backend: Vulkan0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 666 ms [VAE] Tiled decode: 3 tiles (chunk=1024, overlap=64, stride=896) [VAE] Graph: 479 nodes, T_latent=960 [VAE] Upsample factor: 1920.00 (expected ~1920) [VAE] Graph: 479 nodes, T_latent=1024 [VAE] Graph: 479 nodes, T_latent=442 [VAE] Tiled decode done: 3 tiles -> T_audio=4166400 (86.80s @ 48kHz) [VAE-Decode Batch0] Decode: 784.3 ms [Debug] vae_audio: [2, 4166400] first4: 0.000664 0.001393 0.001189 0.001724 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 462 ms (188.05x realtime), 32 threads [MP3] Wrote ggml-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 10:17:19.215 | 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 10:17:20.993 | WARNING | acestep.training.trainer::40 - bitsandbytes not installed. Using standard AdamW. 2026-07-12 10:17:20.995 | 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 10:17:21.063 | 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 10:17:22.648 | 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 10:17:22.648 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:17:22.648 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:17:22.648 | 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 10:17:22.651 | 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 10:17:22.656 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:17:22.861 | 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 10:17:22.863 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:17:22.863 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:17:22.863 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:17:22.863 | 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 10:17:22.863 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:17:22.863 | 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 10:17:22.863 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:17:22.870 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:17:22.885 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:17:22.885 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:17:22.909 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 10:17:22.916 | 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 10:17:23.015 | 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 10:17:23.241 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 10:17:23.242 | 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 10:17:23.242 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.00687718391418457, 'diffusion_time_cost': 0.3253509998321533, 'diffusion_per_step_time_cost': 0.040668874979019165, 'total_time_cost': 0.3322281837463379, 'offload_time_cost': 0.0} 2026-07-12 10:17:23.256 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 10:17:23.259 | 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 10:17:23.259 | 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 10:17:23.259 | 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 10:17:23.259 | 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 10:17:23.259 | 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 10:17:23.259 | 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 10:17:23.558 | 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 10:17:23.720 | 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 10:17:23.724 | 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-Q6_K.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-Q6_K.gguf: 678 tensors, data at offset 56800 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q6_K.gguf: 678 tensors, data at offset 56800 [Synth-Load] Ready: turbo=no, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=50, guidance=1.0, shift=1.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/Qwen3-Embedding-0.6B-BF16.gguf: 310 tensors, data at offset 5337568 [Load] TextEncoder: 28L, H=1024, Nh=16/8 [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 310 tensors, 1136.5 MB into backend [Store] Load TextEnc: 334 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-Q6_K.gguf: 678 tensors, data at offset 56800 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 140 tensors, 476.3 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L), text_proj, null_cond [Store] Load CondEnc: 308 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, 8.0 ms [Debug] enc_hidden: [238, 2048] first4: 1.761343 -0.050573 -0.132979 0.058387 [Store] Unload CondEnc (476.3 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-Q6_K.gguf: 678 tensors, data at offset 56800 [WeightCtx] Loaded 30 tensors, 82.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 58 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 166.7 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.141068 1.454457 0.315091 -0.623697 [Store] Unload FSQ-Detok (82.2 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: -0.141068 1.454457 0.315091 -0.623697 [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-Q6_K.gguf: 678 tensors, data at offset 56800 [DiT] Self-attn: Q+K+V fused [DiT] Cross-attn: Q+K+V fused [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 478 tensors, 1237.2 MB into backend [Load] DiT: 24 layers, H=2048, Nh=16/8, D=128 [Store] Load DiT: 917 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: 0.153861 -0.117528 -0.090110 0.080834 [Debug] temb: [2048] first4: -0.002466 -0.176370 0.004370 -0.002069 [Debug] temb_t: [2048] first4: -0.000999 0.003474 -0.013219 -0.002130 [Debug] temb_r: [2048] first4: -0.001467 -0.179844 0.017589 0.000062 [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: -0.041140 0.030273 0.027836 -0.025459 [Debug] temb_lin1_r: [2048] first4: 0.004272 0.006720 0.000208 -0.005103 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.086914 -0.913330 0.502930 0.445343 [Debug] proj_in_input: [192, 2170] first4: -0.141068 1.454457 0.315091 -0.623697 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.196335 0.882095 0.309570 -0.577072 [Debug] layer0_sa_input: [2048, 1085] first4: -0.912818 -0.710435 -0.040195 0.295128 [Debug] layer0_q_after_rope: [128, 16] first4: -1.804207 -1.364957 -0.180621 0.386244 [Debug] layer0_k_after_rope: [128, 8] first4: -0.196335 0.882095 0.309570 -0.577072 [Debug] layer0_sa_output: [2048, 1085] first4: -1.654297 0.799316 -0.604004 0.526855 [Debug] layer0_attn_out: [2048, 1085] first4: -1.804207 -1.364957 -0.180621 0.386244 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.804207 -1.364957 -0.180621 0.386244 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.953499 -1.016080 -0.441119 0.404280 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.324492 1.273869 56.608513 -1.681727 [Debug] hidden_after_layer6: [2048, 1085] first4: -16.517490 2.367325 57.986778 -1.173687 [Debug] hidden_after_layer12: [2048, 1085] first4: -9.672653 6.375622 -24.099077 -2.695755 [Debug] hidden_after_layer18: [2048, 1085] first4: -4.557255 20.772308 -41.185806 -13.952637 [Debug] hidden_after_layer23: [2048, 1085] first4: 34.651863 61.537109 52.391788 -11.685187 [Debug] dit_step0_vt: [2170, 64] first4: -0.601501 2.507568 -0.217552 1.446964 [Debug] dit_step0_xt: [2170, 64] first4: 0.206366 2.106099 -0.167524 0.818717 [DiT] Step 1/50 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.656219 2.503906 -0.068497 1.279297 [Debug] dit_step1_xt: [2170, 64] first4: 0.219490 2.056020 -0.166154 0.793131 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.652161 2.466187 -0.025574 1.303055 [Debug] dit_step2_xt: [2170, 64] first4: 0.232534 2.006697 -0.165643 0.767070 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.634064 2.402954 0.003052 1.317352 [Debug] dit_step3_xt: [2170, 64] first4: 0.245215 1.958638 -0.165704 0.740723 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.579712 2.322998 -0.006134 1.309219 [Debug] dit_step4_xt: [2170, 64] first4: 0.256809 1.912178 -0.165581 0.714538 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.549683 2.265381 -0.041306 1.304794 [Debug] dit_step5_xt: [2170, 64] first4: 0.267803 1.866870 -0.164755 0.688443 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.528717 2.222290 -0.068329 1.293121 [Debug] dit_step6_xt: [2170, 64] first4: 0.278377 1.822424 -0.163388 0.662580 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.499725 2.164185 -0.091522 1.282471 [Debug] dit_step7_xt: [2170, 64] first4: 0.288372 1.779141 -0.161558 0.636931 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.474426 2.079407 -0.102562 1.272995 [Debug] dit_step8_xt: [2170, 64] first4: 0.297860 1.737553 -0.159507 0.611471 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.416016 2.002014 -0.115929 1.258011 [Debug] dit_step9_xt: [2170, 64] first4: 0.306180 1.697512 -0.157188 0.586311 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.351990 1.912537 -0.097443 1.258530 [Debug] dit_step10_xt: [2170, 64] first4: 0.313220 1.659261 -0.155239 0.561140 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.301697 1.841614 -0.090363 1.256653 [Debug] dit_step11_xt: [2170, 64] first4: 0.319254 1.622429 -0.153432 0.536007 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.271423 1.766937 -0.095398 1.231491 [Debug] dit_step12_xt: [2170, 64] first4: 0.324683 1.587090 -0.151524 0.511377 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.250519 1.685883 -0.106049 1.220062 [Debug] dit_step13_xt: [2170, 64] first4: 0.329693 1.553373 -0.149403 0.486976 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.241241 1.621063 -0.130524 1.201080 [Debug] dit_step14_xt: [2170, 64] first4: 0.334518 1.520952 -0.146792 0.462954 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.246765 1.530609 -0.175232 1.176941 [Debug] dit_step15_xt: [2170, 64] first4: 0.339453 1.490339 -0.143288 0.439416 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.254650 1.461601 -0.200195 1.158096 [Debug] dit_step16_xt: [2170, 64] first4: 0.344546 1.461107 -0.139284 0.416254 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.263466 1.374950 -0.226440 1.143280 [Debug] dit_step17_xt: [2170, 64] first4: 0.349815 1.433608 -0.134755 0.393388 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.265228 1.298164 -0.251587 1.132477 [Debug] dit_step18_xt: [2170, 64] first4: 0.355120 1.407645 -0.129723 0.370739 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.281715 1.217697 -0.276611 1.117401 [Debug] dit_step19_xt: [2170, 64] first4: 0.360754 1.383291 -0.124191 0.348391 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.307564 1.127289 -0.302246 1.084282 [Debug] dit_step20_xt: [2170, 64] first4: 0.366906 1.360745 -0.118146 0.326705 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.327629 1.035736 -0.295288 1.061987 [Debug] dit_step21_xt: [2170, 64] first4: 0.373458 1.340031 -0.112240 0.305465 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.341904 0.929108 -0.307373 1.033506 [Debug] dit_step22_xt: [2170, 64] first4: 0.380296 1.321448 -0.106093 0.284795 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.365715 0.837402 -0.317749 1.007633 [Debug] dit_step23_xt: [2170, 64] first4: 0.387611 1.304700 -0.099738 0.264642 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.386436 0.731323 -0.325195 0.981575 [Debug] dit_step24_xt: [2170, 64] first4: 0.395339 1.290074 -0.093234 0.245011 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.403419 0.626892 -0.317566 0.964088 [Debug] dit_step25_xt: [2170, 64] first4: 0.403408 1.277536 -0.086883 0.225729 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.430286 0.494324 -0.325684 0.939499 [Debug] dit_step26_xt: [2170, 64] first4: 0.412013 1.267650 -0.080369 0.206939 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.449799 0.389923 -0.304016 0.926376 [Debug] dit_step27_xt: [2170, 64] first4: 0.421009 1.259851 -0.074289 0.188412 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.483150 0.273987 -0.299255 0.908310 [Debug] dit_step28_xt: [2170, 64] first4: 0.430672 1.254371 -0.068304 0.170245 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.502817 0.167244 -0.300598 0.895294 [Debug] dit_step29_xt: [2170, 64] first4: 0.440729 1.251027 -0.062292 0.152339 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.525919 0.045662 -0.281464 0.871063 [Debug] dit_step30_xt: [2170, 64] first4: 0.451247 1.250113 -0.056662 0.134918 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.537739 -0.058041 -0.276855 0.863037 [Debug] dit_step31_xt: [2170, 64] first4: 0.462002 1.251274 -0.051125 0.117657 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.547100 -0.163788 -0.265900 0.836700 [Debug] dit_step32_xt: [2170, 64] first4: 0.472944 1.254550 -0.045807 0.100923 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.556519 -0.276703 -0.253174 0.834442 [Debug] dit_step33_xt: [2170, 64] first4: 0.484074 1.260084 -0.040744 0.084235 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.559311 -0.372635 -0.232880 0.804565 [Debug] dit_step34_xt: [2170, 64] first4: 0.495261 1.267537 -0.036086 0.068143 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.579483 -0.476929 -0.205185 0.786591 [Debug] dit_step35_xt: [2170, 64] first4: 0.506850 1.277075 -0.031983 0.052411 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.584763 -0.571457 -0.184233 0.766754 [Debug] dit_step36_xt: [2170, 64] first4: 0.518545 1.288504 -0.028298 0.037076 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.592728 -0.656784 -0.166992 0.746490 [Debug] dit_step37_xt: [2170, 64] first4: 0.530400 1.301640 -0.024958 0.022147 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.589630 -0.742786 -0.142334 0.706787 [Debug] dit_step38_xt: [2170, 64] first4: 0.542193 1.316496 -0.022111 0.008011 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.582153 -0.827654 -0.109314 0.672333 [Debug] dit_step39_xt: [2170, 64] first4: 0.553836 1.333049 -0.019925 -0.005436 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.578615 -0.899073 -0.071594 0.636017 [Debug] dit_step40_xt: [2170, 64] first4: 0.565408 1.351030 -0.018493 -0.018156 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.556297 -0.955750 -0.027427 0.588272 [Debug] dit_step41_xt: [2170, 64] first4: 0.576534 1.370145 -0.017945 -0.029922 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.545105 -1.019773 0.019623 0.543564 [Debug] dit_step42_xt: [2170, 64] first4: 0.587436 1.390541 -0.018337 -0.040793 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.536591 -1.073090 0.064308 0.484711 [Debug] dit_step43_xt: [2170, 64] first4: 0.598168 1.412003 -0.019623 -0.050487 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.519243 -1.128494 0.109482 0.437927 [Debug] dit_step44_xt: [2170, 64] first4: 0.608553 1.434572 -0.021813 -0.059246 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.507538 -1.169708 0.134506 0.375061 [Debug] dit_step45_xt: [2170, 64] first4: 0.618703 1.457967 -0.024503 -0.066747 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.474533 -1.210648 0.172859 0.321289 [Debug] dit_step46_xt: [2170, 64] first4: 0.628194 1.482180 -0.027960 -0.073173 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.451233 -1.252197 0.195976 0.274046 [Debug] dit_step47_xt: [2170, 64] first4: 0.637219 1.507224 -0.031880 -0.078654 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.406464 -1.280365 0.210144 0.228386 [Debug] dit_step48_xt: [2170, 64] first4: 0.645348 1.532831 -0.036083 -0.083221 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.451080 -1.315163 0.183350 0.230469 [Debug] dit_x0: [2170, 64] first4: 0.654370 1.559134 -0.039750 -0.087831 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 1653.8 ms (1653.8 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.654370 1.559134 -0.039750 -0.087831 [Store] Unload DiT (1237.2 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: Vulkan0 (CPU threads: 16) [VAE] Backend: Vulkan0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 675 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: 760.2 ms [Debug] vae_audio: [2, 4166400] first4: -0.001607 -0.001132 -0.001420 -0.000997 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 401 ms (216.55x realtime), 32 threads [MP3] Wrote ggml-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 10:17:30.179 | 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 10:17:32.013 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:17:32.013 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:17:32.014 | 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 10:17:32.017 | 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 10:17:32.021 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:17:32.026 | 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 10:17:32.028 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:17:32.028 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:17:32.028 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:17:32.028 | 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 10:17:32.028 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:17:32.028 | 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 10:17:32.028 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:17:32.031 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:17:32.040 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:17:32.040 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:17:32.046 | 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-Q6_K.gguf [GGML] Running acestep-v15-turbo-Q6_K.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 0.999665 detok_output 0.999971 context 0.999982 noise 1.000000 temb_t 0.999990 hidden_after_proj_in 0.999982 enc_after_cond_emb 0.999691 layer0_sa_output 0.999775 hidden_after_layer0 0.999717 hidden_after_layer6 0.999857 hidden_after_layer12 0.998896 hidden_after_layer18 0.995868 hidden_after_layer23 0.992192 dit_step0_vt 0.970543 dit_step0_xt 0.999935 dit_step1_vt 0.973016 dit_step1_xt 0.999793 dit_step2_vt 0.976125 dit_step2_xt 0.999446 dit_step3_vt 0.976445 dit_step3_xt 0.998660 dit_step4_vt 0.974995 dit_step4_xt 0.996862 dit_step5_vt 0.972620 dit_step5_xt 0.992876 dit_step6_vt 0.969465 dit_step6_xt 0.985355 dit_step7_vt 0.962099 dit_x0 0.974228 vae_audio 0.890325 vae_audio (STFT cosine) 0.969933 [Turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999935 0.147067 0.007355 -0.002249 0.973078 -0.002342 0.972003 dit_step1_xt 0.999793 0.293077 0.012740 -0.005185 0.942937 -0.005313 0.941730 dit_step2_xt 0.999446 0.478717 0.019770 -0.009119 0.909517 -0.009311 0.908527 dit_step3_xt 0.998660 0.736076 0.029449 -0.014401 0.874221 -0.014577 0.873624 dit_step4_xt 0.996862 1.054360 0.043053 -0.021403 0.842412 -0.021660 0.841995 dit_step5_xt 0.992876 1.527437 0.063223 -0.031421 0.825720 -0.032109 0.824593 dit_step6_xt 0.985355 2.179775 0.093596 -0.045617 0.856508 -0.046482 0.855546 [SFT] steps=50, shift=1.0 | acestep-v15-sft-Q6_K.gguf [GGML] Running acestep-v15-sft-Q6_K.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-Q6_K.gguf: 830 tensors, data at offset 69088 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q6_K.gguf: 830 tensors, data at offset 69088 [Synth-Load] Ready: turbo=yes, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=8, guidance=1.0, shift=3.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/Qwen3-Embedding-0.6B-BF16.gguf: 310 tensors, data at offset 5337568 [Load] TextEncoder: 28L, H=1024, Nh=16/8 [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 310 tensors, 1136.5 MB into backend [Store] Load TextEnc: 309 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-Q6_K.gguf: 830 tensors, data at offset 69088 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 141 tensors, 476.3 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L, CLS), text_proj, null_cond [Store] Load CondEnc: 272 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, 220.3 ms [Debug] enc_hidden: [238, 2048] first4: 0.687718 -0.019435 -0.051748 0.022811 [Store] Unload CondEnc (476.3 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-Q6_K.gguf: 830 tensors, data at offset 69088 [WeightCtx] Loaded 30 tensors, 82.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 82 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 165.0 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.141068 1.454457 0.315091 -0.623697 [Store] Unload FSQ-Detok (82.2 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: -0.141068 1.454457 0.315091 -0.623697 [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-Q6_K.gguf: 830 tensors, data at offset 69088 [DiT] Self-attn: Q+K+V fused [DiT] Cross-attn: Q+K+V fused [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 630 tensors, 3267.8 MB into backend [Load] DiT: 32 layers, H=2560, Nh=32/8, D=128 [Store] Load DiT: 3376 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: 0.019342 0.105586 0.065691 -0.074838 [Debug] temb: [2560] first4: 0.026559 -0.022629 -0.016440 0.046999 [Debug] temb_t: [2560] first4: 0.014705 -0.022100 -0.020716 0.048496 [Debug] temb_r: [2560] first4: 0.011854 -0.000528 0.004277 -0.001497 [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: 0.062749 -0.017898 0.137387 -0.111467 [Debug] temb_lin1_r: [2560] first4: -0.011992 0.015308 -0.017417 0.011657 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.338501 -0.580811 0.254028 -0.190186 [Debug] proj_in_input: [192, 2170] first4: -0.141068 1.454457 0.315091 -0.623697 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.030426 0.668320 0.102020 0.862503 [Debug] layer0_sa_input: [2560, 1085] first4: 0.042987 -0.049411 0.095042 -0.135277 [Debug] layer0_q_after_rope: [128, 32] first4: -0.256193 -2.178976 0.150716 0.051828 [Debug] layer0_k_after_rope: [128, 8] first4: 0.562012 2.197266 0.705078 0.185669 [Debug] layer0_sa_output: [2560, 1085] first4: 1.857422 4.464844 -0.962891 5.574219 [Debug] layer0_attn_out: [4096, 1085] first4: 0.487817 -2.312559 -0.447873 -2.875886 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.487817 -2.312559 -0.447873 -2.875886 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.327661 -2.048887 -0.300290 -3.768952 [Debug] hidden_after_layer0: [2560, 1085] first4: 1.648848 -2.464514 238.682541 -3.743355 [Debug] hidden_after_layer6: [2560, 1085] first4: -6.260375 -2.852511 317.295929 -12.720048 [Debug] hidden_after_layer12: [2560, 1085] first4: -10.634361 29.627888 257.109894 31.472534 [Debug] hidden_after_layer18: [2560, 1085] first4: 42.132824 10.771816 273.312897 4.987638 [Debug] hidden_after_layer31: [2560, 1085] first4: -83.018341 -278.274200 -125.700226 216.749130 [Debug] dit_step0_vt: [2170, 64] first4: -1.048584 1.720581 0.838043 1.325562 [Debug] dit_step0_xt: [2170, 64] first4: 0.241999 2.078042 -0.209968 0.787403 [DiT] Step 1/8 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -1.078491 1.936554 1.091003 1.564880 [Debug] dit_step1_xt: [2170, 64] first4: 0.300826 1.972412 -0.269477 0.702046 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: -0.500610 1.410751 0.875992 0.874756 [Debug] dit_step2_xt: [2170, 64] first4: 0.334200 1.878361 -0.327877 0.643729 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: -0.340576 1.375011 0.912453 0.942070 [Debug] dit_step3_xt: [2170, 64] first4: 0.362581 1.763777 -0.403914 0.565223 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: -0.719574 1.478092 1.320892 0.780151 [Debug] dit_step4_xt: [2170, 64] first4: 0.439678 1.605410 -0.545439 0.481636 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: -1.159912 2.325653 0.493408 0.716675 [Debug] dit_step5_xt: [2170, 64] first4: 0.605380 1.273174 -0.615925 0.379254 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: 0.851216 2.056629 0.561623 0.701782 [Debug] dit_step6_xt: [2170, 64] first4: 0.435137 0.861848 -0.728250 0.238897 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: -1.909485 1.894165 1.029694 0.141724 [Debug] dit_x0: [2170, 64] first4: 1.007982 0.293599 -1.037158 0.196380 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 701.6 ms (701.6 ms/sample) [Debug] dit_output: [2170, 64] first4: 1.007982 0.293599 -1.037158 0.196380 [Store] Unload DiT (3267.8 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: 677 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: 762.4 ms [Debug] vae_audio: [2, 4166400] first4: 0.001194 0.003030 0.002848 0.002630 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 441 ms (196.99x realtime), 32 threads [MP3] Wrote ggml-xl-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 10:17:42.425 | 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 0.999665 detok_output 0.999971 context 0.999982 noise 1.000000 temb_t 0.999973 hidden_after_proj_in 0.999981 enc_after_cond_emb 0.999692 layer0_sa_output 0.999789 hidden_after_layer0 0.999778 hidden_after_layer6 0.999732 hidden_after_layer12 0.999284 hidden_after_layer18 0.998483 hidden_after_layer23 0.998801 dit_step0_vt 0.998680 dit_step0_xt 1.000000 dit_step5_vt 0.999060 dit_step5_xt 0.999973 dit_step10_vt 0.998416 dit_step10_xt 0.999904 dit_step15_vt 0.997202 dit_step15_xt 0.999745 dit_step20_vt 0.995440 dit_step20_xt 0.999388 dit_step25_vt 0.992189 dit_step25_xt 0.998753 dit_step30_vt 0.987535 dit_step30_xt 0.997688 dit_step35_vt 0.982236 dit_step35_xt 0.996266 dit_step40_vt 0.978644 dit_step40_xt 0.994750 dit_step45_vt 0.982049 dit_step45_xt 0.993627 dit_step49_vt 0.988029 dit_x0 0.993229 vae_audio 0.967632 vae_audio (STFT cosine) 0.987201 [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 0.020319 0.001499 -0.001652 0.978627 -0.001619 0.978872 dit_step5_xt 0.999973 0.086223 0.005217 -0.008311 0.885672 -0.008480 0.883811 dit_step10_xt 0.999904 0.109421 0.008487 -0.014860 0.802089 -0.015075 0.802324 dit_step15_xt 0.999745 0.152008 0.012644 -0.021316 0.731739 -0.021933 0.731304 dit_step20_xt 0.999388 0.215439 0.018153 -0.027772 0.679004 -0.028882 0.679066 dit_step25_xt 0.998753 0.355048 0.024698 -0.034265 0.648314 -0.035643 0.650780 dit_step30_xt 0.997688 0.594358 0.033162 -0.040771 0.642933 -0.042577 0.647107 dit_step35_xt 0.996266 0.858273 0.043223 -0.047261 0.663617 -0.049471 0.669256 dit_step40_xt 0.994750 1.120473 0.054524 -0.053751 0.707979 -0.056532 0.715783 dit_step45_xt 0.993627 1.354990 0.065298 -0.060315 0.771739 -0.063533 0.780575 [Xl-turbo] steps=8, shift=3.0 | acestep-v15-xl-turbo-Q6_K.gguf [GGML] Running acestep-v15-xl-turbo-Q6_K.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 10:17:45.810 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:17:45.810 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:17:45.810 | 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 10:17:45.813 | INFO | acestep.core.generation.handler.generate_music:_vram_preflight_check:156 - [generate_music] VRAM pre-flight: 76.76 GB free, ~1.23 GB needed (batch=1, duration=88s, mode=xl_turbo). 2026-07-12 10:17:45.820 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:17:45.826 | 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 10:17:45.828 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:17:45.828 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:17:45.828 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:17:45.828 | 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 10:17:45.828 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:17:45.828 | 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 10:17:45.828 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:17:45.832 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:17:45.839 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:17:45.839 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:17:45.846 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 10:17:45.852 | 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 10:17:46.410 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 10:17:46.419 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:64 - [generate_music] pred_latents: torch.Size([1, 2170, 64]), dtype=torch.bfloat16 2026-07-12 10:17:46.419 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.006376981735229492, 'diffusion_time_cost': 0.5568869113922119, 'diffusion_per_step_time_cost': 0.06961086392402649, 'total_time_cost': 0.5632638931274414, 'offload_time_cost': 0.0} 2026-07-12 10:17:46.433 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 10:17:46.442 | DEBUG | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:136 - [generate_music] Before VAE decode: allocated=17.03GB, max=18.33GB 2026-07-12 10:17:46.443 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:154 - [generate_music] Effective free VRAM before VAE decode: 76.67 GB 2026-07-12 10:17:46.443 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:172 - [generate_music] Using tiled VAE decode to reduce VRAM usage... 2026-07-12 10:17:46.443 | DEBUG | acestep.core.generation.handler.memory_utils:_get_auto_decode_chunk_size:108 - [_get_auto_decode_chunk_size] Effective free VRAM: 76.67 GB 2026-07-12 10:17:46.443 | DEBUG | acestep.core.generation.handler.memory_utils:_should_offload_wav_to_cpu:131 - [_should_offload_wav_to_cpu] Effective free VRAM: 76.67 GB 2026-07-12 10:17:46.443 | INFO | acestep.core.generation.handler.vae_decode:tiled_decode:56 - [tiled_decode] chunk_size=512, offload_wav_to_cpu=False, latents_shape=torch.Size([1, 64, 2170]) 2026-07-12 10:17:46.714 | DEBUG | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:193 - [generate_music] After VAE decode: allocated=17.20GB, max=18.49GB 2026-07-12 10:17:46.916 | 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 10:17:46.920 | 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-Q6_K.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-Q6_K.gguf: 830 tensors, data at offset 69056 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q6_K.gguf: 830 tensors, data at offset 69056 [Synth-Load] Ready: turbo=no, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=50, guidance=1.0, shift=1.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/Qwen3-Embedding-0.6B-BF16.gguf: 310 tensors, data at offset 5337568 [Load] TextEncoder: 28L, H=1024, Nh=16/8 [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 310 tensors, 1136.5 MB into backend [Store] Load TextEnc: 470 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-Q6_K.gguf: 830 tensors, data at offset 69056 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 141 tensors, 476.3 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L, CLS), text_proj, null_cond [Store] Load CondEnc: 278 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, 17.9 ms [Debug] enc_hidden: [238, 2048] first4: 0.687718 -0.019435 -0.051748 0.022811 [Store] Unload CondEnc (476.3 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-Q6_K.gguf: 830 tensors, data at offset 69056 [WeightCtx] Loaded 30 tensors, 82.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 57 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 160.7 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.141068 1.454457 0.315091 -0.623697 [Store] Unload FSQ-Detok (82.2 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: -0.141068 1.454457 0.315091 -0.623697 [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-Q6_K.gguf: 830 tensors, data at offset 69056 [DiT] Self-attn: Q+K+V fused [DiT] Cross-attn: Q+K+V fused [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 630 tensors, 3267.8 MB into backend [Load] DiT: 32 layers, H=2560, Nh=32/8, D=128 [Store] Load DiT: 1924 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: 0.010424 0.102531 0.074302 -0.034202 [Debug] temb: [2560] first4: 0.009747 -0.007826 -0.006004 0.004595 [Debug] temb_t: [2560] first4: 0.008220 -0.008037 -0.006556 0.003609 [Debug] temb_r: [2560] first4: 0.001527 0.000211 0.000553 0.000986 [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: 0.033989 -0.083181 0.088318 -0.064778 [Debug] temb_lin1_r: [2560] first4: 0.009306 -0.004837 -0.002965 -0.011076 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.328003 -0.496826 0.124268 -0.213867 [Debug] proj_in_input: [192, 2170] first4: -0.141068 1.454457 0.315091 -0.623697 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.038528 0.641396 0.115234 0.859810 [Debug] layer0_sa_input: [2560, 1085] first4: 0.034297 -0.022830 0.100420 -0.099694 [Debug] layer0_q_after_rope: [128, 32] first4: -0.273979 -2.269409 0.038234 -0.161516 [Debug] layer0_k_after_rope: [128, 8] first4: 0.649414 3.011719 0.793945 1.775391 [Debug] layer0_sa_output: [2560, 1085] first4: 2.236328 4.519531 -1.232422 8.109375 [Debug] layer0_attn_out: [4096, 1085] first4: 0.475312 -2.229294 -0.790678 -4.241121 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.475312 -2.229294 -0.790678 -4.241121 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.285004 -1.943161 -0.728270 -4.301516 [Debug] hidden_after_layer0: [2560, 1085] first4: -1.655906 -1.273209 248.751144 -4.150964 [Debug] hidden_after_layer6: [2560, 1085] first4: -7.759213 3.853369 308.566681 -21.165880 [Debug] hidden_after_layer12: [2560, 1085] first4: -21.793814 38.762608 225.457047 17.860680 [Debug] hidden_after_layer18: [2560, 1085] first4: 10.562727 -13.031769 233.333313 19.542219 [Debug] hidden_after_layer31: [2560, 1085] first4: 66.888657 -104.319061 9.669575 87.487099 [Debug] dit_step0_vt: [2170, 64] first4: -0.314819 2.212646 0.035889 1.679993 [Debug] dit_step0_xt: [2170, 64] first4: 0.200632 2.111997 -0.172593 0.814056 [DiT] Step 1/50 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.343628 2.151489 0.106445 1.533875 [Debug] dit_step1_xt: [2170, 64] first4: 0.207505 2.068967 -0.174722 0.783379 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.323730 2.164490 0.106201 1.546814 [Debug] dit_step2_xt: [2170, 64] first4: 0.213979 2.025678 -0.176846 0.752443 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.269409 2.124939 0.113770 1.545364 [Debug] dit_step3_xt: [2170, 64] first4: 0.219368 1.983179 -0.179121 0.721535 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.267700 2.037476 0.100220 1.566513 [Debug] dit_step4_xt: [2170, 64] first4: 0.224722 1.942429 -0.181126 0.690205 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.252930 2.024689 0.101685 1.576385 [Debug] dit_step5_xt: [2170, 64] first4: 0.229780 1.901936 -0.183159 0.658677 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.251709 1.994202 0.122437 1.594147 [Debug] dit_step6_xt: [2170, 64] first4: 0.234814 1.862052 -0.185608 0.626794 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.255127 1.958893 0.144409 1.593811 [Debug] dit_step7_xt: [2170, 64] first4: 0.239917 1.822874 -0.188496 0.594918 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.270874 1.927673 0.150879 1.592224 [Debug] dit_step8_xt: [2170, 64] first4: 0.245334 1.784320 -0.191514 0.563074 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.302734 1.867523 0.122559 1.583780 [Debug] dit_step9_xt: [2170, 64] first4: 0.251389 1.746970 -0.193965 0.531398 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.313721 1.823273 0.105225 1.580307 [Debug] dit_step10_xt: [2170, 64] first4: 0.257664 1.710504 -0.196069 0.499792 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.304199 1.776276 0.086914 1.563980 [Debug] dit_step11_xt: [2170, 64] first4: 0.263748 1.674979 -0.197808 0.468512 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.292114 1.728977 0.084595 1.551758 [Debug] dit_step12_xt: [2170, 64] first4: 0.269590 1.640399 -0.199500 0.437477 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.294189 1.685463 0.067139 1.537857 [Debug] dit_step13_xt: [2170, 64] first4: 0.275474 1.606690 -0.200842 0.406720 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.284668 1.621246 0.048828 1.515625 [Debug] dit_step14_xt: [2170, 64] first4: 0.281167 1.574265 -0.201819 0.376408 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.277222 1.580658 0.030273 1.514801 [Debug] dit_step15_xt: [2170, 64] first4: 0.286711 1.542652 -0.202424 0.346112 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.252563 1.531189 0.016479 1.509186 [Debug] dit_step16_xt: [2170, 64] first4: 0.291763 1.512028 -0.202754 0.315928 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.215576 1.477051 -0.006592 1.503143 [Debug] dit_step17_xt: [2170, 64] first4: 0.296074 1.482487 -0.202622 0.285865 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.177124 1.417236 -0.015381 1.500977 [Debug] dit_step18_xt: [2170, 64] first4: 0.299617 1.454142 -0.202314 0.255845 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.145142 1.366699 -0.026611 1.488373 [Debug] dit_step19_xt: [2170, 64] first4: 0.302520 1.426808 -0.201782 0.226078 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.117371 1.328796 -0.027100 1.479492 [Debug] dit_step20_xt: [2170, 64] first4: 0.304867 1.400233 -0.201240 0.196488 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.084534 1.279785 -0.023438 1.474243 [Debug] dit_step21_xt: [2170, 64] first4: 0.306558 1.374637 -0.200772 0.167003 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.064087 1.221680 -0.042603 1.454651 [Debug] dit_step22_xt: [2170, 64] first4: 0.307839 1.350203 -0.199919 0.137910 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.050110 1.177063 -0.049438 1.438873 [Debug] dit_step23_xt: [2170, 64] first4: 0.308842 1.326662 -0.198931 0.109133 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.040833 1.124084 -0.054321 1.420898 [Debug] dit_step24_xt: [2170, 64] first4: 0.309658 1.304180 -0.197844 0.080715 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.031616 1.067566 -0.069458 1.395691 [Debug] dit_step25_xt: [2170, 64] first4: 0.310291 1.282829 -0.196455 0.052801 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.038696 1.009949 -0.067810 1.367218 [Debug] dit_step26_xt: [2170, 64] first4: 0.311064 1.262630 -0.195099 0.025457 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.041626 0.944977 -0.083984 1.349976 [Debug] dit_step27_xt: [2170, 64] first4: 0.311897 1.243730 -0.193419 -0.001543 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.045959 0.882568 -0.077820 1.324829 [Debug] dit_step28_xt: [2170, 64] first4: 0.312816 1.226079 -0.191863 -0.028039 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.059937 0.817291 -0.094055 1.297058 [Debug] dit_step29_xt: [2170, 64] first4: 0.314015 1.209733 -0.189982 -0.053981 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.067261 0.734711 -0.106689 1.281738 [Debug] dit_step30_xt: [2170, 64] first4: 0.315360 1.195039 -0.187848 -0.079615 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.078857 0.660034 -0.110291 1.258850 [Debug] dit_step31_xt: [2170, 64] first4: 0.316937 1.181838 -0.185642 -0.104792 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.095215 0.571266 -0.116150 1.224854 [Debug] dit_step32_xt: [2170, 64] first4: 0.318842 1.170413 -0.183319 -0.129289 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.119629 0.493916 -0.103149 1.204041 [Debug] dit_step33_xt: [2170, 64] first4: 0.321234 1.160535 -0.181256 -0.153370 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.143707 0.399086 -0.119858 1.162476 [Debug] dit_step34_xt: [2170, 64] first4: 0.324108 1.152553 -0.178859 -0.176620 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.136261 0.314327 -0.105423 1.153015 [Debug] dit_step35_xt: [2170, 64] first4: 0.326834 1.146266 -0.176751 -0.199680 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.158390 0.202126 -0.096222 1.107056 [Debug] dit_step36_xt: [2170, 64] first4: 0.330001 1.142224 -0.174826 -0.221821 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.165749 0.102234 -0.077759 1.078064 [Debug] dit_step37_xt: [2170, 64] first4: 0.333316 1.140179 -0.173271 -0.243382 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.181122 0.003342 -0.083237 1.033630 [Debug] dit_step38_xt: [2170, 64] first4: 0.336939 1.140112 -0.171606 -0.264055 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.219666 -0.121445 -0.067673 0.995789 [Debug] dit_step39_xt: [2170, 64] first4: 0.341332 1.142541 -0.170253 -0.283971 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.238220 -0.225677 -0.041656 0.963501 [Debug] dit_step40_xt: [2170, 64] first4: 0.346096 1.147055 -0.169420 -0.303241 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.269897 -0.337600 -0.014259 0.914429 [Debug] dit_step41_xt: [2170, 64] first4: 0.351494 1.153807 -0.169134 -0.321529 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.285095 -0.423996 0.002104 0.865112 [Debug] dit_step42_xt: [2170, 64] first4: 0.357196 1.162287 -0.169176 -0.338832 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.320801 -0.533463 0.024017 0.814697 [Debug] dit_step43_xt: [2170, 64] first4: 0.363612 1.172956 -0.169657 -0.355126 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.351013 -0.626686 0.076813 0.760132 [Debug] dit_step44_xt: [2170, 64] first4: 0.370633 1.185490 -0.171193 -0.370328 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.389526 -0.700615 0.117767 0.715088 [Debug] dit_step45_xt: [2170, 64] first4: 0.378423 1.199502 -0.173548 -0.384630 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.438286 -0.775520 0.147339 0.671631 [Debug] dit_step46_xt: [2170, 64] first4: 0.387189 1.215012 -0.176495 -0.398063 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.494186 -0.844978 0.200623 0.636780 [Debug] dit_step47_xt: [2170, 64] first4: 0.397073 1.231912 -0.180508 -0.410798 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.548935 -0.896229 0.258606 0.632202 [Debug] dit_step48_xt: [2170, 64] first4: 0.408051 1.249837 -0.185680 -0.423442 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.545265 -0.886650 0.192444 0.643555 [Debug] dit_x0: [2170, 64] first4: 0.418957 1.267570 -0.189529 -0.436313 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 4086.2 ms (4086.2 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.418957 1.267570 -0.189529 -0.436313 [Store] Unload DiT (3267.8 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: 702 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: 768.9 ms [Debug] vae_audio: [2, 4166400] first4: -0.001417 -0.000703 -0.000832 -0.000345 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 416 ms (208.49x realtime), 32 threads [MP3] Wrote ggml-xl-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 10:17:57.044 | 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 0.999396 detok_output 0.999971 context 0.999982 noise 1.000000 temb_t 0.999978 hidden_after_proj_in 0.999972 enc_after_cond_emb 0.999486 layer0_sa_output 0.989082 hidden_after_layer0 0.998996 hidden_after_layer6 0.999659 hidden_after_layer12 0.994662 hidden_after_layer18 0.995266 hidden_after_layer31 0.986926 dit_step0_vt 0.904214 dit_step0_xt 0.999828 dit_step1_vt 0.919311 dit_step1_xt 0.999452 dit_step2_vt 0.912860 dit_step2_xt 0.998489 dit_step3_vt 0.909887 dit_step3_xt 0.996175 dit_step4_vt 0.905245 dit_step4_xt 0.990561 dit_step5_vt 0.895300 dit_step5_xt 0.977283 dit_step6_vt 0.887474 dit_step6_xt 0.951478 dit_step7_vt 0.864989 dit_x0 0.910392 vae_audio 0.703463 vae_audio (STFT cosine) 0.903330 [Xl-turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999828 0.151054 0.013073 -0.002893 0.978424 -0.002721 0.976534 dit_step1_xt 0.999452 0.309860 0.022808 -0.005841 0.952588 -0.005567 0.949004 dit_step2_xt 0.998489 0.523092 0.036133 -0.009687 0.924039 -0.009330 0.918931 dit_step3_xt 0.996175 0.831504 0.054552 -0.015285 0.893403 -0.014418 0.887241 dit_step4_xt 0.990561 1.334398 0.082131 -0.022966 0.864828 -0.021453 0.857796 dit_step5_xt 0.977283 2.011531 0.124790 -0.034364 0.848542 -0.032240 0.840033 dit_step6_xt 0.951478 2.845881 0.188606 -0.051119 0.872801 -0.049194 0.863964 [Xl-sft] steps=50, shift=1.0 | acestep-v15-xl-sft-Q6_K.gguf [GGML] Running acestep-v15-xl-sft-Q6_K.gguf... [GGML] Done, 131 dump files [Python] Initializing acestep-v15-xl-sft... Loading checkpoint shards: 0%| | 0/4 [00:00 2026-07-12 10:18:01.827 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:18:01.827 | 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 10:18:01.827 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:18:01.831 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:18:01.838 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:18:01.838 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:18:01.845 | 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