[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-Q5_K_M.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-Q5_K_M.gguf: 678 tensors, data at offset 56864 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q5_K_M.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: 327 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-Q5_K_M.gguf: 678 tensors, data at offset 56864 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 140 tensors, 412.5 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L), text_proj, null_cond [Store] Load CondEnc: 716 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, 3549.8 ms [Debug] enc_hidden: [238, 2048] first4: 1.760517 -0.051722 -0.132193 0.058110 [Store] Unload CondEnc (412.5 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] Detokenizer backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q5_K_M.gguf: 678 tensors, data at offset 56864 [WeightCtx] Loaded 30 tensors, 73.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 67 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 568.0 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.126205 1.455492 0.291894 -0.650703 [Store] Unload FSQ-Detok (73.2 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: -0.126205 1.455492 0.291894 -0.650703 [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-Q5_K_M.gguf: 678 tensors, data at offset 56864 [DiT] Self-attn: Q+K fused, V separate [DiT] Cross-attn: all separate [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 478 tensors, 1061.2 MB into backend [Load] DiT: 24 layers, H=2048, Nh=16/8, D=128 [Store] Load DiT: 809 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2063 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [12288] first4: 0.260409 -0.161609 -0.102203 0.051602 [Debug] temb: [2048] first4: -0.000151 -0.132293 -0.035516 0.064751 [Debug] temb_t: [2048] first4: 0.000578 0.026708 -0.052786 0.063514 [Debug] temb_r: [2048] first4: -0.000729 -0.159001 0.017269 0.001237 [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: -0.051153 -0.053631 -0.012192 -0.039024 [Debug] temb_lin1_r: [2048] first4: -0.016165 -0.021121 -0.015801 -0.000525 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.043457 -0.948303 0.536377 0.453583 [Debug] proj_in_input: [192, 2170] first4: -0.126205 1.455492 0.291894 -0.650703 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.155807 0.748947 0.320251 -0.524521 [Debug] layer0_sa_input: [2048, 1085] first4: -0.721759 -0.751600 -0.052395 0.263968 [Debug] layer0_q_after_rope: [128, 16] first4: -0.155807 0.748947 0.320251 -0.524521 [Debug] layer0_k_after_rope: [128, 8] first4: -1.540865 -1.013701 0.146717 0.464547 [Debug] layer0_sa_output: [2048, 1085] first4: -1.498047 0.148193 -0.400635 0.486572 [Debug] layer0_attn_out: [2048, 1085] first4: -1.540865 -1.013701 0.146717 0.464547 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.540865 -1.013701 0.146717 0.464547 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.580019 -0.768828 -0.338879 0.501015 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.068409 0.594532 50.598301 -0.869675 [Debug] hidden_after_layer6: [2048, 1085] first4: -20.365084 0.535518 30.570795 -4.750030 [Debug] hidden_after_layer12: [2048, 1085] first4: -17.782322 -17.607498 73.563065 26.600998 [Debug] hidden_after_layer18: [2048, 1085] first4: -25.504398 9.966521 66.026413 18.954729 [Debug] hidden_after_layer23: [2048, 1085] first4: -3.669283 32.806488 200.278412 155.354309 [Debug] dit_step0_vt: [2170, 64] first4: 0.025371 1.151047 0.333879 2.326447 [Debug] dit_step0_xt: [2170, 64] first4: 0.193183 2.103930 -0.187051 0.741909 [DiT] Step 1/8 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.206284 1.348145 -0.162842 1.874989 [Debug] dit_step1_xt: [2170, 64] first4: 0.204435 2.030395 -0.178169 0.639637 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: -0.047316 1.256003 0.061066 2.357178 [Debug] dit_step2_xt: [2170, 64] first4: 0.207589 1.946661 -0.182240 0.482491 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: 0.279205 1.102295 0.225555 2.597809 [Debug] dit_step3_xt: [2170, 64] first4: 0.184322 1.854803 -0.201036 0.266007 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: 0.328033 0.925766 0.069183 2.673800 [Debug] dit_step4_xt: [2170, 64] first4: 0.149175 1.755614 -0.208449 -0.020471 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: 0.328247 0.794983 -0.328369 2.678589 [Debug] dit_step5_xt: [2170, 64] first4: 0.102283 1.642045 -0.161539 -0.403127 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: 0.206287 0.609131 -0.831482 2.758667 [Debug] dit_step6_xt: [2170, 64] first4: 0.061025 1.520219 0.004757 -0.954860 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: -0.059860 0.266479 -1.305878 3.130249 [Debug] dit_x0: [2170, 64] first4: 0.078984 1.440275 0.396521 -1.893935 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 285.7 ms (285.7 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.078984 1.440275 0.396521 -1.893935 [Store] Unload DiT (1061.2 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: 669 ms [VAE] Tiled decode: 3 tiles (chunk=1024, overlap=64, stride=896) [VAE] Graph: 479 nodes, T_latent=960 [VAE] Upsample factor: 1920.00 (expected ~1920) [VAE] Graph: 479 nodes, T_latent=1024 [VAE] Graph: 479 nodes, T_latent=442 [VAE] Tiled decode done: 3 tiles -> T_audio=4166400 (86.80s @ 48kHz) [VAE-Decode Batch0] Decode: 761.3 ms [Debug] vae_audio: [2, 4166400] first4: 0.000954 0.001687 0.001462 0.002010 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 455 ms (190.94x realtime), 32 threads [MP3] Wrote ggml-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 10:18:16.919 | 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 10:18:18.956 | WARNING | acestep.training.trainer::40 - bitsandbytes not installed. Using standard AdamW. 2026-07-12 10:18:18.961 | 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 10:18:19.031 | 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 10:18:21.626 | 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 10:18:21.626 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:18:21.626 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:18:21.626 | 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 10:18:21.630 | 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 10:18:21.635 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:18:21.850 | 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 10:18:21.852 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:18:21.852 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:18:21.852 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:18:21.852 | 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 10:18:21.852 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:18:21.852 | 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 10:18:21.852 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:18:21.860 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:18:21.894 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:18:21.894 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:18:21.923 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 10:18:21.931 | 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 10:18:22.016 | 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 10:18:22.242 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 10:18:22.243 | 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 10:18:22.243 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.006852149963378906, 'diffusion_time_cost': 0.3118259906768799, 'diffusion_per_step_time_cost': 0.038978248834609985, 'total_time_cost': 0.3186781406402588, 'offload_time_cost': 0.0} 2026-07-12 10:18:22.257 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 10:18:22.259 | 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 10:18:22.259 | 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 10:18:22.259 | 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 10:18:22.260 | 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 10:18:22.260 | 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 10:18:22.260 | 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 10:18:22.536 | 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 10:18:22.697 | 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 10:18:22.700 | 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-Q5_K_M.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-Q5_K_M.gguf: 678 tensors, data at offset 56800 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q5_K_M.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: 295 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-Q5_K_M.gguf: 678 tensors, data at offset 56800 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 140 tensors, 412.5 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L), text_proj, null_cond [Store] Load CondEnc: 693 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, 8.6 ms [Debug] enc_hidden: [238, 2048] first4: 1.760517 -0.051722 -0.132193 0.058110 [Store] Unload CondEnc (412.5 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] Detokenizer backend: Vulkan0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q5_K_M.gguf: 678 tensors, data at offset 56800 [WeightCtx] Loaded 30 tensors, 73.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 50 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 155.5 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.126205 1.455492 0.291894 -0.650703 [Store] Unload FSQ-Detok (73.2 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: -0.126205 1.455492 0.291894 -0.650703 [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-Q5_K_M.gguf: 678 tensors, data at offset 56800 [DiT] Self-attn: Q+K fused, V separate [DiT] Cross-attn: all separate [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 478 tensors, 1061.2 MB into backend [Load] DiT: 24 layers, H=2048, Nh=16/8, D=128 [Store] Load DiT: 773 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2063 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [12288] first4: 0.155137 -0.118305 -0.084248 0.082623 [Debug] temb: [2048] first4: -0.002843 -0.176820 0.004745 -0.001924 [Debug] temb_t: [2048] first4: -0.001351 0.003023 -0.012552 -0.001712 [Debug] temb_r: [2048] first4: -0.001491 -0.179843 0.017298 -0.000212 [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: -0.034754 0.028817 0.027120 -0.031729 [Debug] temb_lin1_r: [2048] first4: -0.002680 0.004202 0.000655 -0.002088 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.093750 -0.904541 0.498291 0.448029 [Debug] proj_in_input: [192, 2170] first4: -0.126205 1.455492 0.291894 -0.650703 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.175980 0.837662 0.353027 -0.564377 [Debug] layer0_sa_input: [2048, 1085] first4: -0.918976 -0.705216 -0.034956 0.298489 [Debug] layer0_q_after_rope: [128, 16] first4: -0.175980 0.837662 0.353027 -0.564377 [Debug] layer0_k_after_rope: [128, 8] first4: -1.804577 -1.343851 -0.172177 0.389113 [Debug] layer0_sa_output: [2048, 1085] first4: -1.642578 0.773438 -0.594727 0.532715 [Debug] layer0_attn_out: [2048, 1085] first4: -1.804577 -1.343851 -0.172177 0.389113 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.804577 -1.343851 -0.172177 0.389113 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.963147 -0.975931 -0.431699 0.408095 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.506083 1.333558 57.889484 -1.651999 [Debug] hidden_after_layer6: [2048, 1085] first4: -14.857037 3.700705 57.577919 -0.212808 [Debug] hidden_after_layer12: [2048, 1085] first4: -12.710327 0.135689 -23.786119 -0.411003 [Debug] hidden_after_layer18: [2048, 1085] first4: -5.429951 15.575186 -43.283718 -9.827290 [Debug] hidden_after_layer23: [2048, 1085] first4: 31.953632 61.223022 49.807259 -14.503943 [Debug] dit_step0_vt: [2170, 64] first4: -0.641724 2.509766 -0.195007 1.440453 [Debug] dit_step0_xt: [2170, 64] first4: 0.207170 2.106055 -0.167975 0.818847 [DiT] Step 1/50 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.707092 2.520752 -0.055862 1.249969 [Debug] dit_step1_xt: [2170, 64] first4: 0.221312 2.055640 -0.166858 0.793848 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.699463 2.483521 -0.010895 1.275681 [Debug] dit_step2_xt: [2170, 64] first4: 0.235302 2.005969 -0.166640 0.768334 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.683350 2.420776 0.027679 1.295914 [Debug] dit_step3_xt: [2170, 64] first4: 0.248968 1.957554 -0.167193 0.742416 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.617676 2.334717 0.029968 1.297104 [Debug] dit_step4_xt: [2170, 64] first4: 0.261322 1.910859 -0.167793 0.716474 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.573364 2.254883 0.008667 1.298843 [Debug] dit_step5_xt: [2170, 64] first4: 0.272789 1.865762 -0.167966 0.690497 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.534851 2.207275 -0.007507 1.293152 [Debug] dit_step6_xt: [2170, 64] first4: 0.283486 1.821616 -0.167816 0.664634 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.506104 2.140503 -0.025757 1.291321 [Debug] dit_step7_xt: [2170, 64] first4: 0.293608 1.778806 -0.167301 0.638808 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.464783 2.063049 -0.026096 1.281891 [Debug] dit_step8_xt: [2170, 64] first4: 0.302904 1.737545 -0.166779 0.613170 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.406372 1.973022 -0.016159 1.270966 [Debug] dit_step9_xt: [2170, 64] first4: 0.311031 1.698085 -0.166456 0.587750 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.351379 1.893433 -0.000069 1.254684 [Debug] dit_step10_xt: [2170, 64] first4: 0.318059 1.660216 -0.166454 0.562657 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.316406 1.820557 -0.000183 1.247009 [Debug] dit_step11_xt: [2170, 64] first4: 0.324387 1.623805 -0.166451 0.537717 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.289673 1.746704 -0.000641 1.228943 [Debug] dit_step12_xt: [2170, 64] first4: 0.330181 1.588871 -0.166438 0.513138 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.278503 1.680603 -0.011887 1.218796 [Debug] dit_step13_xt: [2170, 64] first4: 0.335751 1.555259 -0.166200 0.488762 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.269562 1.610504 -0.021378 1.204819 [Debug] dit_step14_xt: [2170, 64] first4: 0.341142 1.523049 -0.165772 0.464665 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.268494 1.541458 -0.037979 1.190079 [Debug] dit_step15_xt: [2170, 64] first4: 0.346512 1.492220 -0.165013 0.440864 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.257202 1.453674 -0.059326 1.182373 [Debug] dit_step16_xt: [2170, 64] first4: 0.351656 1.463146 -0.163826 0.417216 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.267578 1.378244 -0.073822 1.159747 [Debug] dit_step17_xt: [2170, 64] first4: 0.357007 1.435581 -0.162350 0.394021 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.279449 1.301514 -0.083740 1.167648 [Debug] dit_step18_xt: [2170, 64] first4: 0.362596 1.409551 -0.160675 0.370669 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.288963 1.223618 -0.102936 1.151321 [Debug] dit_step19_xt: [2170, 64] first4: 0.368376 1.385079 -0.158616 0.347642 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.309130 1.142365 -0.106384 1.129257 [Debug] dit_step20_xt: [2170, 64] first4: 0.374558 1.362231 -0.156489 0.325057 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.327896 1.055389 -0.116852 1.118927 [Debug] dit_step21_xt: [2170, 64] first4: 0.381116 1.341124 -0.154152 0.302678 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.343699 0.964478 -0.138092 1.091110 [Debug] dit_step22_xt: [2170, 64] first4: 0.387990 1.321834 -0.151390 0.280856 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.375406 0.871948 -0.128845 1.077972 [Debug] dit_step23_xt: [2170, 64] first4: 0.395498 1.304395 -0.148813 0.259297 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.395601 0.774658 -0.140045 1.049103 [Debug] dit_step24_xt: [2170, 64] first4: 0.403410 1.288902 -0.146012 0.238315 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.430218 0.675964 -0.137299 1.029510 [Debug] dit_step25_xt: [2170, 64] first4: 0.412015 1.275383 -0.143266 0.217724 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.451081 0.574768 -0.138916 1.005829 [Debug] dit_step26_xt: [2170, 64] first4: 0.421036 1.263887 -0.140488 0.197608 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.483080 0.470375 -0.127747 0.985077 [Debug] dit_step27_xt: [2170, 64] first4: 0.430698 1.254480 -0.137933 0.177906 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.506035 0.356218 -0.133118 0.943115 [Debug] dit_step28_xt: [2170, 64] first4: 0.440819 1.247355 -0.135271 0.159044 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.533081 0.237335 -0.121765 0.924622 [Debug] dit_step29_xt: [2170, 64] first4: 0.451480 1.242609 -0.132835 0.140552 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.548103 0.136307 -0.112534 0.900879 [Debug] dit_step30_xt: [2170, 64] first4: 0.462442 1.239883 -0.130585 0.122534 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.560001 0.023422 -0.100266 0.885925 [Debug] dit_step31_xt: [2170, 64] first4: 0.473642 1.239414 -0.128579 0.104816 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.568299 -0.077255 -0.095306 0.859192 [Debug] dit_step32_xt: [2170, 64] first4: 0.485008 1.240959 -0.126673 0.087632 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.581564 -0.188309 -0.084377 0.835266 [Debug] dit_step33_xt: [2170, 64] first4: 0.496640 1.244725 -0.124986 0.070926 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.588802 -0.278351 -0.055026 0.814819 [Debug] dit_step34_xt: [2170, 64] first4: 0.508416 1.250292 -0.123885 0.054630 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.594746 -0.390350 -0.039513 0.780823 [Debug] dit_step35_xt: [2170, 64] first4: 0.520311 1.258099 -0.123095 0.039014 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.599476 -0.495300 -0.017380 0.749451 [Debug] dit_step36_xt: [2170, 64] first4: 0.532300 1.268005 -0.122747 0.024025 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.590794 -0.588013 0.017365 0.711792 [Debug] dit_step37_xt: [2170, 64] first4: 0.544116 1.279766 -0.123094 0.009789 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.596119 -0.675644 0.037170 0.679565 [Debug] dit_step38_xt: [2170, 64] first4: 0.556038 1.293278 -0.123838 -0.003803 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.580641 -0.762772 0.080627 0.641663 [Debug] dit_step39_xt: [2170, 64] first4: 0.567651 1.308534 -0.125450 -0.016636 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.573952 -0.836243 0.113636 0.599915 [Debug] dit_step40_xt: [2170, 64] first4: 0.579130 1.325259 -0.127723 -0.028634 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.558943 -0.921591 0.155464 0.550819 [Debug] dit_step41_xt: [2170, 64] first4: 0.590309 1.343691 -0.130832 -0.039651 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.540726 -0.985313 0.199669 0.502808 [Debug] dit_step42_xt: [2170, 64] first4: 0.601124 1.363397 -0.134826 -0.049707 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.532143 -1.062855 0.239838 0.453857 [Debug] dit_step43_xt: [2170, 64] first4: 0.611766 1.384654 -0.139623 -0.058784 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.520704 -1.116089 0.271423 0.395386 [Debug] dit_step44_xt: [2170, 64] first4: 0.622180 1.406976 -0.145051 -0.066692 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.497375 -1.168381 0.300766 0.324280 [Debug] dit_step45_xt: [2170, 64] first4: 0.632128 1.430343 -0.151066 -0.073177 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.476692 -1.223755 0.319763 0.279377 [Debug] dit_step46_xt: [2170, 64] first4: 0.641662 1.454818 -0.157462 -0.078765 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.449448 -1.260193 0.330574 0.240448 [Debug] dit_step47_xt: [2170, 64] first4: 0.650651 1.480022 -0.164073 -0.083574 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.411346 -1.296421 0.314270 0.199112 [Debug] dit_step48_xt: [2170, 64] first4: 0.658878 1.505951 -0.170359 -0.087556 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.444992 -1.321194 0.295990 0.190460 [Debug] dit_x0: [2170, 64] first4: 0.667777 1.532375 -0.176278 -0.091365 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 1599.0 ms (1599.0 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.667777 1.532375 -0.176278 -0.091365 [Store] Unload DiT (1061.2 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: 672 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: 791.9 ms [Debug] vae_audio: [2, 4166400] first4: -0.001469 -0.000962 -0.001236 -0.000808 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 417 ms (208.27x realtime), 32 threads [MP3] Wrote ggml-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 10:18:29.266 | 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 10:18:31.224 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:18:31.224 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:18:31.225 | 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 10:18:31.228 | 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 10:18:31.233 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:18:31.237 | 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 10:18:31.238 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:18:31.238 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:18:31.238 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:18:31.238 | 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 10:18:31.238 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:18:31.238 | 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 10:18:31.238 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:18:31.242 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:18:31.251 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:18:31.251 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:18:31.258 | 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-Q5_K_M.gguf [GGML] Running acestep-v15-turbo-Q5_K_M.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 0.999133 detok_output 0.999877 context 0.999921 noise 1.000000 temb_t 0.999972 hidden_after_proj_in 0.999959 enc_after_cond_emb 0.999272 layer0_sa_output 0.999443 hidden_after_layer0 0.999648 hidden_after_layer6 0.999757 hidden_after_layer12 0.998599 hidden_after_layer18 0.995453 hidden_after_layer23 0.991433 dit_step0_vt 0.967908 dit_step0_xt 0.999930 dit_step1_vt 0.970597 dit_step1_xt 0.999780 dit_step2_vt 0.968507 dit_step2_xt 0.999353 dit_step3_vt 0.972874 dit_step3_xt 0.998468 dit_step4_vt 0.971710 dit_step4_xt 0.996489 dit_step5_vt 0.967571 dit_step5_xt 0.991964 dit_step6_vt 0.962763 dit_step6_xt 0.983178 dit_step7_vt 0.953879 dit_x0 0.969943 vae_audio 0.876137 vae_audio (STFT cosine) 0.967571 [Turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999930 0.141518 0.007848 -0.002279 0.973166 -0.002342 0.972003 dit_step1_xt 0.999780 0.264755 0.013513 -0.005264 0.943151 -0.005313 0.941730 dit_step2_xt 0.999353 0.458578 0.021657 -0.009248 0.909938 -0.009311 0.908527 dit_step3_xt 0.998468 0.670735 0.032072 -0.014602 0.874875 -0.014577 0.873624 dit_step4_xt 0.996489 0.972677 0.046476 -0.021754 0.843322 -0.021660 0.841995 dit_step5_xt 0.991964 1.450334 0.068500 -0.031978 0.827067 -0.032109 0.824593 dit_step6_xt 0.983178 2.225803 0.102301 -0.046379 0.858371 -0.046482 0.855546 [SFT] steps=50, shift=1.0 | acestep-v15-sft-Q5_K_M.gguf [GGML] Running acestep-v15-sft-Q5_K_M.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-Q5_K_M.gguf: 830 tensors, data at offset 69088 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q5_K_M.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: 324 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-Q5_K_M.gguf: 830 tensors, data at offset 69088 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 141 tensors, 416.2 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L, CLS), text_proj, null_cond [Store] Load CondEnc: 699 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, 359.5 ms [Debug] enc_hidden: [238, 2048] first4: 0.687591 -0.019866 -0.051357 0.022943 [Store] Unload CondEnc (416.2 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-Q5_K_M.gguf: 830 tensors, data at offset 69088 [WeightCtx] Loaded 30 tensors, 73.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 51 ms [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 161.7 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.126205 1.455492 0.291894 -0.650703 [Store] Unload FSQ-Detok (73.2 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: -0.126205 1.455492 0.291894 -0.650703 [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-Q5_K_M.gguf: 830 tensors, data at offset 69088 [DiT] Self-attn: Q+K fused, V separate [DiT] Cross-attn: all separate [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 630 tensors, 2801.1 MB into backend [Load] DiT: 32 layers, H=2560, Nh=32/8, D=128 [Store] Load DiT: 1914 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2729 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [15360] first4: 0.008611 0.102928 0.078232 -0.076587 [Debug] temb: [2560] first4: 0.026536 -0.023384 -0.016406 0.047140 [Debug] temb_t: [2560] first4: 0.014694 -0.022190 -0.020725 0.048421 [Debug] temb_r: [2560] first4: 0.011842 -0.001194 0.004319 -0.001281 [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: 0.063173 -0.019530 0.138760 -0.115951 [Debug] temb_lin1_r: [2560] first4: -0.008961 0.017832 -0.016914 0.012844 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.347290 -0.579834 0.248169 -0.175293 [Debug] proj_in_input: [192, 2170] first4: -0.126205 1.455492 0.291894 -0.650703 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.041519 0.657150 0.097382 0.871948 [Debug] layer0_sa_input: [2560, 1085] first4: 0.032966 -0.052398 0.107496 -0.132416 [Debug] layer0_q_after_rope: [128, 32] first4: -0.864978 -2.239127 -2.331378 0.583767 [Debug] layer0_k_after_rope: [128, 8] first4: 1.817383 4.375000 -0.879395 5.601562 [Debug] layer0_sa_output: [2560, 1085] first4: 1.817383 4.375000 -0.879395 5.601562 [Debug] layer0_attn_out: [4096, 1085] first4: -0.864978 -2.239127 -2.331378 0.583767 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.488784 -2.281324 -0.395761 -2.849367 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.339492 -2.003736 -0.270639 -3.554934 [Debug] hidden_after_layer0: [2560, 1085] first4: 1.961443 -2.383310 233.163849 -3.531876 [Debug] hidden_after_layer6: [2560, 1085] first4: -6.210892 -2.381814 309.576965 -12.394644 [Debug] hidden_after_layer12: [2560, 1085] first4: -12.720728 28.751617 253.120514 28.455772 [Debug] hidden_after_layer18: [2560, 1085] first4: 46.786480 17.452593 251.881866 7.068016 [Debug] hidden_after_layer31: [2560, 1085] first4: -118.962685 -197.781723 -168.264236 112.052963 [Debug] dit_step0_vt: [2170, 64] first4: -1.402466 1.721573 0.760040 1.399658 [Debug] dit_step0_xt: [2170, 64] first4: 0.258084 2.077997 -0.206422 0.784035 [DiT] Step 1/8 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -1.014771 1.379101 0.831535 1.300903 [Debug] dit_step1_xt: [2170, 64] first4: 0.313436 2.002773 -0.251779 0.713077 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: -0.714417 1.363159 0.919462 0.933777 [Debug] dit_step2_xt: [2170, 64] first4: 0.361063 1.911896 -0.313076 0.650825 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: -0.385284 1.430389 0.915997 0.920319 [Debug] dit_step3_xt: [2170, 64] first4: 0.393170 1.792697 -0.389409 0.574132 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: -0.432167 1.431396 1.059799 0.880928 [Debug] dit_step4_xt: [2170, 64] first4: 0.439474 1.639333 -0.502959 0.479747 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: 0.028061 2.614258 0.716888 0.657837 [Debug] dit_step5_xt: [2170, 64] first4: 0.435465 1.265867 -0.605372 0.385770 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: 0.604919 2.383850 0.587473 0.701172 [Debug] dit_step6_xt: [2170, 64] first4: 0.314481 0.789097 -0.722866 0.245536 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: -1.547607 2.700684 1.176147 0.380249 [Debug] dit_x0: [2170, 64] first4: 0.778764 -0.021108 -1.075711 0.131461 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 658.9 ms (658.9 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.778764 -0.021108 -1.075711 0.131461 [Store] Unload DiT (2801.1 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: 667 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: 789.1 ms [Debug] vae_audio: [2, 4166400] first4: 0.001926 0.003796 0.003710 0.003577 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 457 ms (189.81x realtime), 32 threads [MP3] Wrote ggml-xl-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 10:18:40.699 | 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 0.999133 detok_output 0.999877 context 0.999921 noise 1.000000 temb_t 0.999899 hidden_after_proj_in 0.999959 enc_after_cond_emb 0.999269 layer0_sa_output 0.999521 hidden_after_layer0 0.999790 hidden_after_layer6 0.999605 hidden_after_layer12 0.999035 hidden_after_layer18 0.998078 hidden_after_layer23 0.998548 dit_step0_vt 0.998468 dit_step0_xt 1.000000 dit_step5_vt 0.998542 dit_step5_xt 0.999966 dit_step10_vt 0.997855 dit_step10_xt 0.999878 dit_step15_vt 0.996348 dit_step15_xt 0.999676 dit_step20_vt 0.993777 dit_step20_xt 0.999210 dit_step25_vt 0.988960 dit_step25_xt 0.998341 dit_step30_vt 0.982830 dit_step30_xt 0.996865 dit_step35_vt 0.976226 dit_step35_xt 0.994910 dit_step40_vt 0.971887 dit_step40_xt 0.992871 dit_step45_vt 0.976355 dit_step45_xt 0.991383 dit_step49_vt 0.983869 dit_x0 0.990868 vae_audio 0.958914 vae_audio (STFT cosine) 0.983184 [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 0.018047 0.001572 -0.001635 0.978552 -0.001619 0.978872 dit_step5_xt 0.999966 0.069460 0.005797 -0.008235 0.885338 -0.008480 0.883811 dit_step10_xt 0.999878 0.086120 0.009653 -0.014926 0.801678 -0.015075 0.802324 dit_step15_xt 0.999676 0.141800 0.014205 -0.021602 0.731346 -0.021933 0.731304 dit_step20_xt 0.999210 0.229679 0.020421 -0.028340 0.678623 -0.028882 0.679066 dit_step25_xt 0.998341 0.384278 0.028102 -0.035174 0.647830 -0.035643 0.650780 dit_step30_xt 0.996865 0.602514 0.038019 -0.042047 0.642320 -0.042577 0.647107 dit_step35_xt 0.994910 0.849962 0.049697 -0.048897 0.662980 -0.049471 0.669256 dit_step40_xt 0.992871 1.097433 0.062726 -0.055731 0.707499 -0.056532 0.715783 dit_step45_xt 0.991383 1.320407 0.075127 -0.062614 0.771567 -0.063533 0.780575 [Xl-turbo] steps=8, shift=3.0 | acestep-v15-xl-turbo-Q5_K_M.gguf [GGML] Running acestep-v15-xl-turbo-Q5_K_M.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 10:18:43.818 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:18:43.818 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:18:43.819 | 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 10:18:43.821 | INFO | acestep.core.generation.handler.generate_music:_vram_preflight_check:156 - [generate_music] VRAM pre-flight: 76.75 GB free, ~1.23 GB needed (batch=1, duration=88s, mode=xl_turbo). 2026-07-12 10:18:43.826 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:18:43.831 | 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 10:18:43.834 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:18:43.834 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:18:43.834 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:18:43.834 | 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 10:18:43.834 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:18:43.834 | 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 10:18:43.834 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:18:43.837 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:18:43.845 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:18:43.845 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:18:43.851 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 10:18:43.858 | 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 10:18:44.412 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 10:18:44.413 | 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 10:18:44.413 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.006398439407348633, 'diffusion_time_cost': 0.5539302825927734, 'diffusion_per_step_time_cost': 0.06924128532409668, 'total_time_cost': 0.5603287220001221, 'offload_time_cost': 0.0} 2026-07-12 10:18:44.436 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 10:18:44.437 | DEBUG | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:136 - [generate_music] Before VAE decode: allocated=17.03GB, max=18.34GB 2026-07-12 10:18:44.437 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:154 - [generate_music] Effective free VRAM before VAE decode: 76.66 GB 2026-07-12 10:18:44.437 | 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 10:18:44.438 | DEBUG | acestep.core.generation.handler.memory_utils:_get_auto_decode_chunk_size:108 - [_get_auto_decode_chunk_size] Effective free VRAM: 76.66 GB 2026-07-12 10:18:44.438 | DEBUG | acestep.core.generation.handler.memory_utils:_should_offload_wav_to_cpu:131 - [_should_offload_wav_to_cpu] Effective free VRAM: 76.66 GB 2026-07-12 10:18:44.438 | 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 10:18:44.711 | DEBUG | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:193 - [generate_music] After VAE decode: allocated=17.21GB, max=18.49GB 2026-07-12 10:18:44.911 | 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 10:18:44.914 | 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-Q5_K_M.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-Q5_K_M.gguf: 830 tensors, data at offset 69056 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q5_K_M.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: 333 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-Q5_K_M.gguf: 830 tensors, data at offset 69056 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 141 tensors, 416.2 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L, CLS), text_proj, null_cond [Store] Load CondEnc: 718 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, 8.4 ms [Debug] enc_hidden: [238, 2048] first4: 0.687591 -0.019866 -0.051357 0.022943 [Store] Unload CondEnc (416.2 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-Q5_K_M.gguf: 830 tensors, data at offset 69056 [WeightCtx] Loaded 30 tensors, 73.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 55 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: -0.126205 1.455492 0.291894 -0.650703 [Store] Unload FSQ-Detok (73.2 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: -0.126205 1.455492 0.291894 -0.650703 [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-Q5_K_M.gguf: 830 tensors, data at offset 69056 [DiT] Self-attn: Q+K fused, V separate [DiT] Cross-attn: all separate [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 630 tensors, 2801.1 MB into backend [Load] DiT: 32 layers, H=2560, Nh=32/8, D=128 [Store] Load DiT: 2003 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2729 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [15360] first4: 0.005156 0.089698 0.077651 -0.035042 [Debug] temb: [2560] first4: 0.009955 -0.008020 -0.006084 0.004950 [Debug] temb_t: [2560] first4: 0.008419 -0.008148 -0.006723 0.003948 [Debug] temb_r: [2560] first4: 0.001536 0.000128 0.000639 0.001002 [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: 0.028513 -0.079992 0.086633 -0.070856 [Debug] temb_lin1_r: [2560] first4: 0.006506 -0.006323 -0.000918 -0.013460 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.337036 -0.495361 0.120361 -0.198975 [Debug] proj_in_input: [192, 2170] first4: -0.126205 1.455492 0.291894 -0.650703 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.052719 0.649559 0.105347 0.866692 [Debug] layer0_sa_input: [2560, 1085] first4: 0.029791 -0.035540 0.103631 -0.096009 [Debug] layer0_q_after_rope: [128, 32] first4: -0.867970 -2.225284 -2.790891 0.572552 [Debug] layer0_k_after_rope: [128, 8] first4: 2.195312 4.488281 -1.233398 8.062500 [Debug] layer0_sa_output: [2560, 1085] first4: 2.195312 4.488281 -1.233398 8.062500 [Debug] layer0_attn_out: [4096, 1085] first4: -0.867970 -2.225284 -2.790891 0.572552 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.474917 -2.224739 -0.792218 -4.254282 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.285342 -1.939095 -0.732342 -4.281595 [Debug] hidden_after_layer0: [2560, 1085] first4: -1.411742 -1.287817 245.778427 -4.117364 [Debug] hidden_after_layer6: [2560, 1085] first4: -7.633467 3.076976 303.753357 -20.899485 [Debug] hidden_after_layer12: [2560, 1085] first4: -21.389183 38.040333 219.691864 17.137894 [Debug] hidden_after_layer18: [2560, 1085] first4: 11.735584 -15.480383 227.798523 19.725605 [Debug] hidden_after_layer31: [2560, 1085] first4: 67.860855 -114.579346 -0.892647 87.634003 [Debug] dit_step0_vt: [2170, 64] first4: -0.340698 2.263306 0.039551 1.717499 [Debug] dit_step0_xt: [2170, 64] first4: 0.201150 2.110984 -0.172666 0.813306 [DiT] Step 1/50 t=1.000 [Debug] dit_step1_vt: [2170, 64] first4: -0.366089 2.194214 0.102295 1.563385 [Debug] dit_step1_xt: [2170, 64] first4: 0.208472 2.067100 -0.174712 0.782039 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.350586 2.198120 0.111328 1.552673 [Debug] dit_step2_xt: [2170, 64] first4: 0.215483 2.023137 -0.176938 0.750985 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.298218 2.099426 0.128784 1.541931 [Debug] dit_step3_xt: [2170, 64] first4: 0.221448 1.981149 -0.179514 0.720147 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.306641 2.057617 0.096069 1.573410 [Debug] dit_step4_xt: [2170, 64] first4: 0.227581 1.939996 -0.181436 0.688678 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.296143 2.036438 0.109253 1.574615 [Debug] dit_step5_xt: [2170, 64] first4: 0.233503 1.899267 -0.183621 0.657186 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.294922 2.007812 0.128296 1.597961 [Debug] dit_step6_xt: [2170, 64] first4: 0.239402 1.859111 -0.186187 0.625227 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.295654 1.968903 0.163940 1.599792 [Debug] dit_step7_xt: [2170, 64] first4: 0.245315 1.819733 -0.189465 0.593231 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.312500 1.924988 0.185303 1.600510 [Debug] dit_step8_xt: [2170, 64] first4: 0.251565 1.781233 -0.193171 0.561221 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.340332 1.867340 0.162231 1.582851 [Debug] dit_step9_xt: [2170, 64] first4: 0.258372 1.743887 -0.196416 0.529564 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.349609 1.825562 0.143677 1.575897 [Debug] dit_step10_xt: [2170, 64] first4: 0.265364 1.707375 -0.199290 0.498046 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.346924 1.784607 0.132202 1.562317 [Debug] dit_step11_xt: [2170, 64] first4: 0.272302 1.671683 -0.201934 0.466799 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.346558 1.728760 0.117188 1.544540 [Debug] dit_step12_xt: [2170, 64] first4: 0.279233 1.637108 -0.204277 0.435909 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.330322 1.683357 0.098877 1.529327 [Debug] dit_step13_xt: [2170, 64] first4: 0.285840 1.603441 -0.206255 0.405322 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.324097 1.633286 0.077759 1.523193 [Debug] dit_step14_xt: [2170, 64] first4: 0.292322 1.570775 -0.207810 0.374858 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.299316 1.578735 0.050171 1.507721 [Debug] dit_step15_xt: [2170, 64] first4: 0.298308 1.539201 -0.208813 0.344704 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.268555 1.529358 0.030884 1.497559 [Debug] dit_step16_xt: [2170, 64] first4: 0.303679 1.508613 -0.209431 0.314753 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.235229 1.476715 0.006470 1.494385 [Debug] dit_step17_xt: [2170, 64] first4: 0.308384 1.479079 -0.209561 0.284865 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.195679 1.425964 -0.000366 1.477417 [Debug] dit_step18_xt: [2170, 64] first4: 0.312297 1.450560 -0.209553 0.255317 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.164124 1.371460 -0.013672 1.482056 [Debug] dit_step19_xt: [2170, 64] first4: 0.315580 1.423131 -0.209280 0.225675 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.127563 1.333252 -0.020508 1.473206 [Debug] dit_step20_xt: [2170, 64] first4: 0.318131 1.396466 -0.208870 0.196211 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.100464 1.283936 -0.027954 1.456329 [Debug] dit_step21_xt: [2170, 64] first4: 0.320140 1.370787 -0.208311 0.167085 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.071289 1.233643 -0.040894 1.432709 [Debug] dit_step22_xt: [2170, 64] first4: 0.321566 1.346114 -0.207493 0.138431 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.042847 1.186768 -0.043945 1.415863 [Debug] dit_step23_xt: [2170, 64] first4: 0.322423 1.322379 -0.206614 0.110113 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.025818 1.138550 -0.053223 1.396820 [Debug] dit_step24_xt: [2170, 64] first4: 0.322939 1.299608 -0.205549 0.082177 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.011322 1.080627 -0.060791 1.377777 [Debug] dit_step25_xt: [2170, 64] first4: 0.323166 1.277995 -0.204334 0.054621 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.011169 1.019989 -0.072510 1.353058 [Debug] dit_step26_xt: [2170, 64] first4: 0.323389 1.257595 -0.202883 0.027560 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.012268 0.948425 -0.080627 1.326782 [Debug] dit_step27_xt: [2170, 64] first4: 0.323635 1.238627 -0.201271 0.001025 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.013550 0.906769 -0.086304 1.303101 [Debug] dit_step28_xt: [2170, 64] first4: 0.323906 1.220492 -0.199545 -0.025037 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.016174 0.829041 -0.098145 1.286682 [Debug] dit_step29_xt: [2170, 64] first4: 0.324229 1.203911 -0.197582 -0.050771 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.030396 0.754982 -0.116333 1.276733 [Debug] dit_step30_xt: [2170, 64] first4: 0.324837 1.188811 -0.195255 -0.076306 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.049072 0.685066 -0.098022 1.246552 [Debug] dit_step31_xt: [2170, 64] first4: 0.325818 1.175110 -0.193295 -0.101237 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.041931 0.606918 -0.109436 1.214722 [Debug] dit_step32_xt: [2170, 64] first4: 0.326657 1.162971 -0.191106 -0.125531 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.041870 0.522072 -0.111786 1.186951 [Debug] dit_step33_xt: [2170, 64] first4: 0.327495 1.152530 -0.188870 -0.149270 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.060272 0.430695 -0.120544 1.155945 [Debug] dit_step34_xt: [2170, 64] first4: 0.328700 1.143916 -0.186459 -0.172389 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.081757 0.332596 -0.126602 1.126312 [Debug] dit_step35_xt: [2170, 64] first4: 0.330335 1.137264 -0.183927 -0.194915 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.103500 0.229736 -0.112553 1.087860 [Debug] dit_step36_xt: [2170, 64] first4: 0.332405 1.132670 -0.181676 -0.216673 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.124882 0.123627 -0.126328 1.062744 [Debug] dit_step37_xt: [2170, 64] first4: 0.334903 1.130197 -0.179150 -0.237927 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.150455 0.019623 -0.101562 1.013489 [Debug] dit_step38_xt: [2170, 64] first4: 0.337912 1.129805 -0.177118 -0.258197 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.164871 -0.089508 -0.085571 0.982849 [Debug] dit_step39_xt: [2170, 64] first4: 0.341209 1.131595 -0.175407 -0.277854 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.199799 -0.199615 -0.079025 0.932983 [Debug] dit_step40_xt: [2170, 64] first4: 0.345205 1.135587 -0.173827 -0.296514 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.219299 -0.286095 -0.061401 0.891724 [Debug] dit_step41_xt: [2170, 64] first4: 0.349591 1.141309 -0.172599 -0.314348 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.238281 -0.386032 -0.021469 0.838318 [Debug] dit_step42_xt: [2170, 64] first4: 0.354357 1.149030 -0.172169 -0.331115 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.276886 -0.479294 0.001894 0.783142 [Debug] dit_step43_xt: [2170, 64] first4: 0.359895 1.158616 -0.172207 -0.346778 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.299683 -0.570953 0.057632 0.718140 [Debug] dit_step44_xt: [2170, 64] first4: 0.365888 1.170035 -0.173360 -0.361140 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.327682 -0.646957 0.094086 0.655945 [Debug] dit_step45_xt: [2170, 64] first4: 0.372442 1.182974 -0.175241 -0.374259 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.360474 -0.718628 0.131775 0.620850 [Debug] dit_step46_xt: [2170, 64] first4: 0.379651 1.197347 -0.177877 -0.386676 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.408968 -0.800018 0.187805 0.583557 [Debug] dit_step47_xt: [2170, 64] first4: 0.387831 1.213347 -0.181633 -0.398347 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.448593 -0.856491 0.251770 0.563599 [Debug] dit_step48_xt: [2170, 64] first4: 0.396803 1.230477 -0.186668 -0.409619 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.446228 -0.842178 0.176453 0.585327 [Debug] dit_x0: [2170, 64] first4: 0.405727 1.247320 -0.190197 -0.421326 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 3892.1 ms (3892.1 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.405727 1.247320 -0.190197 -0.421326 [Store] Unload DiT (2801.1 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: Vulkan0 (CPU threads: 16) [VAE] Backend: Vulkan0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 668 ms [VAE] Tiled decode: 3 tiles (chunk=1024, overlap=64, stride=896) [VAE] Graph: 479 nodes, T_latent=960 [VAE] Upsample factor: 1920.00 (expected ~1920) [VAE] Graph: 479 nodes, T_latent=1024 [VAE] Graph: 479 nodes, T_latent=442 [VAE] Tiled decode done: 3 tiles -> T_audio=4166400 (86.80s @ 48kHz) [VAE-Decode Batch0] Decode: 788.8 ms [Debug] vae_audio: [2, 4166400] first4: -0.001532 -0.000818 -0.000920 -0.000402 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 403 ms (215.13x realtime), 32 threads [MP3] Wrote ggml-xl-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 10:18:55.124 | 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 0.998895 detok_output 0.999877 context 0.999921 noise 1.000000 temb_t 0.999907 hidden_after_proj_in 0.999940 enc_after_cond_emb 0.998871 layer0_sa_output 0.988768 hidden_after_layer0 0.999253 hidden_after_layer6 0.999612 hidden_after_layer12 0.994731 hidden_after_layer18 0.995475 hidden_after_layer31 0.987338 dit_step0_vt 0.908273 dit_step0_xt 0.999835 dit_step1_vt 0.921604 dit_step1_xt 0.999486 dit_step2_vt 0.918673 dit_step2_xt 0.998626 dit_step3_vt 0.922855 dit_step3_xt 0.996766 dit_step4_vt 0.918148 dit_step4_xt 0.992262 dit_step5_vt 0.911560 dit_step5_xt 0.981613 dit_step6_vt 0.902635 dit_step6_xt 0.960373 dit_step7_vt 0.880709 dit_x0 0.925475 vae_audio 0.745040 vae_audio (STFT cosine) 0.924731 [Xl-turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999835 0.233049 0.012942 -0.002876 0.978384 -0.002721 0.976534 dit_step1_xt 0.999486 0.275895 0.022581 -0.005755 0.952721 -0.005567 0.949004 dit_step2_xt 0.998626 0.459889 0.035632 -0.009350 0.924358 -0.009330 0.918931 dit_step3_xt 0.996766 0.778180 0.052084 -0.014882 0.893903 -0.014418 0.887241 dit_step4_xt 0.992262 1.371272 0.077106 -0.022565 0.865832 -0.021453 0.857796 dit_step5_xt 0.981613 2.038996 0.115692 -0.033932 0.849562 -0.032240 0.840033 dit_step6_xt 0.960373 2.861604 0.175036 -0.050712 0.874252 -0.049194 0.863964 [Xl-sft] steps=50, shift=1.0 | acestep-v15-xl-sft-Q5_K_M.gguf [GGML] Running acestep-v15-xl-sft-Q5_K_M.gguf... [GGML] Done, 131 dump files [Python] Initializing acestep-v15-xl-sft... Loading checkpoint shards: 0%| | 0/4 [00:00 2026-07-12 10:19:00.225 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:19:00.225 | 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 10:19:00.225 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:19:00.229 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:19:00.237 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:19:00.237 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:19:00.243 | 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