[Request] Parsed ggml-turbo/request0.json [Request] seed=42 lm_batch=1 synth_batch=1 [Request] caption: An upbeat and anthemic pop-rock track driven by bright, slig... [Request] lyrics: 490 bytes [Request] bpm=83 dur=88 key=G major ts=4 lang=fr [Request] lm: temp=0.85 cfg=2.0 top_p=0.90 top_k=0 [Request] dit: steps=8 guidance=0.0 shift=3.0 [Request] task_type: text2music [Request] solver: euler (stork_substeps=10) [Request] lm_mode: generate [Request] output_format: mp3 [Request] synth_model: acestep-v15-turbo-Q6_K.gguf [Request] vae: vae-BF16.gguf [Request] audio_codes: (present) [Ace-Synth] Batch: 1 request(s) [Registry] Qwen3-Embedding-0.6B-BF16.gguf -> Text-Enc [Registry] Qwen3-Embedding-0.6B-Q8_0.gguf -> Text-Enc [Registry] acestep-5Hz-lm-0.6B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-0.6B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-4B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q5_K_M.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q6_K.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q8_0.gguf -> LM [Registry] acestep-v15-base-BF16.gguf -> DiT [Registry] acestep-v15-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-base-Q6_K.gguf -> DiT [Registry] acestep-v15-base-Q8_0.gguf -> DiT [Registry] acestep-v15-sft-BF16.gguf -> DiT [Registry] acestep-v15-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-BF16.gguf -> DiT [Registry] acestep-v15-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-continuous-BF16.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift1-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift3-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-base-BF16.gguf -> DiT [Registry] acestep-v15-xl-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-base-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sft-BF16.gguf -> DiT [Registry] acestep-v15-xl-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-turbo-BF16.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q8_0.gguf -> DiT [Registry] scragvae-BF16.gguf -> VAE [Registry] vae-BF16.gguf -> VAE [Store] Created (policy=STRICT) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q6_K.gguf: 678 tensors, data at offset 56864 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q6_K.gguf: 678 tensors, data at offset 56864 [Synth-Load] Ready: turbo=yes, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=8, guidance=1.0, shift=3.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: CUDA0 (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: 269 ms [Debug] text_hidden: [70, 1024] first4: 3.652379 1.048645 0.229843 -13.063987 [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: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q6_K.gguf: 678 tensors, data at offset 56864 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 140 tensors, 476.3 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L), text_proj, null_cond [Store] Load CondEnc: 321 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, 17.5 ms [Debug] enc_hidden: [238, 2048] first4: 1.760692 -0.050084 -0.132974 0.057858 [Store] Unload CondEnc (476.3 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] Detokenizer backend: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q6_K.gguf: 678 tensors, data at offset 56864 [WeightCtx] Loaded 30 tensors, 82.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 56 ms CUDA Graph id 8 reused ggml_backend_cuda_graph_compute: CUDA graph warmup complete CUDA Graph id 10 reused ggml_backend_cuda_graph_compute: CUDA graph warmup complete CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 80.9 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.138872 1.452165 0.311136 -0.633141 [Store] Unload FSQ-Detok (82.2 MB) [Init-Noise Batch0] Philox noise seed=42, [2170, 64] solver=euler [Debug] noise: [2170, 64] first4: 0.194336 2.156250 -0.171875 0.847656 [Debug] context: [2170, 128] first4: -0.138872 1.452165 0.311136 -0.633141 [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: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q6_K.gguf: 678 tensors, data at offset 56864 [DiT] Self-attn: Q+K+V fused [DiT] Cross-attn: Q+K+V fused [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 478 tensors, 1237.2 MB into backend [Load] DiT: 24 layers, H=2048, Nh=16/8, D=128 [Store] Load DiT: 776 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2129 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [12288] first4: 0.259936 -0.161027 -0.098424 0.051532 [Debug] temb: [2048] first4: 0.000362 -0.132329 -0.035400 0.064685 [Debug] temb_t: [2048] first4: 0.001493 0.026964 -0.052786 0.063738 [Debug] temb_r: [2048] first4: -0.001131 -0.159293 0.017385 0.000947 [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: -0.049350 -0.051345 -0.017496 -0.036550 [Debug] temb_lin1_r: [2048] first4: -0.014407 -0.020607 -0.015728 0.003874 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.035786 -0.947470 0.541442 0.448331 [Debug] proj_in_input: [192, 2170] first4: -0.138872 1.452165 0.311136 -0.633141 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.172973 0.808179 0.316906 -0.565211 [Debug] layer0_sa_input: [2048, 1085] first4: -0.715028 -0.750239 -0.048129 0.261506 [Debug] layer0_q_after_rope: [128, 16] first4: -1.538383 -1.034582 0.180243 0.458721 [Debug] layer0_k_after_rope: [128, 8] first4: -0.172973 0.808179 0.316906 -0.565211 [Debug] layer0_sa_output: [2048, 1085] first4: -1.505207 0.196705 -0.369789 0.517696 [Debug] layer0_attn_out: [2048, 1085] first4: -1.538383 -1.034582 0.180243 0.458721 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.538383 -1.034582 0.180243 0.458721 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.597656 -0.814493 -0.319301 0.494606 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.117565 0.547766 51.825462 -0.820052 [Debug] hidden_after_layer6: [2048, 1085] first4: -21.199974 -0.005882 33.427658 -4.062068 [Debug] hidden_after_layer12: [2048, 1085] first4: -15.117359 -16.393244 75.374779 29.863274 [Debug] hidden_after_layer18: [2048, 1085] first4: -27.528477 12.100023 64.311127 19.518600 [Debug] hidden_after_layer23: [2048, 1085] first4: -16.013800 47.237381 197.732361 135.147873 [Debug] dit_step0_vt: [2170, 64] first4: 0.094602 1.137793 0.345273 2.376239 [Debug] dit_step0_xt: [2170, 64] first4: 0.190036 2.104532 -0.187569 0.739645 [DiT] Step 1/8 t=1.000 ggml_backend_cuda_graph_compute: CUDA graph warmup complete [Debug] dit_step1_vt: [2170, 64] first4: -0.144124 1.309091 -0.176263 1.953844 [Debug] dit_step1_xt: [2170, 64] first4: 0.197897 2.033127 -0.177955 0.633072 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: 0.045820 1.244032 0.113861 2.373163 [Debug] dit_step2_xt: [2170, 64] first4: 0.194843 1.950192 -0.185546 0.474861 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: 0.299730 1.099803 0.261519 2.643507 [Debug] dit_step3_xt: [2170, 64] first4: 0.169865 1.858541 -0.207339 0.254569 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: 0.328432 1.033373 0.084045 2.729582 [Debug] dit_step4_xt: [2170, 64] first4: 0.134676 1.747823 -0.216344 -0.037886 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: 0.310009 0.900694 -0.309269 2.738043 [Debug] dit_step5_xt: [2170, 64] first4: 0.090389 1.619152 -0.172162 -0.429035 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: 0.260761 0.626324 -0.805546 2.744401 [Debug] dit_step6_xt: [2170, 64] first4: 0.038237 1.493888 -0.011053 -0.977915 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: 0.140841 0.117065 -1.339300 2.990012 [Debug] dit_x0: [2170, 64] first4: -0.004015 1.458768 0.390737 -1.874919 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 316.1 ms (316.1 ms/sample) [Debug] dit_output: [2170, 64] first4: -0.004015 1.458768 0.390737 -1.874919 [Store] Unload DiT (1237.2 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: CUDA0 (CPU threads: 16) [VAE] Backend: CUDA0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 671 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: 616.2 ms [Debug] vae_audio: [2, 4166400] first4: 0.000691 0.001293 0.001082 0.001582 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 457 ms (190.02x realtime), 32 threads [MP3] Wrote ggml-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 10:13:18.251 | 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 10:13:20.308 | WARNING | acestep.training.trainer::40 - bitsandbytes not installed. Using standard AdamW. 2026-07-12 10:13:20.313 | 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 10:13:20.382 | 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 10:13:21.990 | 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 10:13:21.990 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:13:21.990 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:13:21.990 | 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 10:13:21.993 | 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 10:13:21.999 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:13:22.195 | 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 10:13:22.197 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:13:22.197 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:13:22.197 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:13:22.197 | 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 10:13:22.197 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:13:22.197 | 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 10:13:22.197 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:13:22.204 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:13:22.218 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:13:22.219 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:13:22.254 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 10:13:22.262 | 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 10:13:22.346 | 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 10:13:22.572 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 10:13:22.573 | 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 10:13:22.573 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.006876230239868164, 'diffusion_time_cost': 0.3104565143585205, 'diffusion_per_step_time_cost': 0.038807064294815063, 'total_time_cost': 0.31733274459838867, 'offload_time_cost': 0.0} 2026-07-12 10:13:22.587 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 10:13:22.589 | 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 10:13:22.589 | 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 10:13:22.589 | 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 10:13:22.589 | 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 10:13:22.589 | 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 10:13:22.589 | 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 10:13:22.866 | 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 10:13:23.024 | 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 10:13:23.027 | INFO | acestep.core.generation.handler.generate_music_payload:_build_generate_music_success_payload:50 - [generate_music] Done! Generated 1 audio tensors. [Request] Parsed ggml-sft/request0.json [Request] seed=42 lm_batch=1 synth_batch=1 [Request] caption: An upbeat and anthemic pop-rock track driven by bright, slig... [Request] lyrics: 490 bytes [Request] bpm=83 dur=88 key=G major ts=4 lang=fr [Request] lm: temp=0.85 cfg=2.0 top_p=0.90 top_k=0 [Request] dit: steps=50 guidance=1.0 shift=1.0 [Request] task_type: text2music [Request] solver: euler (stork_substeps=10) [Request] lm_mode: generate [Request] output_format: mp3 [Request] synth_model: acestep-v15-sft-Q6_K.gguf [Request] vae: vae-BF16.gguf [Request] audio_codes: (present) [Ace-Synth] Batch: 1 request(s) [Registry] Qwen3-Embedding-0.6B-BF16.gguf -> Text-Enc [Registry] Qwen3-Embedding-0.6B-Q8_0.gguf -> Text-Enc [Registry] acestep-5Hz-lm-0.6B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-0.6B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-4B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q5_K_M.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q6_K.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q8_0.gguf -> LM [Registry] acestep-v15-base-BF16.gguf -> DiT [Registry] acestep-v15-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-base-Q6_K.gguf -> DiT [Registry] acestep-v15-base-Q8_0.gguf -> DiT [Registry] acestep-v15-sft-BF16.gguf -> DiT [Registry] acestep-v15-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-BF16.gguf -> DiT [Registry] acestep-v15-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-continuous-BF16.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift1-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift3-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-base-BF16.gguf -> DiT [Registry] acestep-v15-xl-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-base-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sft-BF16.gguf -> DiT [Registry] acestep-v15-xl-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-turbo-BF16.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q8_0.gguf -> DiT [Registry] scragvae-BF16.gguf -> VAE [Registry] vae-BF16.gguf -> VAE [Store] Created (policy=STRICT) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q6_K.gguf: 678 tensors, data at offset 56800 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q6_K.gguf: 678 tensors, data at offset 56800 [Synth-Load] Ready: turbo=no, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=50, guidance=1.0, shift=1.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: CUDA0 (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: 264 ms [Debug] text_hidden: [70, 1024] first4: 3.652379 1.048645 0.229843 -13.063987 [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: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q6_K.gguf: 678 tensors, data at offset 56800 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 140 tensors, 476.3 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L), text_proj, null_cond [Store] Load CondEnc: 304 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, 16.1 ms [Debug] enc_hidden: [238, 2048] first4: 1.760692 -0.050084 -0.132974 0.057858 [Store] Unload CondEnc (476.3 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] Detokenizer backend: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q6_K.gguf: 678 tensors, data at offset 56800 [WeightCtx] Loaded 30 tensors, 82.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 54 ms CUDA Graph id 8 reused ggml_backend_cuda_graph_compute: CUDA graph warmup complete CUDA Graph id 10 reused ggml_backend_cuda_graph_compute: CUDA graph warmup complete CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 76.6 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.138872 1.452165 0.311136 -0.633141 [Store] Unload FSQ-Detok (82.2 MB) [Init-Noise Batch0] Philox noise seed=42, [2170, 64] solver=euler [Debug] noise: [2170, 64] first4: 0.194336 2.156250 -0.171875 0.847656 [Debug] context: [2170, 128] first4: -0.138872 1.452165 0.311136 -0.633141 [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: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q6_K.gguf: 678 tensors, data at offset 56800 [DiT] Self-attn: Q+K+V fused [DiT] Cross-attn: Q+K+V fused [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 478 tensors, 1237.2 MB into backend [Load] DiT: 24 layers, H=2048, Nh=16/8, D=128 [Store] Load DiT: 757 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2129 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [12288] first4: 0.154808 -0.117030 -0.090352 0.080481 [Debug] temb: [2048] first4: -0.002344 -0.176478 0.004328 -0.002122 [Debug] temb_t: [2048] first4: -0.000923 0.003440 -0.013277 -0.002148 [Debug] temb_r: [2048] first4: -0.001421 -0.179918 0.017604 0.000026 [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: -0.041380 0.030081 0.028122 -0.024844 [Debug] temb_lin1_r: [2048] first4: 0.004272 0.006720 0.000208 -0.005103 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.086624 -0.904253 0.503391 0.442051 [Debug] proj_in_input: [192, 2170] first4: -0.138872 1.452165 0.311136 -0.633141 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.199570 0.874754 0.316708 -0.571101 [Debug] layer0_sa_input: [2048, 1085] first4: -0.910745 -0.704020 -0.040318 0.293133 [Debug] layer0_q_after_rope: [128, 16] first4: -1.802282 -1.351642 -0.156203 0.382316 [Debug] layer0_k_after_rope: [128, 8] first4: -0.199570 0.874754 0.316708 -0.571101 [Debug] layer0_sa_output: [2048, 1085] first4: -1.653265 0.791985 -0.584073 0.532764 [Debug] layer0_attn_out: [2048, 1085] first4: -1.802282 -1.351642 -0.156203 0.382316 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.802282 -1.351642 -0.156203 0.382316 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.969991 -1.020857 -0.438228 0.389828 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.361176 1.290457 57.309540 -1.704920 [Debug] hidden_after_layer6: [2048, 1085] first4: -17.171047 2.590428 58.682625 -1.280477 [Debug] hidden_after_layer12: [2048, 1085] first4: -10.460129 6.957541 -24.641716 -3.044174 [Debug] hidden_after_layer18: [2048, 1085] first4: -5.832457 20.281548 -42.479450 -14.793241 [Debug] hidden_after_layer23: [2048, 1085] first4: 35.322731 62.939636 51.963326 -12.880016 [Debug] dit_step0_vt: [2170, 64] first4: -0.627825 2.542034 -0.213104 1.444499 [Debug] dit_step0_xt: [2170, 64] first4: 0.206892 2.105409 -0.167613 0.818766 [DiT] Step 1/50 t=1.000 ggml_backend_cuda_graph_compute: CUDA graph warmup complete [Debug] dit_step1_vt: [2170, 64] first4: -0.690862 2.516313 -0.058985 1.274892 [Debug] dit_step1_xt: [2170, 64] first4: 0.220710 2.055083 -0.166433 0.793268 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.655266 2.463712 -0.012028 1.314226 [Debug] dit_step2_xt: [2170, 64] first4: 0.233815 2.005809 -0.166193 0.766984 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.659177 2.428759 -0.003529 1.312791 [Debug] dit_step3_xt: [2170, 64] first4: 0.246999 1.957234 -0.166122 0.740728 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.615125 2.357382 -0.009183 1.303123 [Debug] dit_step4_xt: [2170, 64] first4: 0.259301 1.910086 -0.165938 0.714666 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.567008 2.270495 -0.036029 1.294512 [Debug] dit_step5_xt: [2170, 64] first4: 0.270641 1.864676 -0.165218 0.688775 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.540169 2.242677 -0.064433 1.283841 [Debug] dit_step6_xt: [2170, 64] first4: 0.281445 1.819823 -0.163929 0.663099 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.522388 2.171490 -0.070519 1.274468 [Debug] dit_step7_xt: [2170, 64] first4: 0.291892 1.776393 -0.162519 0.637609 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.479816 2.093430 -0.098517 1.257552 [Debug] dit_step8_xt: [2170, 64] first4: 0.301489 1.734524 -0.160548 0.612458 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.415475 1.999766 -0.097237 1.248299 [Debug] dit_step9_xt: [2170, 64] first4: 0.309798 1.694529 -0.158604 0.587492 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.353542 1.925806 -0.072687 1.253501 [Debug] dit_step10_xt: [2170, 64] first4: 0.316869 1.656013 -0.157150 0.562422 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.319029 1.833341 -0.105754 1.226157 [Debug] dit_step11_xt: [2170, 64] first4: 0.323250 1.619346 -0.155035 0.537899 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.273029 1.758821 -0.096431 1.214706 [Debug] dit_step12_xt: [2170, 64] first4: 0.328710 1.584170 -0.153106 0.513605 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.266960 1.683219 -0.102588 1.177485 [Debug] dit_step13_xt: [2170, 64] first4: 0.334049 1.550505 -0.151055 0.490055 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.246134 1.612492 -0.158022 1.172917 [Debug] dit_step14_xt: [2170, 64] first4: 0.338972 1.518255 -0.147894 0.466597 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.246173 1.510010 -0.195429 1.161920 [Debug] dit_step15_xt: [2170, 64] first4: 0.343895 1.488055 -0.143985 0.443358 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.252102 1.458717 -0.217965 1.156380 [Debug] dit_step16_xt: [2170, 64] first4: 0.348938 1.458881 -0.139626 0.420231 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.254568 1.368449 -0.232921 1.135936 [Debug] dit_step17_xt: [2170, 64] first4: 0.354029 1.431512 -0.134968 0.397512 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.276478 1.311423 -0.268885 1.097750 [Debug] dit_step18_xt: [2170, 64] first4: 0.359558 1.405283 -0.129590 0.375557 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.294669 1.220377 -0.289366 1.086673 [Debug] dit_step19_xt: [2170, 64] first4: 0.365452 1.380876 -0.123803 0.353824 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.307631 1.122684 -0.305688 1.039096 [Debug] dit_step20_xt: [2170, 64] first4: 0.371604 1.358422 -0.117689 0.333042 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.320067 1.027049 -0.304527 1.019557 [Debug] dit_step21_xt: [2170, 64] first4: 0.378006 1.337881 -0.111598 0.312651 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.350087 0.918397 -0.329622 0.969550 [Debug] dit_step22_xt: [2170, 64] first4: 0.385008 1.319513 -0.105006 0.293260 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.365085 0.829263 -0.336877 0.973421 [Debug] dit_step23_xt: [2170, 64] first4: 0.392309 1.302928 -0.098268 0.273791 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.395747 0.714851 -0.351823 0.920987 [Debug] dit_step24_xt: [2170, 64] first4: 0.400224 1.288631 -0.091232 0.255371 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.407578 0.627121 -0.325457 0.930331 [Debug] dit_step25_xt: [2170, 64] first4: 0.408376 1.276089 -0.084723 0.236765 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.427626 0.480583 -0.322349 0.882701 [Debug] dit_step26_xt: [2170, 64] first4: 0.416928 1.266477 -0.078276 0.219111 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.453244 0.378268 -0.310436 0.878720 [Debug] dit_step27_xt: [2170, 64] first4: 0.425993 1.258911 -0.072067 0.201536 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.484197 0.286652 -0.310500 0.846769 [Debug] dit_step28_xt: [2170, 64] first4: 0.435677 1.253178 -0.065857 0.184601 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.484182 0.147622 -0.320656 0.844611 [Debug] dit_step29_xt: [2170, 64] first4: 0.445361 1.250226 -0.059444 0.167709 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.508160 0.050777 -0.287975 0.822147 [Debug] dit_step30_xt: [2170, 64] first4: 0.455524 1.249210 -0.053685 0.151266 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.526775 -0.067995 -0.286453 0.815522 [Debug] dit_step31_xt: [2170, 64] first4: 0.466059 1.250570 -0.047955 0.134955 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.536907 -0.164483 -0.254206 0.791782 [Debug] dit_step32_xt: [2170, 64] first4: 0.476798 1.253860 -0.042871 0.119120 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.554756 -0.304226 -0.230142 0.777507 [Debug] dit_step33_xt: [2170, 64] first4: 0.487893 1.259945 -0.038269 0.103570 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.573295 -0.425320 -0.236831 0.755118 [Debug] dit_step34_xt: [2170, 64] first4: 0.499359 1.268451 -0.033532 0.088467 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.573044 -0.497246 -0.197240 0.744287 [Debug] dit_step35_xt: [2170, 64] first4: 0.510819 1.278396 -0.029587 0.073582 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.585109 -0.592640 -0.168198 0.726437 [Debug] dit_step36_xt: [2170, 64] first4: 0.522522 1.290249 -0.026223 0.059053 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.587709 -0.676584 -0.197042 0.700205 [Debug] dit_step37_xt: [2170, 64] first4: 0.534276 1.303780 -0.022282 0.045049 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.554357 -0.768880 -0.150155 0.656067 [Debug] dit_step38_xt: [2170, 64] first4: 0.545363 1.319158 -0.019279 0.031927 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.579639 -0.834131 -0.095934 0.619093 [Debug] dit_step39_xt: [2170, 64] first4: 0.556956 1.335841 -0.017361 0.019546 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.559745 -0.917781 -0.068388 0.570466 [Debug] dit_step40_xt: [2170, 64] first4: 0.568151 1.354196 -0.015993 0.008136 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.545133 -0.978838 -0.005740 0.529431 [Debug] dit_step41_xt: [2170, 64] first4: 0.579053 1.373773 -0.015878 -0.002452 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.571913 -1.037350 0.026274 0.511354 [Debug] dit_step42_xt: [2170, 64] first4: 0.590492 1.394520 -0.016403 -0.012679 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.536747 -1.085640 0.067391 0.442781 [Debug] dit_step43_xt: [2170, 64] first4: 0.601226 1.416233 -0.017751 -0.021535 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.516402 -1.133917 0.094046 0.397485 [Debug] dit_step44_xt: [2170, 64] first4: 0.611554 1.438911 -0.019632 -0.029485 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.491958 -1.175158 0.141107 0.345619 [Debug] dit_step45_xt: [2170, 64] first4: 0.621394 1.462415 -0.022454 -0.036397 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.483012 -1.242182 0.198092 0.297223 [Debug] dit_step46_xt: [2170, 64] first4: 0.631054 1.487258 -0.026416 -0.042342 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.444894 -1.265309 0.203886 0.275377 [Debug] dit_step47_xt: [2170, 64] first4: 0.639952 1.512564 -0.030494 -0.047849 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.413320 -1.285746 0.219039 0.217924 [Debug] dit_step48_xt: [2170, 64] first4: 0.648218 1.538279 -0.034875 -0.052208 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.439540 -1.310946 0.194238 0.212056 [Debug] dit_x0: [2170, 64] first4: 0.657009 1.564498 -0.038759 -0.056449 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 1828.7 ms (1828.7 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.657009 1.564498 -0.038759 -0.056449 [Store] Unload DiT (1237.2 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: CUDA0 (CPU threads: 16) [VAE] Backend: CUDA0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 666 ms [VAE] Tiled decode: 3 tiles (chunk=1024, overlap=64, stride=896) [VAE] Graph: 479 nodes, T_latent=960 [VAE] Upsample factor: 1920.00 (expected ~1920) [VAE] Graph: 479 nodes, T_latent=1024 [VAE] Graph: 479 nodes, T_latent=442 [VAE] Tiled decode done: 3 tiles -> T_audio=4166400 (86.80s @ 48kHz) [VAE-Decode Batch0] Decode: 628.8 ms [Debug] vae_audio: [2, 4166400] first4: -0.001554 -0.000981 -0.001207 -0.000807 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 403 ms (215.51x realtime), 32 threads [MP3] Wrote ggml-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 10:13:29.114 | 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 10:13:30.324 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:13:30.325 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:13:30.325 | 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 10:13:30.328 | 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 10:13:30.333 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:13:30.337 | 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 10:13:30.339 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:13:30.339 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:13:30.339 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:13:30.339 | 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 10:13:30.339 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:13:30.339 | 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 10:13:30.339 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:13:30.342 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:13:30.377 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:13:30.378 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:13:30.384 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... [Request] Loaded request0.json [Turbo] steps=8, shift=3.0 | acestep-v15-turbo-Q6_K.gguf [GGML] Running acestep-v15-turbo-Q6_K.gguf... [GGML] Done, 47 dump files [Python] Initializing acestep-v15-turbo... [Python] Generating (acestep-v15-turbo, 8 steps)... Using precomputed LM hints Using precomputed LM hints [Python] Wrote python-turbo/output.wav: 4166400 samples (86.80s @ 48kHz stereo) [Python] Done, 40 dump files [Turbo] Cosine similarities GGML vs Python stage GGML vs Python text_hidden 0.999797 lyric_embed 1.000000 enc_hidden 0.999634 detok_output 0.999961 context 0.999976 noise 1.000000 temb_t 0.999990 hidden_after_proj_in 0.999980 enc_after_cond_emb 0.999645 layer0_sa_output 0.999763 hidden_after_layer0 0.999877 hidden_after_layer6 0.999858 hidden_after_layer12 0.999025 hidden_after_layer18 0.996061 hidden_after_layer23 0.992505 dit_step0_vt 0.972054 dit_step0_xt 0.999938 dit_step1_vt 0.975107 dit_step1_xt 0.999805 dit_step2_vt 0.978115 dit_step2_xt 0.999480 dit_step3_vt 0.977404 dit_step3_xt 0.998726 dit_step4_vt 0.975793 dit_step4_xt 0.997024 dit_step5_vt 0.972429 dit_step5_xt 0.993123 dit_step6_vt 0.969075 dit_step6_xt 0.985638 dit_step7_vt 0.961565 dit_x0 0.974389 vae_audio 0.892850 vae_audio (STFT cosine) 0.969453 [Turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999938 0.148313 0.007155 -0.002267 0.972930 -0.002342 0.972003 dit_step1_xt 0.999805 0.289104 0.012387 -0.005205 0.942683 -0.005313 0.941730 dit_step2_xt 0.999480 0.473056 0.019197 -0.009175 0.909091 -0.009311 0.908527 dit_step3_xt 0.998726 0.730935 0.028763 -0.014471 0.873593 -0.014577 0.873624 dit_step4_xt 0.997024 1.057992 0.042176 -0.021567 0.841570 -0.021660 0.841995 dit_step5_xt 0.993123 1.530323 0.062504 -0.031710 0.824663 -0.032109 0.824593 dit_step6_xt 0.985638 2.183978 0.093149 -0.046075 0.855391 -0.046482 0.855546 [SFT] steps=50, shift=1.0 | acestep-v15-sft-Q6_K.gguf [GGML] Running acestep-v15-sft-Q6_K.gguf... [GGML] Done, 131 dump files [Python] Initializing acestep-v15-sft... [Python] Generating (acestep-v15-sft, 50 steps)... Using precomputed LM hints Using precomputed LM hints 0%| | 0/50 [00:00 Text-Enc [Registry] Qwen3-Embedding-0.6B-Q8_0.gguf -> Text-Enc [Registry] acestep-5Hz-lm-0.6B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-0.6B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-4B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q5_K_M.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q6_K.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q8_0.gguf -> LM [Registry] acestep-v15-base-BF16.gguf -> DiT [Registry] acestep-v15-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-base-Q6_K.gguf -> DiT [Registry] acestep-v15-base-Q8_0.gguf -> DiT [Registry] acestep-v15-sft-BF16.gguf -> DiT [Registry] acestep-v15-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-BF16.gguf -> DiT [Registry] acestep-v15-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-continuous-BF16.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift1-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift3-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-base-BF16.gguf -> DiT [Registry] acestep-v15-xl-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-base-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sft-BF16.gguf -> DiT [Registry] acestep-v15-xl-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-turbo-BF16.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q8_0.gguf -> DiT [Registry] scragvae-BF16.gguf -> VAE [Registry] vae-BF16.gguf -> VAE [Store] Created (policy=STRICT) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q6_K.gguf: 830 tensors, data at offset 69088 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q6_K.gguf: 830 tensors, data at offset 69088 [Synth-Load] Ready: turbo=yes, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=8, guidance=1.0, shift=3.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: CUDA0 (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: 267 ms [Debug] text_hidden: [70, 1024] first4: 3.652379 1.048645 0.229843 -13.063987 [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: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q6_K.gguf: 830 tensors, data at offset 69088 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 141 tensors, 476.3 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L, CLS), text_proj, null_cond [Store] Load CondEnc: 297 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, 15.2 ms [Debug] enc_hidden: [238, 2048] first4: 0.687235 -0.019297 -0.051811 0.022688 [Store] Unload CondEnc (476.3 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] Detokenizer backend: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q6_K.gguf: 830 tensors, data at offset 69088 [WeightCtx] Loaded 30 tensors, 82.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 53 ms CUDA Graph id 8 reused ggml_backend_cuda_graph_compute: CUDA graph warmup complete CUDA Graph id 10 reused ggml_backend_cuda_graph_compute: CUDA graph warmup complete CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 75.9 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.138872 1.452165 0.311136 -0.633141 [Store] Unload FSQ-Detok (82.2 MB) [Init-Noise Batch0] Philox noise seed=42, [2170, 64] solver=euler [Debug] noise: [2170, 64] first4: 0.194336 2.156250 -0.171875 0.847656 [Debug] context: [2170, 128] first4: -0.138872 1.452165 0.311136 -0.633141 [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: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q6_K.gguf: 830 tensors, data at offset 69088 [DiT] Self-attn: Q+K+V fused [DiT] Cross-attn: Q+K+V fused [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 630 tensors, 3267.8 MB into backend [Load] DiT: 32 layers, H=2560, Nh=32/8, D=128 [Store] Load DiT: 1934 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2825 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [15360] first4: 0.017359 0.105443 0.062030 -0.074535 [Debug] temb: [2560] first4: 0.026573 -0.022718 -0.016417 0.046948 [Debug] temb_t: [2560] first4: 0.014720 -0.022140 -0.020709 0.048493 [Debug] temb_r: [2560] first4: 0.011852 -0.000578 0.004292 -0.001545 [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: 0.062568 -0.018019 0.137460 -0.111293 [Debug] temb_lin1_r: [2560] first4: -0.011992 0.015307 -0.017416 0.011656 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.339717 -0.580721 0.251656 -0.188251 [Debug] proj_in_input: [192, 2170] first4: -0.138872 1.452165 0.311136 -0.633141 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.024957 0.664129 0.106195 0.858601 [Debug] layer0_sa_input: [2560, 1085] first4: 0.041100 -0.049584 0.091245 -0.134373 [Debug] layer0_q_after_rope: [128, 32] first4: -0.252849 -2.080972 0.303769 0.208942 [Debug] layer0_k_after_rope: [128, 8] first4: 0.047886 2.156067 0.729569 0.536962 [Debug] layer0_sa_output: [2560, 1085] first4: 1.679846 4.281067 -0.522369 5.346257 [Debug] layer0_attn_out: [4096, 1085] first4: 0.476246 -2.239920 -0.129601 -2.768826 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.476246 -2.239920 -0.129601 -2.768826 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.338366 -2.000539 0.041954 -3.725858 [Debug] hidden_after_layer0: [2560, 1085] first4: 1.742994 -2.412603 241.561249 -3.700965 [Debug] hidden_after_layer6: [2560, 1085] first4: -6.410078 -2.372195 319.891571 -12.513138 [Debug] hidden_after_layer12: [2560, 1085] first4: -10.713213 29.206194 264.577362 30.847910 [Debug] hidden_after_layer18: [2560, 1085] first4: 41.627663 14.022497 281.020599 7.850004 [Debug] hidden_after_layer31: [2560, 1085] first4: -61.533188 -246.610138 -139.306381 173.200012 [Debug] dit_step0_vt: [2170, 64] first4: -1.260313 1.674315 0.500180 1.244653 [Debug] dit_step0_xt: [2170, 64] first4: 0.251623 2.080145 -0.194610 0.791081 [DiT] Step 1/8 t=1.000 ggml_backend_cuda_graph_compute: CUDA graph warmup complete [Debug] dit_step1_vt: [2170, 64] first4: -0.901732 2.499394 0.774408 1.196849 [Debug] dit_step1_xt: [2170, 64] first4: 0.300808 1.943814 -0.236851 0.725798 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: -1.423434 2.045938 1.247103 1.098293 [Debug] dit_step2_xt: [2170, 64] first4: 0.395704 1.807418 -0.319991 0.652579 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: -0.827302 1.502673 1.215600 0.887736 [Debug] dit_step3_xt: [2170, 64] first4: 0.464646 1.682195 -0.421291 0.578601 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: 0.322908 2.612352 1.218055 0.788015 [Debug] dit_step4_xt: [2170, 64] first4: 0.430049 1.402301 -0.551797 0.494171 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: 0.485048 1.949302 1.249014 0.671570 [Debug] dit_step5_xt: [2170, 64] first4: 0.360756 1.123829 -0.730228 0.398232 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: -1.551279 2.270134 0.598492 0.942397 [Debug] dit_step6_xt: [2170, 64] first4: 0.671012 0.669802 -0.849926 0.209753 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: -1.538725 1.414308 0.713661 0.492929 [Debug] dit_x0: [2170, 64] first4: 1.132629 0.245510 -1.064024 0.061874 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 747.4 ms (747.4 ms/sample) [Debug] dit_output: [2170, 64] first4: 1.132629 0.245510 -1.064024 0.061874 [Store] Unload DiT (3267.8 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: CUDA0 (CPU threads: 16) [VAE] Backend: CUDA0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 673 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: 615.8 ms [Debug] vae_audio: [2, 4166400] first4: 0.001888 0.004482 0.004372 0.003374 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 467 ms (185.81x realtime), 32 threads [MP3] Wrote ggml-xl-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 10:13:38.828 | 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 0.999797 lyric_embed 1.000000 enc_hidden 0.999634 detok_output 0.999961 context 0.999976 noise 1.000000 temb_t 0.999970 hidden_after_proj_in 0.999980 enc_after_cond_emb 0.999646 layer0_sa_output 0.999770 hidden_after_layer0 0.999907 hidden_after_layer6 0.999778 hidden_after_layer12 0.999314 hidden_after_layer18 0.998524 hidden_after_layer23 0.998802 dit_step0_vt 0.998703 dit_step0_xt 1.000000 dit_step5_vt 0.999132 dit_step5_xt 0.999974 dit_step10_vt 0.998509 dit_step10_xt 0.999909 dit_step15_vt 0.996965 dit_step15_xt 0.999758 dit_step20_vt 0.994477 dit_step20_xt 0.999384 dit_step25_vt 0.990332 dit_step25_xt 0.998652 dit_step30_vt 0.985238 dit_step30_xt 0.997389 dit_step35_vt 0.979842 dit_step35_xt 0.995716 dit_step40_vt 0.976347 dit_step40_xt 0.993981 dit_step45_vt 0.980324 dit_step45_xt 0.992728 dit_step49_vt 0.986679 dit_x0 0.992293 vae_audio 0.956442 vae_audio (STFT cosine) 0.980874 [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 0.019750 0.001494 -0.001665 0.978611 -0.001619 0.978872 dit_step5_xt 0.999974 0.084931 0.005109 -0.008398 0.885584 -0.008480 0.883811 dit_step10_xt 0.999909 0.101447 0.008222 -0.015068 0.802022 -0.015075 0.802324 dit_step15_xt 0.999758 0.140242 0.012242 -0.021672 0.731791 -0.021933 0.731304 dit_step20_xt 0.999384 0.229565 0.017897 -0.028284 0.679302 -0.028882 0.679066 dit_step25_xt 0.998652 0.386364 0.024861 -0.034945 0.648993 -0.035643 0.650780 dit_step30_xt 0.997389 0.589262 0.033825 -0.041630 0.644066 -0.042577 0.647107 dit_step35_xt 0.995716 0.829774 0.044366 -0.048307 0.665206 -0.049471 0.669256 dit_step40_xt 0.993981 1.088289 0.056050 -0.054996 0.710012 -0.056532 0.715783 dit_step45_xt 0.992728 1.319601 0.067155 -0.061759 0.774198 -0.063533 0.780575 [Xl-turbo] steps=8, shift=3.0 | acestep-v15-xl-turbo-Q6_K.gguf [GGML] Running acestep-v15-xl-turbo-Q6_K.gguf... [GGML] Done, 47 dump files [Python] Initializing acestep-v15-xl-turbo... Loading checkpoint shards: 0%| | 0/4 [00:00 1.0 (turbo does not use CFG). 2026-07-12 10:13:41.569 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:13:41.569 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:13:41.570 | 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 10:13:41.573 | INFO | acestep.core.generation.handler.generate_music:_vram_preflight_check:156 - [generate_music] VRAM pre-flight: 76.76 GB free, ~1.23 GB needed (batch=1, duration=88s, mode=xl_turbo). 2026-07-12 10:13:41.577 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:13:41.587 | 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 10:13:41.588 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:13:41.588 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:13:41.588 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:13:41.588 | 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 10:13:41.588 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:13:41.588 | 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 10:13:41.588 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:13:41.592 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:13:41.599 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:13:41.599 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:13:41.606 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 10:13:41.613 | 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 10:13:42.168 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 10:13:42.169 | 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 10:13:42.169 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.006384372711181641, 'diffusion_time_cost': 0.5550234317779541, 'diffusion_per_step_time_cost': 0.06937792897224426, 'total_time_cost': 0.5614078044891357, 'offload_time_cost': 0.0} 2026-07-12 10:13:42.192 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 10:13:42.193 | 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 10:13:42.193 | 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 10:13:42.193 | 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 10:13:42.194 | 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 10:13:42.194 | 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 10:13:42.194 | 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 10:13:42.467 | DEBUG | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:193 - [generate_music] After VAE decode: allocated=17.21GB, max=18.50GB 2026-07-12 10:13:42.669 | 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 10:13:42.672 | INFO | acestep.core.generation.handler.generate_music_payload:_build_generate_music_success_payload:50 - [generate_music] Done! Generated 1 audio tensors. [Request] Parsed ggml-xl-sft/request0.json [Request] seed=42 lm_batch=1 synth_batch=1 [Request] caption: An upbeat and anthemic pop-rock track driven by bright, slig... [Request] lyrics: 490 bytes [Request] bpm=83 dur=88 key=G major ts=4 lang=fr [Request] lm: temp=0.85 cfg=2.0 top_p=0.90 top_k=0 [Request] dit: steps=50 guidance=1.0 shift=1.0 [Request] task_type: text2music [Request] solver: euler (stork_substeps=10) [Request] lm_mode: generate [Request] output_format: mp3 [Request] synth_model: acestep-v15-xl-sft-Q6_K.gguf [Request] vae: vae-BF16.gguf [Request] audio_codes: (present) [Ace-Synth] Batch: 1 request(s) [Registry] Qwen3-Embedding-0.6B-BF16.gguf -> Text-Enc [Registry] Qwen3-Embedding-0.6B-Q8_0.gguf -> Text-Enc [Registry] acestep-5Hz-lm-0.6B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-0.6B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-4B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q5_K_M.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q6_K.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q8_0.gguf -> LM [Registry] acestep-v15-base-BF16.gguf -> DiT [Registry] acestep-v15-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-base-Q6_K.gguf -> DiT [Registry] acestep-v15-base-Q8_0.gguf -> DiT [Registry] acestep-v15-sft-BF16.gguf -> DiT [Registry] acestep-v15-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-BF16.gguf -> DiT [Registry] acestep-v15-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-continuous-BF16.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift1-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift3-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-base-BF16.gguf -> DiT [Registry] acestep-v15-xl-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-base-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sft-BF16.gguf -> DiT [Registry] acestep-v15-xl-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-turbo-BF16.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q8_0.gguf -> DiT [Registry] scragvae-BF16.gguf -> VAE [Registry] vae-BF16.gguf -> VAE [Store] Created (policy=STRICT) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q6_K.gguf: 830 tensors, data at offset 69056 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q6_K.gguf: 830 tensors, data at offset 69056 [Synth-Load] Ready: turbo=no, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=50, guidance=1.0, shift=1.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: CUDA0 (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: 280 ms [Debug] text_hidden: [70, 1024] first4: 3.652379 1.048645 0.229843 -13.063987 [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: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q6_K.gguf: 830 tensors, data at offset 69056 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K+V fused [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 141 tensors, 476.3 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L, CLS), text_proj, null_cond [Store] Load CondEnc: 321 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, 15.6 ms [Debug] enc_hidden: [238, 2048] first4: 0.687235 -0.019297 -0.051811 0.022688 [Store] Unload CondEnc (476.3 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] Detokenizer backend: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q6_K.gguf: 830 tensors, data at offset 69056 [WeightCtx] Loaded 30 tensors, 82.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 57 ms CUDA Graph id 8 reused ggml_backend_cuda_graph_compute: CUDA graph warmup complete CUDA Graph id 10 reused ggml_backend_cuda_graph_compute: CUDA graph warmup complete CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused CUDA Graph id 8 reused CUDA Graph id 10 reused [Context] Decoded: 434 codes -> 2170 frames (86.8s @ 25Hz) [Build-Context Batch0] Detokenizer: 76.8 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.138872 1.452165 0.311136 -0.633141 [Store] Unload FSQ-Detok (82.2 MB) [Init-Noise Batch0] Philox noise seed=42, [2170, 64] solver=euler [Debug] noise: [2170, 64] first4: 0.194336 2.156250 -0.171875 0.847656 [Debug] context: [2170, 128] first4: -0.138872 1.452165 0.311136 -0.633141 [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: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q6_K.gguf: 830 tensors, data at offset 69056 [DiT] Self-attn: Q+K+V fused [DiT] Cross-attn: Q+K+V fused [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 630 tensors, 3267.8 MB into backend [Load] DiT: 32 layers, H=2560, Nh=32/8, D=128 [Store] Load DiT: 2047 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2825 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [15360] first4: 0.010151 0.102272 0.074128 -0.034786 [Debug] temb: [2560] first4: 0.009691 -0.007795 -0.006042 0.004612 [Debug] temb_t: [2560] first4: 0.008209 -0.007992 -0.006583 0.003656 [Debug] temb_r: [2560] first4: 0.001482 0.000197 0.000541 0.000956 [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: 0.033809 -0.083329 0.088373 -0.064607 [Debug] temb_lin1_r: [2560] first4: 0.009306 -0.004837 -0.002965 -0.011075 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.329120 -0.496207 0.122636 -0.211715 [Debug] proj_in_input: [192, 2170] first4: -0.138872 1.452165 0.311136 -0.633141 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.032805 0.637510 0.118305 0.856546 [Debug] layer0_sa_input: [2560, 1085] first4: 0.034109 -0.022918 0.100164 -0.099605 [Debug] layer0_q_after_rope: [128, 32] first4: -0.276097 -2.157666 0.231967 0.048108 [Debug] layer0_k_after_rope: [128, 8] first4: 0.752848 2.962177 1.231689 1.936058 [Debug] layer0_sa_output: [2560, 1085] first4: 1.974614 4.398660 -0.777708 7.638257 [Debug] layer0_attn_out: [4096, 1085] first4: 0.459360 -2.182381 -0.453893 -4.010606 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.459360 -2.182381 -0.453893 -4.010606 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.285458 -1.908571 -0.374283 -4.100399 [Debug] hidden_after_layer0: [2560, 1085] first4: -1.619007 -1.254476 244.554718 -3.998252 [Debug] hidden_after_layer6: [2560, 1085] first4: -7.643999 3.823754 305.992523 -21.724653 [Debug] hidden_after_layer12: [2560, 1085] first4: -22.661005 37.601696 224.604065 15.789473 [Debug] hidden_after_layer18: [2560, 1085] first4: 10.281745 -13.777548 226.570953 19.817894 [Debug] hidden_after_layer31: [2560, 1085] first4: 65.021729 -106.501411 1.724251 81.214783 [Debug] dit_step0_vt: [2170, 64] first4: -0.316067 2.192628 0.037308 1.673512 [Debug] dit_step0_xt: [2170, 64] first4: 0.200657 2.112397 -0.172621 0.814186 [DiT] Step 1/50 t=1.000 ggml_backend_cuda_graph_compute: CUDA graph warmup complete [Debug] dit_step1_vt: [2170, 64] first4: -0.346052 2.120087 0.088884 1.543391 [Debug] dit_step1_xt: [2170, 64] first4: 0.207578 2.069996 -0.174399 0.783318 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.326836 2.151063 0.081904 1.558125 [Debug] dit_step2_xt: [2170, 64] first4: 0.214115 2.026974 -0.176037 0.752156 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.272440 2.127313 0.083486 1.555750 [Debug] dit_step3_xt: [2170, 64] first4: 0.219564 1.984428 -0.177707 0.721041 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.265475 2.050473 0.054241 1.570176 [Debug] dit_step4_xt: [2170, 64] first4: 0.224873 1.943419 -0.178791 0.689637 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.263569 2.053598 0.077525 1.590581 [Debug] dit_step5_xt: [2170, 64] first4: 0.230145 1.902347 -0.180342 0.657825 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.256271 2.013772 0.088298 1.602440 [Debug] dit_step6_xt: [2170, 64] first4: 0.235270 1.862071 -0.182108 0.625777 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.249435 1.976493 0.129031 1.596092 [Debug] dit_step7_xt: [2170, 64] first4: 0.240259 1.822541 -0.184689 0.593855 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.266758 1.932611 0.150367 1.593687 [Debug] dit_step8_xt: [2170, 64] first4: 0.245594 1.783889 -0.187696 0.561981 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.287645 1.868685 0.137257 1.587183 [Debug] dit_step9_xt: [2170, 64] first4: 0.251347 1.746516 -0.190441 0.530237 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.306527 1.827356 0.125327 1.580803 [Debug] dit_step10_xt: [2170, 64] first4: 0.257477 1.709968 -0.192948 0.498621 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.306439 1.776734 0.113146 1.568733 [Debug] dit_step11_xt: [2170, 64] first4: 0.263606 1.674434 -0.195210 0.467247 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.303816 1.721917 0.099447 1.548697 [Debug] dit_step12_xt: [2170, 64] first4: 0.269682 1.639995 -0.197199 0.436273 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.304374 1.673868 0.083867 1.531562 [Debug] dit_step13_xt: [2170, 64] first4: 0.275770 1.606518 -0.198877 0.405642 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.283001 1.624147 0.049378 1.530697 [Debug] dit_step14_xt: [2170, 64] first4: 0.281430 1.574035 -0.199864 0.375028 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.273944 1.568276 0.045763 1.528582 [Debug] dit_step15_xt: [2170, 64] first4: 0.286909 1.542670 -0.200780 0.344456 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.246314 1.526303 0.022635 1.529150 [Debug] dit_step16_xt: [2170, 64] first4: 0.291835 1.512143 -0.201232 0.313873 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.227878 1.469722 0.002723 1.518573 [Debug] dit_step17_xt: [2170, 64] first4: 0.296393 1.482749 -0.201287 0.283501 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.180847 1.417195 0.012639 1.505873 [Debug] dit_step18_xt: [2170, 64] first4: 0.300010 1.454405 -0.201540 0.253384 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.162145 1.365161 -0.013610 1.485702 [Debug] dit_step19_xt: [2170, 64] first4: 0.303253 1.427102 -0.201267 0.223670 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.125838 1.323303 -0.019397 1.481134 [Debug] dit_step20_xt: [2170, 64] first4: 0.305769 1.400636 -0.200879 0.194047 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.109043 1.273069 -0.046435 1.480630 [Debug] dit_step21_xt: [2170, 64] first4: 0.307950 1.375175 -0.199951 0.164435 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.091522 1.237471 -0.040751 1.464821 [Debug] dit_step22_xt: [2170, 64] first4: 0.309781 1.350425 -0.199136 0.135138 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.048124 1.169007 -0.035027 1.457165 [Debug] dit_step23_xt: [2170, 64] first4: 0.310743 1.327045 -0.198435 0.105995 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.043557 1.128490 -0.059191 1.432740 [Debug] dit_step24_xt: [2170, 64] first4: 0.311614 1.304475 -0.197251 0.077340 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.042960 1.071704 -0.068339 1.416435 [Debug] dit_step25_xt: [2170, 64] first4: 0.312473 1.283041 -0.195884 0.049011 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.057339 1.015468 -0.077961 1.385843 [Debug] dit_step26_xt: [2170, 64] first4: 0.313620 1.262732 -0.194325 0.021295 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.057690 0.951553 -0.077699 1.374532 [Debug] dit_step27_xt: [2170, 64] first4: 0.314774 1.243701 -0.192771 -0.006196 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.065791 0.901671 -0.113930 1.344916 [Debug] dit_step28_xt: [2170, 64] first4: 0.316090 1.225667 -0.190493 -0.033094 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.079240 0.796226 -0.130486 1.316266 [Debug] dit_step29_xt: [2170, 64] first4: 0.317675 1.209743 -0.187883 -0.059420 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.065978 0.741877 -0.140991 1.289006 [Debug] dit_step30_xt: [2170, 64] first4: 0.318994 1.194905 -0.185063 -0.085200 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.095300 0.672658 -0.154950 1.245313 [Debug] dit_step31_xt: [2170, 64] first4: 0.320900 1.181452 -0.181964 -0.110106 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.102566 0.574676 -0.131005 1.226383 [Debug] dit_step32_xt: [2170, 64] first4: 0.322952 1.169958 -0.179344 -0.134634 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.117988 0.496552 -0.163723 1.208066 [Debug] dit_step33_xt: [2170, 64] first4: 0.325311 1.160027 -0.176070 -0.158795 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.152328 0.395106 -0.150882 1.185050 [Debug] dit_step34_xt: [2170, 64] first4: 0.328358 1.152125 -0.173052 -0.182496 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.157734 0.326846 -0.145120 1.149352 [Debug] dit_step35_xt: [2170, 64] first4: 0.331513 1.145588 -0.170150 -0.205483 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.137977 0.223650 -0.138674 1.128036 [Debug] dit_step36_xt: [2170, 64] first4: 0.334272 1.141115 -0.167376 -0.228044 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.174962 0.103992 -0.131920 1.076288 [Debug] dit_step37_xt: [2170, 64] first4: 0.337771 1.139035 -0.164738 -0.249570 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.183660 0.013020 -0.115287 1.050973 [Debug] dit_step38_xt: [2170, 64] first4: 0.341444 1.138775 -0.162432 -0.270589 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.229824 -0.098910 -0.100294 0.986662 [Debug] dit_step39_xt: [2170, 64] first4: 0.346041 1.140753 -0.160426 -0.290322 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.245596 -0.212491 -0.082895 0.938103 [Debug] dit_step40_xt: [2170, 64] first4: 0.350953 1.145003 -0.158768 -0.309084 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.251332 -0.324740 -0.044176 0.893726 [Debug] dit_step41_xt: [2170, 64] first4: 0.355980 1.151498 -0.157885 -0.326959 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.280275 -0.421558 -0.040252 0.870646 [Debug] dit_step42_xt: [2170, 64] first4: 0.361585 1.159929 -0.157080 -0.344372 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.325324 -0.530322 -0.040450 0.813340 [Debug] dit_step43_xt: [2170, 64] first4: 0.368092 1.170535 -0.156271 -0.360639 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.334626 -0.620794 0.021875 0.750188 [Debug] dit_step44_xt: [2170, 64] first4: 0.374784 1.182951 -0.156708 -0.375642 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.382806 -0.681148 0.110001 0.707148 [Debug] dit_step45_xt: [2170, 64] first4: 0.382440 1.196574 -0.158908 -0.389785 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.415717 -0.747008 0.136931 0.672017 [Debug] dit_step46_xt: [2170, 64] first4: 0.390755 1.211514 -0.161647 -0.403226 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.495854 -0.835213 0.202784 0.631986 [Debug] dit_step47_xt: [2170, 64] first4: 0.400672 1.228219 -0.165702 -0.415865 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.537848 -0.897493 0.264666 0.631197 [Debug] dit_step48_xt: [2170, 64] first4: 0.411429 1.246168 -0.170996 -0.428489 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.547714 -0.877415 0.170556 0.642186 [Debug] dit_x0: [2170, 64] first4: 0.422383 1.263717 -0.174407 -0.441333 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 4431.9 ms (4431.9 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.422383 1.263717 -0.174407 -0.441333 [Store] Unload DiT (3267.8 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: CUDA0 (CPU threads: 16) [VAE] Backend: CUDA0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 666 ms [VAE] Tiled decode: 3 tiles (chunk=1024, overlap=64, stride=896) [VAE] Graph: 479 nodes, T_latent=960 [VAE] Upsample factor: 1920.00 (expected ~1920) [VAE] Graph: 479 nodes, T_latent=1024 [VAE] Graph: 479 nodes, T_latent=442 [VAE] Tiled decode done: 3 tiles -> T_audio=4166400 (86.80s @ 48kHz) [VAE-Decode Batch0] Decode: 615.3 ms [Debug] vae_audio: [2, 4166400] first4: -0.001303 -0.000684 -0.000679 -0.000196 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 416 ms (208.85x realtime), 32 threads [MP3] Wrote ggml-xl-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 10:13:52.760 | 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 0.999797 lyric_embed 1.000000 enc_hidden 0.999363 detok_output 0.999961 context 0.999976 noise 1.000000 temb_t 0.999974 hidden_after_proj_in 0.999970 enc_after_cond_emb 0.999433 layer0_sa_output 0.999214 hidden_after_layer0 0.999301 hidden_after_layer6 0.999798 hidden_after_layer12 0.996920 hidden_after_layer18 0.996160 hidden_after_layer31 0.988712 dit_step0_vt 0.916352 dit_step0_xt 0.999849 dit_step1_vt 0.930602 dit_step1_xt 0.999547 dit_step2_vt 0.932367 dit_step2_xt 0.998841 dit_step3_vt 0.932471 dit_step3_xt 0.997271 dit_step4_vt 0.925951 dit_step4_xt 0.993230 dit_step5_vt 0.915972 dit_step5_xt 0.983281 dit_step6_vt 0.909652 dit_step6_xt 0.963363 dit_step7_vt 0.891261 dit_x0 0.931059 vae_audio 0.739813 vae_audio (STFT cosine) 0.925935 [Xl-turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999849 0.149059 0.012235 -0.002972 0.978336 -0.002721 0.976534 dit_step1_xt 0.999547 0.306811 0.020854 -0.005952 0.951997 -0.005567 0.949004 dit_step2_xt 0.998841 0.524989 0.031960 -0.009783 0.923100 -0.009330 0.918931 dit_step3_xt 0.997271 0.785000 0.046718 -0.015495 0.892124 -0.014418 0.887241 dit_step4_xt 0.993230 1.155480 0.070325 -0.023405 0.862608 -0.021453 0.857796 dit_step5_xt 0.983281 1.736475 0.107997 -0.035171 0.845667 -0.032240 0.840033 dit_step6_xt 0.963363 2.506787 0.165096 -0.052857 0.870366 -0.049194 0.863964 [Xl-sft] steps=50, shift=1.0 | acestep-v15-xl-sft-Q6_K.gguf [GGML] Running acestep-v15-xl-sft-Q6_K.gguf... [GGML] Done, 131 dump files [Python] Initializing acestep-v15-xl-sft... Loading checkpoint shards: 0%| | 0/4 [00:00 2026-07-12 10:13:55.455 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:13:55.455 | 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 10:13:55.455 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:13:55.459 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:13:55.466 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:13:55.466 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:13:55.473 | 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