[Request] Parsed ggml-turbo/request0.json [Request] seed=42 lm_batch=1 synth_batch=1 [Request] caption: An upbeat and anthemic pop-rock track driven by bright, slig... [Request] lyrics: 490 bytes [Request] bpm=83 dur=88 key=G major ts=4 lang=fr [Request] lm: temp=0.85 cfg=2.0 top_p=0.90 top_k=0 [Request] dit: steps=8 guidance=0.0 shift=3.0 [Request] task_type: text2music [Request] solver: euler (stork_substeps=10) [Request] lm_mode: generate [Request] output_format: mp3 [Request] synth_model: acestep-v15-turbo-Q5_K_M.gguf [Request] vae: vae-BF16.gguf [Request] audio_codes: (present) [Ace-Synth] Batch: 1 request(s) [Registry] Qwen3-Embedding-0.6B-BF16.gguf -> Text-Enc [Registry] Qwen3-Embedding-0.6B-Q8_0.gguf -> Text-Enc [Registry] acestep-5Hz-lm-0.6B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-0.6B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-4B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q5_K_M.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q6_K.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q8_0.gguf -> LM [Registry] acestep-v15-base-BF16.gguf -> DiT [Registry] acestep-v15-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-base-Q6_K.gguf -> DiT [Registry] acestep-v15-base-Q8_0.gguf -> DiT [Registry] acestep-v15-sft-BF16.gguf -> DiT [Registry] acestep-v15-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-BF16.gguf -> DiT [Registry] acestep-v15-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-continuous-BF16.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift1-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift3-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-base-BF16.gguf -> DiT [Registry] acestep-v15-xl-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-base-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sft-BF16.gguf -> DiT [Registry] acestep-v15-xl-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-turbo-BF16.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q8_0.gguf -> DiT [Registry] scragvae-BF16.gguf -> VAE [Registry] vae-BF16.gguf -> VAE [Store] Created (policy=STRICT) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q5_K_M.gguf: 678 tensors, data at offset 56864 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q5_K_M.gguf: 678 tensors, data at offset 56864 [Synth-Load] Ready: turbo=yes, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=8, guidance=1.0, shift=3.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: 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: 287 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-Q5_K_M.gguf: 678 tensors, data at offset 56864 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 140 tensors, 412.5 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L), text_proj, null_cond [Store] Load CondEnc: 282 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, 20.5 ms [Debug] enc_hidden: [238, 2048] first4: 1.760428 -0.050975 -0.131559 0.057932 [Store] Unload CondEnc (412.5 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] Detokenizer backend: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-turbo-Q5_K_M.gguf: 678 tensors, data at offset 56864 [WeightCtx] Loaded 30 tensors, 73.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 51 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: 79.0 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.123923 1.460390 0.290576 -0.653908 [Store] Unload FSQ-Detok (73.2 MB) [Init-Noise Batch0] Philox noise seed=42, [2170, 64] solver=euler [Debug] noise: [2170, 64] first4: 0.194336 2.156250 -0.171875 0.847656 [Debug] context: [2170, 128] first4: -0.123923 1.460390 0.290576 -0.653908 [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-Q5_K_M.gguf: 678 tensors, data at offset 56864 [DiT] Self-attn: Q+K fused, V separate [DiT] Cross-attn: all separate [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 478 tensors, 1061.2 MB into backend [Load] DiT: 24 layers, H=2048, Nh=16/8, D=128 [Store] Load DiT: 685 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2063 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [12288] first4: 0.260130 -0.161413 -0.102271 0.051211 [Debug] temb: [2048] first4: -0.000033 -0.132307 -0.035515 0.064775 [Debug] temb_t: [2048] first4: 0.000653 0.026699 -0.052806 0.063542 [Debug] temb_r: [2048] first4: -0.000685 -0.159005 0.017290 0.001234 [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: -0.051436 -0.053873 -0.011918 -0.038393 [Debug] temb_lin1_r: [2048] first4: -0.016164 -0.021120 -0.015800 -0.000525 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.043332 -0.944784 0.540302 0.454851 [Debug] proj_in_input: [192, 2170] first4: -0.123923 1.460390 0.290576 -0.653908 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.153065 0.748113 0.330677 -0.520041 [Debug] layer0_sa_input: [2048, 1085] first4: -0.721768 -0.749294 -0.052004 0.264110 [Debug] layer0_q_after_rope: [128, 16] first4: -0.153065 0.748113 0.330677 -0.520041 [Debug] layer0_k_after_rope: [128, 8] first4: -1.545187 -1.001738 0.157244 0.465787 [Debug] layer0_sa_output: [2048, 1085] first4: -1.512387 0.128997 -0.393924 0.474906 [Debug] layer0_attn_out: [2048, 1085] first4: -1.545187 -1.001738 0.157244 0.465787 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.545187 -1.001738 0.157244 0.465787 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.592080 -0.792732 -0.329140 0.496456 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.120616 0.585962 51.064106 -0.814500 [Debug] hidden_after_layer6: [2048, 1085] first4: -20.266460 0.438827 28.112282 -5.063588 [Debug] hidden_after_layer12: [2048, 1085] first4: -18.454622 -18.145733 69.498116 24.758778 [Debug] hidden_after_layer18: [2048, 1085] first4: -23.937134 9.779724 65.806046 18.612759 [Debug] hidden_after_layer23: [2048, 1085] first4: -1.096336 29.231785 202.331970 155.837204 [Debug] dit_step0_vt: [2170, 64] first4: 0.011483 1.175653 0.292105 2.369347 [Debug] dit_step0_xt: [2170, 64] first4: 0.193814 2.102811 -0.185153 0.739959 [DiT] Step 1/8 t=1.000 ggml_backend_cuda_graph_compute: CUDA graph warmup complete [Debug] dit_step1_vt: [2170, 64] first4: -0.208771 1.400575 -0.147506 1.878888 [Debug] dit_step1_xt: [2170, 64] first4: 0.205202 2.026416 -0.177107 0.637474 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: -0.054715 1.273130 0.083181 2.373883 [Debug] dit_step2_xt: [2170, 64] first4: 0.208849 1.941541 -0.182652 0.479215 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: 0.320739 1.085841 0.254760 2.632028 [Debug] dit_step3_xt: [2170, 64] first4: 0.182121 1.851054 -0.203882 0.259879 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: 0.367912 0.907669 0.059071 2.666218 [Debug] dit_step4_xt: [2170, 64] first4: 0.142702 1.753804 -0.210211 -0.025787 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: 0.364670 0.737234 -0.379900 2.695297 [Debug] dit_step5_xt: [2170, 64] first4: 0.090606 1.648485 -0.155940 -0.410829 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: 0.185300 0.650105 -0.946220 2.800602 [Debug] dit_step6_xt: [2170, 64] first4: 0.053546 1.518464 0.033304 -0.970950 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: 0.057908 0.300911 -1.435219 3.188086 [Debug] dit_x0: [2170, 64] first4: 0.036174 1.428190 0.463870 -1.927376 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 295.4 ms (295.4 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.036174 1.428190 0.463870 -1.927376 [Store] Unload DiT (1061.2 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: CUDA0 (CPU threads: 16) [VAE] Backend: CUDA0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 665 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: 616.6 ms [Debug] vae_audio: [2, 4166400] first4: 0.001009 0.001681 0.001451 0.001845 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 466 ms (186.39x realtime), 32 threads [MP3] Wrote ggml-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 10:14:05.604 | 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 10:14:07.265 | WARNING | acestep.training.trainer::40 - bitsandbytes not installed. Using standard AdamW. 2026-07-12 10:14:07.274 | 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 10:14:07.344 | 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 10:14:08.816 | 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 10:14:08.816 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:14:08.816 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:14:08.816 | 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 10:14:08.820 | 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 10:14:08.824 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:14:09.016 | 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 10:14:09.018 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:14:09.018 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:14:09.018 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:14:09.018 | 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 10:14:09.019 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:14:09.019 | 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 10:14:09.019 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:14:09.036 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:14:09.050 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:14:09.050 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:14:09.086 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 10:14:09.094 | 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 10:14:09.178 | 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 10:14:09.404 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 10:14:09.405 | 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 10:14:09.405 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.006878376007080078, 'diffusion_time_cost': 0.3108069896697998, 'diffusion_per_step_time_cost': 0.038850873708724976, 'total_time_cost': 0.3176853656768799, 'offload_time_cost': 0.0} 2026-07-12 10:14:09.419 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 10:14:09.421 | 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 10:14:09.421 | 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 10:14:09.421 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:172 - [generate_music] Using tiled VAE decode to reduce VRAM usage... 2026-07-12 10:14:09.422 | 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 10:14:09.422 | 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 10:14:09.422 | 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 10:14:09.698 | 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 10:14:09.855 | 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 10:14:09.859 | INFO | acestep.core.generation.handler.generate_music_payload:_build_generate_music_success_payload:50 - [generate_music] Done! Generated 1 audio tensors. [Request] Parsed ggml-sft/request0.json [Request] seed=42 lm_batch=1 synth_batch=1 [Request] caption: An upbeat and anthemic pop-rock track driven by bright, slig... [Request] lyrics: 490 bytes [Request] bpm=83 dur=88 key=G major ts=4 lang=fr [Request] lm: temp=0.85 cfg=2.0 top_p=0.90 top_k=0 [Request] dit: steps=50 guidance=1.0 shift=1.0 [Request] task_type: text2music [Request] solver: euler (stork_substeps=10) [Request] lm_mode: generate [Request] output_format: mp3 [Request] synth_model: acestep-v15-sft-Q5_K_M.gguf [Request] vae: vae-BF16.gguf [Request] audio_codes: (present) [Ace-Synth] Batch: 1 request(s) [Registry] Qwen3-Embedding-0.6B-BF16.gguf -> Text-Enc [Registry] Qwen3-Embedding-0.6B-Q8_0.gguf -> Text-Enc [Registry] acestep-5Hz-lm-0.6B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-0.6B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-4B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q5_K_M.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q6_K.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q8_0.gguf -> LM [Registry] acestep-v15-base-BF16.gguf -> DiT [Registry] acestep-v15-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-base-Q6_K.gguf -> DiT [Registry] acestep-v15-base-Q8_0.gguf -> DiT [Registry] acestep-v15-sft-BF16.gguf -> DiT [Registry] acestep-v15-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-BF16.gguf -> DiT [Registry] acestep-v15-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-continuous-BF16.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift1-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift3-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-base-BF16.gguf -> DiT [Registry] acestep-v15-xl-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-base-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sft-BF16.gguf -> DiT [Registry] acestep-v15-xl-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-turbo-BF16.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q8_0.gguf -> DiT [Registry] scragvae-BF16.gguf -> VAE [Registry] vae-BF16.gguf -> VAE [Store] Created (policy=STRICT) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q5_K_M.gguf: 678 tensors, data at offset 56800 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q5_K_M.gguf: 678 tensors, data at offset 56800 [Synth-Load] Ready: turbo=no, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=50, guidance=1.0, shift=1.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: 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-Q5_K_M.gguf: 678 tensors, data at offset 56800 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 140 tensors, 412.5 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L), text_proj, null_cond [Store] Load CondEnc: 264 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, 18.7 ms [Debug] enc_hidden: [238, 2048] first4: 1.760428 -0.050975 -0.131559 0.057932 [Store] Unload CondEnc (412.5 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] Detokenizer backend: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-sft-Q5_K_M.gguf: 678 tensors, data at offset 56800 [WeightCtx] Loaded 30 tensors, 73.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 49 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: 78.4 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.123923 1.460390 0.290576 -0.653908 [Store] Unload FSQ-Detok (73.2 MB) [Init-Noise Batch0] Philox noise seed=42, [2170, 64] solver=euler [Debug] noise: [2170, 64] first4: 0.194336 2.156250 -0.171875 0.847656 [Debug] context: [2170, 128] first4: -0.123923 1.460390 0.290576 -0.653908 [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-Q5_K_M.gguf: 678 tensors, data at offset 56800 [DiT] Self-attn: Q+K fused, V separate [DiT] Cross-attn: all separate [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 478 tensors, 1061.2 MB into backend [Load] DiT: 24 layers, H=2048, Nh=16/8, D=128 [Store] Load DiT: 660 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2063 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [12288] first4: 0.155849 -0.117613 -0.084447 0.082327 [Debug] temb: [2048] first4: -0.002771 -0.176872 0.004783 -0.001896 [Debug] temb_t: [2048] first4: -0.001252 0.002996 -0.012525 -0.001720 [Debug] temb_r: [2048] first4: -0.001519 -0.179868 0.017307 -0.000176 [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: -0.034977 0.028613 0.027414 -0.031142 [Debug] temb_lin1_r: [2048] first4: -0.002680 0.004202 0.000655 -0.002088 [Debug] hidden_after_proj_in: [2048, 1085] first4: -1.093960 -0.901712 0.502693 0.448468 [Debug] proj_in_input: [192, 2170] first4: -0.123923 1.460390 0.290576 -0.653908 [Debug] enc_after_cond_emb: [2048, 238] first4: -0.173654 0.836852 0.363378 -0.560410 [Debug] layer0_sa_input: [2048, 1085] first4: -0.917982 -0.702556 -0.034536 0.298304 [Debug] layer0_q_after_rope: [128, 16] first4: -0.173654 0.836852 0.363378 -0.560410 [Debug] layer0_k_after_rope: [128, 8] first4: -1.803552 -1.336632 -0.140250 0.389456 [Debug] layer0_sa_output: [2048, 1085] first4: -1.640800 0.765481 -0.571317 0.532140 [Debug] layer0_attn_out: [2048, 1085] first4: -1.803552 -1.336632 -0.140250 0.389456 [Debug] layer0_after_self_attn: [2048, 1085] first4: -1.803552 -1.336632 -0.140250 0.389456 [Debug] layer0_after_cross_attn: [2048, 1085] first4: -1.975525 -0.993182 -0.428240 0.397368 [Debug] hidden_after_layer0: [2048, 1085] first4: -9.603395 1.313110 58.217678 -1.698679 [Debug] hidden_after_layer6: [2048, 1085] first4: -15.990807 4.089139 57.017509 0.740930 [Debug] hidden_after_layer12: [2048, 1085] first4: -12.277161 -0.061552 -23.602268 0.146515 [Debug] hidden_after_layer18: [2048, 1085] first4: -4.105232 13.851622 -48.453953 -8.530821 [Debug] hidden_after_layer23: [2048, 1085] first4: 32.344223 60.453835 44.542774 -14.415376 [Debug] dit_step0_vt: [2170, 64] first4: -0.598885 2.481843 -0.200003 1.430893 [Debug] dit_step0_xt: [2170, 64] first4: 0.206314 2.106613 -0.167875 0.819038 [DiT] Step 1/50 t=1.000 ggml_backend_cuda_graph_compute: CUDA graph warmup complete [Debug] dit_step1_vt: [2170, 64] first4: -0.722226 2.546410 -0.045075 1.249809 [Debug] dit_step1_xt: [2170, 64] first4: 0.220758 2.055685 -0.166973 0.794042 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.707185 2.516510 -0.025942 1.283760 [Debug] dit_step2_xt: [2170, 64] first4: 0.234902 2.005355 -0.166455 0.768367 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.693223 2.454171 0.018907 1.305743 [Debug] dit_step3_xt: [2170, 64] first4: 0.248766 1.956271 -0.166833 0.742252 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.723299 2.364829 0.032437 1.317539 [Debug] dit_step4_xt: [2170, 64] first4: 0.263232 1.908975 -0.167481 0.715901 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.538837 2.219769 -0.010621 1.274968 [Debug] dit_step5_xt: [2170, 64] first4: 0.274009 1.864579 -0.167269 0.690402 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.482324 2.184248 -0.020701 1.273998 [Debug] dit_step6_xt: [2170, 64] first4: 0.283655 1.820894 -0.166855 0.664922 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.484739 2.150861 -0.035792 1.266761 [Debug] dit_step7_xt: [2170, 64] first4: 0.293350 1.777877 -0.166139 0.639587 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.448088 2.038056 -0.042797 1.266751 [Debug] dit_step8_xt: [2170, 64] first4: 0.302312 1.737116 -0.165283 0.614252 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.416353 1.946485 -0.016520 1.267176 [Debug] dit_step9_xt: [2170, 64] first4: 0.310639 1.698186 -0.164953 0.588908 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.371370 1.926938 -0.015435 1.248008 [Debug] dit_step10_xt: [2170, 64] first4: 0.318066 1.659647 -0.164644 0.563948 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.350025 1.819170 0.027618 1.245770 [Debug] dit_step11_xt: [2170, 64] first4: 0.325067 1.623264 -0.165197 0.539033 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.271804 1.746165 0.038049 1.246100 [Debug] dit_step12_xt: [2170, 64] first4: 0.330503 1.588341 -0.165957 0.514111 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.219700 1.640215 -0.031091 1.173178 [Debug] dit_step13_xt: [2170, 64] first4: 0.334897 1.555537 -0.165336 0.490647 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.306865 1.642073 -0.015670 1.181101 [Debug] dit_step14_xt: [2170, 64] first4: 0.341034 1.522695 -0.165022 0.467025 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.285823 1.534333 -0.060504 1.157573 [Debug] dit_step15_xt: [2170, 64] first4: 0.346751 1.492008 -0.163812 0.443874 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.312448 1.448498 -0.052756 1.150130 [Debug] dit_step16_xt: [2170, 64] first4: 0.353000 1.463038 -0.162757 0.420871 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.302144 1.380733 -0.073626 1.207761 [Debug] dit_step17_xt: [2170, 64] first4: 0.359043 1.435424 -0.161285 0.396716 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.249111 1.280910 -0.135654 1.120540 [Debug] dit_step18_xt: [2170, 64] first4: 0.364025 1.409806 -0.158571 0.374305 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.235894 1.182589 -0.104290 1.103547 [Debug] dit_step19_xt: [2170, 64] first4: 0.368743 1.386154 -0.156486 0.352234 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.303730 1.099488 -0.117972 1.073260 [Debug] dit_step20_xt: [2170, 64] first4: 0.374817 1.364164 -0.154126 0.330769 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.366585 1.053582 -0.151024 1.118861 [Debug] dit_step21_xt: [2170, 64] first4: 0.382149 1.343092 -0.151106 0.308392 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.347673 0.959593 -0.124846 1.047356 [Debug] dit_step22_xt: [2170, 64] first4: 0.389103 1.323901 -0.148609 0.287445 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.397120 0.874537 -0.136443 1.055794 [Debug] dit_step23_xt: [2170, 64] first4: 0.397045 1.306410 -0.145880 0.266329 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.396540 0.763922 -0.121717 1.007129 [Debug] dit_step24_xt: [2170, 64] first4: 0.404976 1.291131 -0.143446 0.246186 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.408251 0.670536 -0.167363 1.010911 [Debug] dit_step25_xt: [2170, 64] first4: 0.413141 1.277721 -0.140098 0.225968 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.454777 0.575630 -0.161889 0.936652 [Debug] dit_step26_xt: [2170, 64] first4: 0.422236 1.266208 -0.136861 0.207235 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.500971 0.436795 -0.106646 0.934626 [Debug] dit_step27_xt: [2170, 64] first4: 0.432256 1.257472 -0.134728 0.188542 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: -0.532059 0.323710 -0.139381 0.938730 [Debug] dit_step28_xt: [2170, 64] first4: 0.442897 1.250998 -0.131940 0.169768 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.544721 0.236158 -0.152185 0.877221 [Debug] dit_step29_xt: [2170, 64] first4: 0.453791 1.246275 -0.128896 0.152223 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: -0.547521 0.123259 -0.089830 0.811955 [Debug] dit_step30_xt: [2170, 64] first4: 0.464742 1.243810 -0.127100 0.135984 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.516022 -0.025290 -0.101965 0.836572 [Debug] dit_step31_xt: [2170, 64] first4: 0.475062 1.244316 -0.125060 0.119253 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.585766 -0.097383 -0.110544 0.790573 [Debug] dit_step32_xt: [2170, 64] first4: 0.486778 1.246263 -0.122850 0.103441 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.628357 -0.223386 -0.082620 0.798364 [Debug] dit_step33_xt: [2170, 64] first4: 0.499345 1.250731 -0.121197 0.087474 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.613702 -0.312331 -0.066017 0.772573 [Debug] dit_step34_xt: [2170, 64] first4: 0.511619 1.256977 -0.119877 0.072023 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.558879 -0.442187 -0.036877 0.726849 [Debug] dit_step35_xt: [2170, 64] first4: 0.522796 1.265821 -0.119139 0.057486 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.549527 -0.541382 -0.023357 0.734102 [Debug] dit_step36_xt: [2170, 64] first4: 0.533787 1.276649 -0.118672 0.042804 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.567534 -0.560511 0.019391 0.683110 [Debug] dit_step37_xt: [2170, 64] first4: 0.545138 1.287859 -0.119060 0.029141 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.590213 -0.706433 0.054139 0.653592 [Debug] dit_step38_xt: [2170, 64] first4: 0.556942 1.301988 -0.120143 0.016070 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.555574 -0.805554 0.094045 0.581749 [Debug] dit_step39_xt: [2170, 64] first4: 0.568053 1.318099 -0.122024 0.004435 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.587503 -0.848795 0.133121 0.576783 [Debug] dit_step40_xt: [2170, 64] first4: 0.579803 1.335075 -0.124686 -0.007101 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.566197 -0.945432 0.166331 0.510777 [Debug] dit_step41_xt: [2170, 64] first4: 0.591127 1.353983 -0.128013 -0.017317 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.499275 -1.018076 0.255062 0.512903 [Debug] dit_step42_xt: [2170, 64] first4: 0.601113 1.374345 -0.133114 -0.027575 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.539464 -1.049839 0.223339 0.398312 [Debug] dit_step43_xt: [2170, 64] first4: 0.611902 1.395342 -0.137581 -0.035541 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.519098 -1.162097 0.264549 0.363149 [Debug] dit_step44_xt: [2170, 64] first4: 0.622284 1.418584 -0.142872 -0.042804 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.497965 -1.202720 0.330253 0.349676 [Debug] dit_step45_xt: [2170, 64] first4: 0.632243 1.442638 -0.149477 -0.049797 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.460436 -1.212318 0.385633 0.278805 [Debug] dit_step46_xt: [2170, 64] first4: 0.641452 1.466884 -0.157189 -0.055374 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.428477 -1.269999 0.299510 0.191235 [Debug] dit_step47_xt: [2170, 64] first4: 0.650022 1.492284 -0.163180 -0.059198 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.443385 -1.324937 0.311900 0.221528 [Debug] dit_step48_xt: [2170, 64] first4: 0.658889 1.518783 -0.169418 -0.063629 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.434431 -1.315081 0.299059 0.157148 [Debug] dit_x0: [2170, 64] first4: 0.667578 1.545085 -0.175399 -0.066772 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 1660.0 ms (1660.0 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.667578 1.545085 -0.175399 -0.066772 [Store] Unload DiT (1061.2 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: CUDA0 (CPU threads: 16) [VAE] Backend: CUDA0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 667 ms [VAE] Tiled decode: 3 tiles (chunk=1024, overlap=64, stride=896) [VAE] Graph: 479 nodes, T_latent=960 [VAE] Upsample factor: 1920.00 (expected ~1920) [VAE] Graph: 479 nodes, T_latent=1024 [VAE] Graph: 479 nodes, T_latent=442 [VAE] Tiled decode done: 3 tiles -> T_audio=4166400 (86.80s @ 48kHz) [VAE-Decode Batch0] Decode: 616.8 ms [Debug] vae_audio: [2, 4166400] first4: -0.001445 -0.000855 -0.001125 -0.000724 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 397 ms (218.81x realtime), 32 threads [MP3] Wrote ggml-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 10:14:15.633 | 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 10:14:17.370 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:14:17.370 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:14:17.371 | 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 10:14:17.373 | 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 10:14:17.379 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:14:17.384 | 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 10:14:17.386 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:14:17.386 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:14:17.386 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:14:17.386 | 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 10:14:17.386 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:14:17.386 | 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 10:14:17.386 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:14:17.390 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:14:17.399 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:14:17.399 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:14:17.405 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... [Request] Loaded request0.json [Turbo] steps=8, shift=3.0 | acestep-v15-turbo-Q5_K_M.gguf [GGML] Running acestep-v15-turbo-Q5_K_M.gguf... [GGML] Done, 47 dump files [Python] Initializing acestep-v15-turbo... [Python] Generating (acestep-v15-turbo, 8 steps)... Using precomputed LM hints Using precomputed LM hints [Python] Wrote python-turbo/output.wav: 4166400 samples (86.80s @ 48kHz stereo) [Python] Done, 40 dump files [Turbo] Cosine similarities GGML vs Python stage GGML vs Python text_hidden 0.999797 lyric_embed 1.000000 enc_hidden 0.999029 detok_output 0.999875 context 0.999920 noise 1.000000 temb_t 0.999972 hidden_after_proj_in 0.999959 enc_after_cond_emb 0.999131 layer0_sa_output 0.999385 hidden_after_layer0 0.999814 hidden_after_layer6 0.999747 hidden_after_layer12 0.998566 hidden_after_layer18 0.995381 hidden_after_layer23 0.991251 dit_step0_vt 0.967956 dit_step0_xt 0.999930 dit_step1_vt 0.971167 dit_step1_xt 0.999784 dit_step2_vt 0.970900 dit_step2_xt 0.999387 dit_step3_vt 0.974197 dit_step3_xt 0.998558 dit_step4_vt 0.972379 dit_step4_xt 0.996671 dit_step5_vt 0.967283 dit_step5_xt 0.992226 dit_step6_vt 0.963048 dit_step6_xt 0.983539 dit_step7_vt 0.954328 dit_x0 0.970460 vae_audio 0.883568 vae_audio (STFT cosine) 0.969614 [Turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999930 0.136959 0.007850 -0.002290 0.973041 -0.002342 0.972003 dit_step1_xt 0.999784 0.260928 0.013401 -0.005310 0.942932 -0.005313 0.941730 dit_step2_xt 0.999387 0.453701 0.021139 -0.009320 0.909560 -0.009311 0.908527 dit_step3_xt 0.998558 0.682602 0.031120 -0.014717 0.874310 -0.014577 0.873624 dit_step4_xt 0.996671 1.021173 0.045231 -0.021938 0.842541 -0.021660 0.841995 dit_step5_xt 0.992226 1.481974 0.067165 -0.032340 0.826100 -0.032109 0.824593 dit_step6_xt 0.983539 2.115583 0.100745 -0.046905 0.857373 -0.046482 0.855546 [SFT] steps=50, shift=1.0 | acestep-v15-sft-Q5_K_M.gguf [GGML] Running acestep-v15-sft-Q5_K_M.gguf... [GGML] Done, 131 dump files [Python] Initializing acestep-v15-sft... [Python] Generating (acestep-v15-sft, 50 steps)... Using precomputed LM hints Using precomputed LM hints 0%| | 0/50 [00:00 Text-Enc [Registry] Qwen3-Embedding-0.6B-Q8_0.gguf -> Text-Enc [Registry] acestep-5Hz-lm-0.6B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-0.6B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-4B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q5_K_M.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q6_K.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q8_0.gguf -> LM [Registry] acestep-v15-base-BF16.gguf -> DiT [Registry] acestep-v15-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-base-Q6_K.gguf -> DiT [Registry] acestep-v15-base-Q8_0.gguf -> DiT [Registry] acestep-v15-sft-BF16.gguf -> DiT [Registry] acestep-v15-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-BF16.gguf -> DiT [Registry] acestep-v15-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-continuous-BF16.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift1-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift3-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-base-BF16.gguf -> DiT [Registry] acestep-v15-xl-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-base-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sft-BF16.gguf -> DiT [Registry] acestep-v15-xl-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-turbo-BF16.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q8_0.gguf -> DiT [Registry] scragvae-BF16.gguf -> VAE [Registry] vae-BF16.gguf -> VAE [Store] Created (policy=STRICT) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q5_K_M.gguf: 830 tensors, data at offset 69088 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q5_K_M.gguf: 830 tensors, data at offset 69088 [Synth-Load] Ready: turbo=yes, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=8, guidance=1.0, shift=3.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: 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: 273 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-Q5_K_M.gguf: 830 tensors, data at offset 69088 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 141 tensors, 416.2 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L, CLS), text_proj, null_cond [Store] Load CondEnc: 263 ms [CondEnc] Lyric sliding mask: 167x167, window=128 [CondEnc] Timbre sliding mask: 2x2, window=128 (CLS) [Encode] Packed: lyric=167 + timbre=1 + text=70 = 238 tokens [Encode-Text Batch0] 70+167 tokens -> enc_S=238, 17.9 ms [Debug] enc_hidden: [238, 2048] first4: 0.687547 -0.019583 -0.050590 0.023056 [Store] Unload CondEnc (416.2 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] Detokenizer backend: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-turbo-Q5_K_M.gguf: 830 tensors, data at offset 69088 [WeightCtx] Loaded 30 tensors, 73.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 48 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: 83.0 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.123923 1.460390 0.290576 -0.653908 [Store] Unload FSQ-Detok (73.2 MB) [Init-Noise Batch0] Philox noise seed=42, [2170, 64] solver=euler [Debug] noise: [2170, 64] first4: 0.194336 2.156250 -0.171875 0.847656 [Debug] context: [2170, 128] first4: -0.123923 1.460390 0.290576 -0.653908 [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-Q5_K_M.gguf: 830 tensors, data at offset 69088 [DiT] Self-attn: Q+K fused, V separate [DiT] Cross-attn: all separate [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 630 tensors, 2801.1 MB into backend [Load] DiT: 32 layers, H=2560, Nh=32/8, D=128 [Store] Load DiT: 1664 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2729 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [15360] first4: 0.007712 0.102719 0.074757 -0.076147 [Debug] temb: [2560] first4: 0.026508 -0.023413 -0.016350 0.047091 [Debug] temb_t: [2560] first4: 0.014671 -0.022221 -0.020670 0.048385 [Debug] temb_r: [2560] first4: 0.011837 -0.001192 0.004320 -0.001294 [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: 0.063001 -0.019660 0.138829 -0.115781 [Debug] temb_lin1_r: [2560] first4: -0.008961 0.017831 -0.016913 0.012843 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.345921 -0.578886 0.249678 -0.175482 [Debug] proj_in_input: [192, 2170] first4: -0.123923 1.460390 0.290576 -0.653908 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.044055 0.663473 0.113305 0.875277 [Debug] layer0_sa_input: [2560, 1085] first4: 0.031957 -0.052276 0.104071 -0.132012 [Debug] layer0_q_after_rope: [128, 32] first4: -0.852972 -2.203073 -2.025851 0.596684 [Debug] layer0_k_after_rope: [128, 8] first4: 1.636017 4.264533 -0.529837 5.329281 [Debug] layer0_sa_output: [2560, 1085] first4: 1.636017 4.264533 -0.529837 5.329281 [Debug] layer0_attn_out: [4096, 1085] first4: -0.852972 -2.203073 -2.025851 0.596684 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.473983 -2.235947 -0.138305 -2.722099 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.348925 -1.977113 0.004606 -3.476543 [Debug] hidden_after_layer0: [2560, 1085] first4: 2.072834 -2.354451 234.576370 -3.456183 [Debug] hidden_after_layer6: [2560, 1085] first4: -6.698165 -2.014523 313.496613 -12.148424 [Debug] hidden_after_layer12: [2560, 1085] first4: -13.566050 28.171444 259.746735 27.469402 [Debug] hidden_after_layer18: [2560, 1085] first4: 48.342758 21.642435 261.330566 9.697443 [Debug] hidden_after_layer31: [2560, 1085] first4: -116.185379 -200.565689 -105.924301 79.352737 [Debug] dit_step0_vt: [2170, 64] first4: -1.426639 1.935840 0.583195 1.323876 [Debug] dit_step0_xt: [2170, 64] first4: 0.259183 2.068257 -0.198384 0.787480 [DiT] Step 1/8 t=1.000 ggml_backend_cuda_graph_compute: CUDA graph warmup complete [Debug] dit_step1_vt: [2170, 64] first4: -0.723360 2.497524 0.794920 0.986940 [Debug] dit_step1_xt: [2170, 64] first4: 0.298639 1.932029 -0.241743 0.733647 [DiT] Step 2/8 t=0.955 [Debug] dit_step2_vt: [2170, 64] first4: -0.727988 2.024858 1.146908 0.979205 [Debug] dit_step2_xt: [2170, 64] first4: 0.347172 1.797038 -0.318204 0.668367 [DiT] Step 3/8 t=0.900 [Debug] dit_step3_vt: [2170, 64] first4: -1.407346 1.878366 1.589833 1.218770 [Debug] dit_step3_xt: [2170, 64] first4: 0.464451 1.640508 -0.450690 0.566803 [DiT] Step 4/8 t=0.833 [Debug] dit_step4_vt: [2170, 64] first4: -0.885981 2.362860 1.037356 0.232174 [Debug] dit_step4_xt: [2170, 64] first4: 0.559377 1.387344 -0.561835 0.541927 [DiT] Step 5/8 t=0.750 [Debug] dit_step5_vt: [2170, 64] first4: -1.191545 2.296031 1.274162 1.351497 [Debug] dit_step5_xt: [2170, 64] first4: 0.729598 1.059340 -0.743858 0.348856 [DiT] Step 6/8 t=0.643 [Debug] dit_step6_vt: [2170, 64] first4: -1.225613 2.462377 0.761850 0.810354 [Debug] dit_step6_xt: [2170, 64] first4: 0.974720 0.566864 -0.896228 0.186785 [DiT] Step 7/8 t=0.500 [Debug] dit_step7_vt: [2170, 64] first4: -1.124249 1.483996 0.739541 0.441340 [Debug] dit_x0: [2170, 64] first4: 1.311995 0.121666 -1.118091 0.054383 [DiT] Step 8/8 t=0.300 [DiT-Generate] Total: 681.4 ms (681.4 ms/sample) [Debug] dit_output: [2170, 64] first4: 1.311995 0.121666 -1.118091 0.054383 [Store] Unload DiT (2801.1 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: CUDA0 (CPU threads: 16) [VAE] Backend: CUDA0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 670 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: 642.2 ms [Debug] vae_audio: [2, 4166400] first4: 0.000463 0.003677 0.004036 0.003027 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 471 ms (184.34x realtime), 32 threads [MP3] Wrote ggml-xl-turbo/request00.mp3 [Ace-Synth] All done 2026-07-12 10:14:25.544 | 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 0.999029 detok_output 0.999875 context 0.999920 noise 1.000000 temb_t 0.999900 hidden_after_proj_in 0.999960 enc_after_cond_emb 0.999131 layer0_sa_output 0.999446 hidden_after_layer0 0.999861 hidden_after_layer6 0.999571 hidden_after_layer12 0.998959 hidden_after_layer18 0.997911 hidden_after_layer23 0.998400 dit_step0_vt 0.998296 dit_step0_xt 1.000000 dit_step5_vt 0.998409 dit_step5_xt 0.999965 dit_step10_vt 0.997636 dit_step10_xt 0.999871 dit_step15_vt 0.995434 dit_step15_xt 0.999643 dit_step20_vt 0.991571 dit_step20_xt 0.999075 dit_step25_vt 0.985835 dit_step25_xt 0.997958 dit_step30_vt 0.978792 dit_step30_xt 0.996080 dit_step35_vt 0.971572 dit_step35_xt 0.993649 dit_step40_vt 0.967290 dit_step40_xt 0.991194 dit_step45_vt 0.972670 dit_step45_xt 0.989465 dit_step49_vt 0.981039 dit_x0 0.988884 vae_audio 0.950679 vae_audio (STFT cosine) 0.977855 [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 0.018260 0.001605 -0.001650 0.978544 -0.001619 0.978872 dit_step5_xt 0.999965 0.069438 0.005877 -0.008356 0.885268 -0.008480 0.883811 dit_step10_xt 0.999871 0.106721 0.009776 -0.015227 0.801656 -0.015075 0.802324 dit_step15_xt 0.999643 0.188042 0.014522 -0.022169 0.731467 -0.021933 0.731304 dit_step20_xt 0.999075 0.382507 0.021081 -0.029219 0.678993 -0.028882 0.679066 dit_step25_xt 0.997958 0.641991 0.029315 -0.036369 0.648507 -0.035643 0.650780 dit_step30_xt 0.996080 0.921765 0.039877 -0.043566 0.643327 -0.042577 0.647107 dit_step35_xt 0.993649 1.218068 0.052268 -0.050739 0.664237 -0.049471 0.669256 dit_step40_xt 0.991194 1.497953 0.065930 -0.057893 0.709005 -0.056532 0.715783 dit_step45_xt 0.989465 1.740591 0.078873 -0.065090 0.773289 -0.063533 0.780575 [Xl-turbo] steps=8, shift=3.0 | acestep-v15-xl-turbo-Q5_K_M.gguf [GGML] Running acestep-v15-xl-turbo-Q5_K_M.gguf... [GGML] Done, 47 dump files [Python] Initializing acestep-v15-xl-turbo... Loading checkpoint shards: 0%| | 0/4 [00:00 1.0 (turbo does not use CFG). 2026-07-12 10:14:28.912 | INFO | acestep.core.generation.handler.generate_music:generate_music:304 - [generate_music] Starting generation... 2026-07-12 10:14:28.912 | INFO | acestep.core.generation.handler.generate_music:generate_music:307 - [generate_music] Preparing inputs... 2026-07-12 10:14:28.913 | 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 10:14:28.917 | INFO | acestep.core.generation.handler.generate_music:_vram_preflight_check:156 - [generate_music] VRAM pre-flight: 76.75 GB free, ~1.23 GB needed (batch=1, duration=88s, mode=xl_turbo). 2026-07-12 10:14:28.921 | INFO | acestep.core.generation.handler.conditioning_target:_prepare_target_latents_and_wavs:78 - [generate_music] Decoding audio codes for item 0... 2026-07-12 10:14:28.927 | 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 10:14:28.928 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:119 - ====================================================================== 2026-07-12 10:14:28.928 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:120 - πŸ” [DEBUG] DiT TEXT ENCODER INPUT (Inference) 2026-07-12 10:14:28.928 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:121 - ====================================================================== 2026-07-12 10:14:28.928 | 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 10:14:28.928 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:14:28.928 | 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 10:14:28.929 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:14:28.932 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:14:28.940 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:14:28.940 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:14:28.946 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:263 - [service_generate] DiT diffusion via PyTorch (cuda)... 2026-07-12 10:14:28.953 | 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 10:14:29.525 | INFO | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:42 - [generate_music] Model generation completed. Decoding latents... 2026-07-12 10:14:29.526 | 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 10:14:29.526 | DEBUG | acestep.core.generation.handler.generate_music_decode:_prepare_generate_music_decode_state:65 - [generate_music] time_costs: {'encoder_time_cost': 0.006379127502441406, 'diffusion_time_cost': 0.5721137523651123, 'diffusion_per_step_time_cost': 0.07151421904563904, 'total_time_cost': 0.5784928798675537, 'offload_time_cost': 0.0} 2026-07-12 10:14:29.549 | INFO | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:127 - [generate_music] Decoding latents with VAE... 2026-07-12 10:14:29.552 | DEBUG | acestep.core.generation.handler.generate_music_decode:_decode_generate_music_pred_latents:136 - [generate_music] Before VAE decode: allocated=17.04GB, max=18.34GB 2026-07-12 10:14:29.552 | 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 10:14:29.552 | 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 10:14:29.552 | 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 10:14:29.552 | 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 10:14:29.552 | 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 10:14:29.829 | 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 10:14:30.033 | 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 10:14:30.036 | INFO | acestep.core.generation.handler.generate_music_payload:_build_generate_music_success_payload:50 - [generate_music] Done! Generated 1 audio tensors. [Request] Parsed ggml-xl-sft/request0.json [Request] seed=42 lm_batch=1 synth_batch=1 [Request] caption: An upbeat and anthemic pop-rock track driven by bright, slig... [Request] lyrics: 490 bytes [Request] bpm=83 dur=88 key=G major ts=4 lang=fr [Request] lm: temp=0.85 cfg=2.0 top_p=0.90 top_k=0 [Request] dit: steps=50 guidance=1.0 shift=1.0 [Request] task_type: text2music [Request] solver: euler (stork_substeps=10) [Request] lm_mode: generate [Request] output_format: mp3 [Request] synth_model: acestep-v15-xl-sft-Q5_K_M.gguf [Request] vae: vae-BF16.gguf [Request] audio_codes: (present) [Ace-Synth] Batch: 1 request(s) [Registry] Qwen3-Embedding-0.6B-BF16.gguf -> Text-Enc [Registry] Qwen3-Embedding-0.6B-Q8_0.gguf -> Text-Enc [Registry] acestep-5Hz-lm-0.6B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-0.6B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-1.7B-Q8_0.gguf -> LM [Registry] acestep-5Hz-lm-4B-BF16.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q5_K_M.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q6_K.gguf -> LM [Registry] acestep-5Hz-lm-4B-Q8_0.gguf -> LM [Registry] acestep-v15-base-BF16.gguf -> DiT [Registry] acestep-v15-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-base-Q6_K.gguf -> DiT [Registry] acestep-v15-base-Q8_0.gguf -> DiT [Registry] acestep-v15-sft-BF16.gguf -> DiT [Registry] acestep-v15-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-BF16.gguf -> DiT [Registry] acestep-v15-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-continuous-BF16.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-continuous-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift1-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift1-Q8_0.gguf -> DiT [Registry] acestep-v15-turbo-shift3-BF16.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q4_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q5_K_M.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q6_K.gguf -> DiT [Registry] acestep-v15-turbo-shift3-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-base-BF16.gguf -> DiT [Registry] acestep-v15-xl-base-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-base-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-base-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sft-BF16.gguf -> DiT [Registry] acestep-v15-xl-sft-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sft-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sft-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-BF16.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-sftturbo50-Q8_0.gguf -> DiT [Registry] acestep-v15-xl-turbo-BF16.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q4_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q5_K_M.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q6_K.gguf -> DiT [Registry] acestep-v15-xl-turbo-Q8_0.gguf -> DiT [Registry] scragvae-BF16.gguf -> VAE [Registry] vae-BF16.gguf -> VAE [Store] Created (policy=STRICT) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q5_K_M.gguf: 830 tensors, data at offset 69056 [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q5_K_M.gguf: 830 tensors, data at offset 69056 [Synth-Load] Ready: turbo=no, fa=no, batch_cfg=yes [Resolve-Params] max audio codes across batch: 434 (86.8s @ 5Hz) [Resolve-T] T=2170, S=1085 [Resolve-T] seed=42, steps=50, guidance=1.0, shift=1.0, duration=88.0s [BPE] Loaded from GGUF: 151643 vocab, 151387 merges ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB): Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so ggml_vulkan: Found 1 Vulkan devices: ggml_vulkan: 0 = NVIDIA RTX PRO 6000 Blackwell Workstation Edition (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2 load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] TextEncoder backend: 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: 279 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-Q5_K_M.gguf: 830 tensors, data at offset 69056 [Load] LyricEncoder: 8L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [Load] TimbreEncoder: 4L [Qwen3] Attn: Q+K fused, V separate [Qwen3] MLP: gate+up fused [WeightCtx] Loaded 141 tensors, 416.2 MB into backend [Load] CondEncoder: lyric(8L), timbre(4L, CLS), text_proj, null_cond [Store] Load CondEnc: 277 ms [CondEnc] Lyric sliding mask: 167x167, window=128 [CondEnc] Timbre sliding mask: 2x2, window=128 (CLS) [Encode] Packed: lyric=167 + timbre=1 + text=70 = 238 tokens [Encode-Text Batch0] 70+167 tokens -> enc_S=238, 18.9 ms [Debug] enc_hidden: [238, 2048] first4: 0.687547 -0.019583 -0.050590 0.023056 [Store] Unload CondEnc (416.2 MB) load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] Detokenizer backend: CUDA0 (CPU threads: 16) [GGUF] /mnt/workspace/git/acestep.cpp/models/acestep-v15-xl-sft-Q5_K_M.gguf: 830 tensors, data at offset 69056 [WeightCtx] Loaded 30 tensors, 73.2 MB into backend [Load] Detokenizer: FSQ(6->2048) + 2L encoder(S=5, 2048->64) [Store] Load FSQ-Detok: 51 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: 78.5 ms, 434 codes [Debug] detok_output: [2170, 64] first4: -0.123923 1.460390 0.290576 -0.653908 [Store] Unload FSQ-Detok (73.2 MB) [Init-Noise Batch0] Philox noise seed=42, [2170, 64] solver=euler [Debug] noise: [2170, 64] first4: 0.194336 2.156250 -0.171875 0.847656 [Debug] context: [2170, 128] first4: -0.123923 1.460390 0.290576 -0.653908 [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-Q5_K_M.gguf: 830 tensors, data at offset 69056 [DiT] Self-attn: Q+K fused, V separate [DiT] Cross-attn: all separate [DiT] MLP: gate+up fused [Load] null_condition_emb found (CFG available) [WeightCtx] Loaded 630 tensors, 2801.1 MB into backend [Load] DiT: 32 layers, H=2560, Nh=32/8, D=128 [Store] Load DiT: 1723 ms [DiT] Batch N=1, T=2170, S=1085, enc_S=238 [DiT] Graph: 2729 nodes [DiT] Solver: ODE Euler (1 NFE/step, order 1) [Debug] tproj: [15360] first4: 0.004657 0.088920 0.077683 -0.035812 [Debug] temb: [2560] first4: 0.009920 -0.007980 -0.006100 0.004947 [Debug] temb_t: [2560] first4: 0.008431 -0.008081 -0.006747 0.003949 [Debug] temb_r: [2560] first4: 0.001489 0.000101 0.000647 0.000998 [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: 0.028321 -0.080133 0.086678 -0.070687 [Debug] temb_lin1_r: [2560] first4: 0.006506 -0.006322 -0.000918 -0.013459 [Debug] hidden_after_proj_in: [2560, 1085] first4: 0.335268 -0.494534 0.121045 -0.198807 [Debug] proj_in_input: [192, 2170] first4: -0.123923 1.460390 0.290576 -0.653908 [Debug] enc_after_cond_emb: [2560, 238] first4: 0.055182 0.655642 0.120549 0.870686 [Debug] layer0_sa_input: [2560, 1085] first4: 0.029160 -0.036116 0.103694 -0.096734 [Debug] layer0_q_after_rope: [128, 32] first4: -0.879093 -2.053107 -2.858946 0.777861 [Debug] layer0_k_after_rope: [128, 8] first4: 2.007242 4.405107 -0.766803 7.804473 [Debug] layer0_sa_output: [2560, 1085] first4: 2.007242 4.405107 -0.766803 7.804473 [Debug] layer0_attn_out: [4096, 1085] first4: -0.879093 -2.053107 -2.858946 0.777861 [Debug] layer0_after_self_attn: [2560, 1085] first4: 0.461472 -2.190605 -0.445395 -4.132349 [Debug] layer0_after_cross_attn: [2560, 1085] first4: 0.291429 -1.916080 -0.371009 -4.176170 [Debug] hidden_after_layer0: [2560, 1085] first4: -1.419206 -1.269824 243.988464 -4.070871 [Debug] hidden_after_layer6: [2560, 1085] first4: -7.385107 3.258244 303.973602 -21.687317 [Debug] hidden_after_layer12: [2560, 1085] first4: -20.920511 37.196354 222.099899 15.860222 [Debug] hidden_after_layer18: [2560, 1085] first4: 12.357943 -16.971991 227.058609 19.074316 [Debug] hidden_after_layer31: [2560, 1085] first4: 64.164536 -115.760475 -7.805725 86.182961 [Debug] dit_step0_vt: [2170, 64] first4: -0.353781 2.234020 0.056670 1.716218 [Debug] dit_step0_xt: [2170, 64] first4: 0.201412 2.111570 -0.173008 0.813332 [DiT] Step 1/50 t=1.000 ggml_backend_cuda_graph_compute: CUDA graph warmup complete [Debug] dit_step1_vt: [2170, 64] first4: -0.379588 2.197823 0.105507 1.541791 [Debug] dit_step1_xt: [2170, 64] first4: 0.209003 2.067613 -0.175119 0.782496 [DiT] Step 2/50 t=0.980 [Debug] dit_step2_vt: [2170, 64] first4: -0.361178 2.192315 0.104834 1.558212 [Debug] dit_step2_xt: [2170, 64] first4: 0.216227 2.023767 -0.177215 0.751332 [DiT] Step 3/50 t=0.960 [Debug] dit_step3_vt: [2170, 64] first4: -0.319306 2.143630 0.078354 1.564876 [Debug] dit_step3_xt: [2170, 64] first4: 0.222613 1.980894 -0.178782 0.720034 [DiT] Step 4/50 t=0.940 [Debug] dit_step4_vt: [2170, 64] first4: -0.313112 2.073701 0.056295 1.556895 [Debug] dit_step4_xt: [2170, 64] first4: 0.228875 1.939420 -0.179908 0.688896 [DiT] Step 5/50 t=0.920 [Debug] dit_step5_vt: [2170, 64] first4: -0.296990 2.056541 0.083264 1.592337 [Debug] dit_step5_xt: [2170, 64] first4: 0.234815 1.898289 -0.181574 0.657050 [DiT] Step 6/50 t=0.900 [Debug] dit_step6_vt: [2170, 64] first4: -0.297873 2.035826 0.091434 1.615180 [Debug] dit_step6_xt: [2170, 64] first4: 0.240773 1.857573 -0.183402 0.624746 [DiT] Step 7/50 t=0.880 [Debug] dit_step7_vt: [2170, 64] first4: -0.287747 1.989983 0.146563 1.613355 [Debug] dit_step7_xt: [2170, 64] first4: 0.246527 1.817773 -0.186333 0.592479 [DiT] Step 8/50 t=0.860 [Debug] dit_step8_vt: [2170, 64] first4: -0.308704 1.922363 0.183966 1.618611 [Debug] dit_step8_xt: [2170, 64] first4: 0.252702 1.779326 -0.190013 0.560107 [DiT] Step 9/50 t=0.840 [Debug] dit_step9_vt: [2170, 64] first4: -0.342480 1.874891 0.148204 1.581906 [Debug] dit_step9_xt: [2170, 64] first4: 0.259551 1.741828 -0.192977 0.528469 [DiT] Step 10/50 t=0.820 [Debug] dit_step10_vt: [2170, 64] first4: -0.347371 1.810045 0.143854 1.569329 [Debug] dit_step10_xt: [2170, 64] first4: 0.266499 1.705627 -0.195854 0.497082 [DiT] Step 11/50 t=0.800 [Debug] dit_step11_vt: [2170, 64] first4: -0.351709 1.759107 0.114552 1.556755 [Debug] dit_step11_xt: [2170, 64] first4: 0.273533 1.670445 -0.198145 0.465947 [DiT] Step 12/50 t=0.780 [Debug] dit_step12_vt: [2170, 64] first4: -0.343074 1.718905 0.097588 1.542628 [Debug] dit_step12_xt: [2170, 64] first4: 0.280394 1.636067 -0.200097 0.435094 [DiT] Step 13/50 t=0.760 [Debug] dit_step13_vt: [2170, 64] first4: -0.345222 1.667212 0.114111 1.544209 [Debug] dit_step13_xt: [2170, 64] first4: 0.287299 1.602723 -0.202379 0.404210 [DiT] Step 14/50 t=0.740 [Debug] dit_step14_vt: [2170, 64] first4: -0.334928 1.618853 0.070356 1.531120 [Debug] dit_step14_xt: [2170, 64] first4: 0.293997 1.570346 -0.203786 0.373588 [DiT] Step 15/50 t=0.720 [Debug] dit_step15_vt: [2170, 64] first4: -0.320637 1.574285 0.059569 1.525364 [Debug] dit_step15_xt: [2170, 64] first4: 0.300410 1.538860 -0.204977 0.343081 [DiT] Step 16/50 t=0.700 [Debug] dit_step16_vt: [2170, 64] first4: -0.285145 1.524498 0.033380 1.502578 [Debug] dit_step16_xt: [2170, 64] first4: 0.306113 1.508370 -0.205645 0.313029 [DiT] Step 17/50 t=0.680 [Debug] dit_step17_vt: [2170, 64] first4: -0.251070 1.490252 0.042136 1.502482 [Debug] dit_step17_xt: [2170, 64] first4: 0.311134 1.478565 -0.206488 0.282979 [DiT] Step 18/50 t=0.660 [Debug] dit_step18_vt: [2170, 64] first4: -0.222496 1.432354 0.009511 1.486087 [Debug] dit_step18_xt: [2170, 64] first4: 0.315584 1.449918 -0.206678 0.253258 [DiT] Step 19/50 t=0.640 [Debug] dit_step19_vt: [2170, 64] first4: -0.181504 1.371475 -0.031198 1.482056 [Debug] dit_step19_xt: [2170, 64] first4: 0.319214 1.422488 -0.206054 0.223616 [DiT] Step 20/50 t=0.620 [Debug] dit_step20_vt: [2170, 64] first4: -0.148796 1.328794 -0.020605 1.468277 [Debug] dit_step20_xt: [2170, 64] first4: 0.322190 1.395912 -0.205642 0.194251 [DiT] Step 21/50 t=0.600 [Debug] dit_step21_vt: [2170, 64] first4: -0.121421 1.285650 -0.049342 1.450442 [Debug] dit_step21_xt: [2170, 64] first4: 0.324619 1.370199 -0.204655 0.165242 [DiT] Step 22/50 t=0.580 [Debug] dit_step22_vt: [2170, 64] first4: -0.089767 1.256400 -0.069547 1.439301 [Debug] dit_step22_xt: [2170, 64] first4: 0.326414 1.345071 -0.203264 0.136456 [DiT] Step 23/50 t=0.560 [Debug] dit_step23_vt: [2170, 64] first4: -0.061627 1.186475 -0.058876 1.411836 [Debug] dit_step23_xt: [2170, 64] first4: 0.327646 1.321342 -0.202087 0.108219 [DiT] Step 24/50 t=0.540 [Debug] dit_step24_vt: [2170, 64] first4: -0.030048 1.147390 -0.044150 1.389533 [Debug] dit_step24_xt: [2170, 64] first4: 0.328247 1.298394 -0.201204 0.080429 [DiT] Step 25/50 t=0.520 [Debug] dit_step25_vt: [2170, 64] first4: -0.009638 1.109962 -0.079680 1.368076 [Debug] dit_step25_xt: [2170, 64] first4: 0.328440 1.276195 -0.199610 0.053067 [DiT] Step 26/50 t=0.500 [Debug] dit_step26_vt: [2170, 64] first4: -0.005787 1.046394 -0.083872 1.335997 [Debug] dit_step26_xt: [2170, 64] first4: 0.328556 1.255267 -0.197933 0.026347 [DiT] Step 27/50 t=0.480 [Debug] dit_step27_vt: [2170, 64] first4: -0.011942 0.961428 -0.087800 1.328463 [Debug] dit_step27_xt: [2170, 64] first4: 0.328795 1.236038 -0.196177 -0.000222 [DiT] Step 28/50 t=0.460 [Debug] dit_step28_vt: [2170, 64] first4: 0.008341 0.935667 -0.174369 1.287243 [Debug] dit_step28_xt: [2170, 64] first4: 0.328628 1.217325 -0.192689 -0.025967 [DiT] Step 29/50 t=0.440 [Debug] dit_step29_vt: [2170, 64] first4: -0.006268 0.814412 -0.154866 1.289366 [Debug] dit_step29_xt: [2170, 64] first4: 0.328753 1.201037 -0.189592 -0.051754 [DiT] Step 30/50 t=0.420 [Debug] dit_step30_vt: [2170, 64] first4: 0.002100 0.729612 -0.203881 1.280860 [Debug] dit_step30_xt: [2170, 64] first4: 0.328711 1.186445 -0.185514 -0.077371 [DiT] Step 31/50 t=0.400 [Debug] dit_step31_vt: [2170, 64] first4: -0.028591 0.717589 -0.166166 1.230846 [Debug] dit_step31_xt: [2170, 64] first4: 0.329283 1.172093 -0.182191 -0.101988 [DiT] Step 32/50 t=0.380 [Debug] dit_step32_vt: [2170, 64] first4: -0.067151 0.586936 -0.156711 1.183615 [Debug] dit_step32_xt: [2170, 64] first4: 0.330626 1.160354 -0.179057 -0.125661 [DiT] Step 33/50 t=0.360 [Debug] dit_step33_vt: [2170, 64] first4: -0.034281 0.525394 -0.203134 1.195561 [Debug] dit_step33_xt: [2170, 64] first4: 0.331312 1.149846 -0.174994 -0.149572 [DiT] Step 34/50 t=0.340 [Debug] dit_step34_vt: [2170, 64] first4: -0.068247 0.452481 -0.193849 1.137017 [Debug] dit_step34_xt: [2170, 64] first4: 0.332677 1.140797 -0.171117 -0.172312 [DiT] Step 35/50 t=0.320 [Debug] dit_step35_vt: [2170, 64] first4: -0.090713 0.342806 -0.182613 1.102554 [Debug] dit_step35_xt: [2170, 64] first4: 0.334491 1.133940 -0.167465 -0.194363 [DiT] Step 36/50 t=0.300 [Debug] dit_step36_vt: [2170, 64] first4: -0.108528 0.253984 -0.247407 1.080858 [Debug] dit_step36_xt: [2170, 64] first4: 0.336661 1.128861 -0.162517 -0.215980 [DiT] Step 37/50 t=0.280 [Debug] dit_step37_vt: [2170, 64] first4: -0.130257 0.144284 -0.173920 1.054290 [Debug] dit_step37_xt: [2170, 64] first4: 0.339267 1.125975 -0.159038 -0.237066 [DiT] Step 38/50 t=0.260 [Debug] dit_step38_vt: [2170, 64] first4: -0.162549 -0.014001 -0.155289 1.020695 [Debug] dit_step38_xt: [2170, 64] first4: 0.342518 1.126255 -0.155933 -0.257480 [DiT] Step 39/50 t=0.240 [Debug] dit_step39_vt: [2170, 64] first4: -0.174740 -0.101155 -0.121567 0.994627 [Debug] dit_step39_xt: [2170, 64] first4: 0.346012 1.128278 -0.153501 -0.277373 [DiT] Step 40/50 t=0.220 [Debug] dit_step40_vt: [2170, 64] first4: -0.159963 -0.172928 -0.122356 0.942750 [Debug] dit_step40_xt: [2170, 64] first4: 0.349212 1.131737 -0.151054 -0.296228 [DiT] Step 41/50 t=0.200 [Debug] dit_step41_vt: [2170, 64] first4: -0.241654 -0.290387 -0.117310 0.874794 [Debug] dit_step41_xt: [2170, 64] first4: 0.354045 1.137545 -0.148708 -0.313724 [DiT] Step 42/50 t=0.180 [Debug] dit_step42_vt: [2170, 64] first4: -0.246419 -0.352279 -0.088571 0.803472 [Debug] dit_step42_xt: [2170, 64] first4: 0.358973 1.144590 -0.146936 -0.329793 [DiT] Step 43/50 t=0.160 [Debug] dit_step43_vt: [2170, 64] first4: -0.242872 -0.505493 -0.054259 0.754805 [Debug] dit_step43_xt: [2170, 64] first4: 0.363831 1.154700 -0.145851 -0.344889 [DiT] Step 44/50 t=0.140 [Debug] dit_step44_vt: [2170, 64] first4: -0.277420 -0.569724 -0.018621 0.691538 [Debug] dit_step44_xt: [2170, 64] first4: 0.369379 1.166095 -0.145479 -0.358720 [DiT] Step 45/50 t=0.120 [Debug] dit_step45_vt: [2170, 64] first4: -0.299468 -0.641958 0.033377 0.651296 [Debug] dit_step45_xt: [2170, 64] first4: 0.375368 1.178934 -0.146146 -0.371746 [DiT] Step 46/50 t=0.100 [Debug] dit_step46_vt: [2170, 64] first4: -0.321493 -0.717342 0.047816 0.602033 [Debug] dit_step46_xt: [2170, 64] first4: 0.381798 1.193281 -0.147103 -0.383786 [DiT] Step 47/50 t=0.080 [Debug] dit_step47_vt: [2170, 64] first4: -0.401090 -0.787850 0.094034 0.561207 [Debug] dit_step47_xt: [2170, 64] first4: 0.389820 1.209038 -0.148983 -0.395011 [DiT] Step 48/50 t=0.060 [Debug] dit_step48_vt: [2170, 64] first4: -0.446572 -0.841793 0.217490 0.533705 [Debug] dit_step48_xt: [2170, 64] first4: 0.398751 1.225873 -0.153333 -0.405685 [DiT] Step 49/50 t=0.040 [Debug] dit_step49_vt: [2170, 64] first4: -0.428528 -0.833031 0.146004 0.550341 [Debug] dit_x0: [2170, 64] first4: 0.407322 1.242534 -0.156253 -0.416692 [DiT] Step 50/50 t=0.020 [DiT-Generate] Total: 4037.2 ms (4037.2 ms/sample) [Debug] dit_output: [2170, 64] first4: 0.407322 1.242534 -0.156253 -0.416692 [Store] Unload DiT (2801.1 MB) [GGUF] /mnt/workspace/git/acestep.cpp/models/vae-BF16.gguf: 365 tensors, data at offset 30048 load_backend: loaded CUDA backend from /mnt/workspace/git/acestep.cpp/build/libggml-cuda.so load_backend: loaded Vulkan backend from /mnt/workspace/git/acestep.cpp/build/libggml-vulkan.so load_backend: loaded CPU backend from /mnt/workspace/git/acestep.cpp/build/libggml-cpu-zen4.so [Load] VAE backend: CUDA0 (CPU threads: 16) [VAE] Backend: CUDA0, Weight buffer: 161.1 MB [VAE] Loaded: 5 blocks, upsample=1920x, F32 activations [Store] Load VAE-Dec: 670 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: 615.6 ms [Debug] vae_audio: [2, 4166400] first4: -0.001460 -0.000821 -0.000808 -0.000309 [Store] Unload VAE-Dec (161.1 MB) [MP3] Encoding 86.8s @ 128 kbps, 48000 Hz stereo [MP3] 1388928 bytes (12.0:1), 408 ms (212.60x realtime), 32 threads [MP3] Wrote ggml-xl-sft/request00.mp3 [Ace-Synth] All done 2026-07-12 10:14:39.337 | 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 0.998790 detok_output 0.999875 context 0.999920 noise 1.000000 temb_t 0.999914 hidden_after_proj_in 0.999941 enc_after_cond_emb 0.998676 layer0_sa_output 0.998951 hidden_after_layer0 0.999330 hidden_after_layer6 0.999694 hidden_after_layer12 0.996782 hidden_after_layer18 0.995998 hidden_after_layer31 0.988017 dit_step0_vt 0.912695 dit_step0_xt 0.999842 dit_step1_vt 0.925523 dit_step1_xt 0.999523 dit_step2_vt 0.925170 dit_step2_xt 0.998706 dit_step3_vt 0.922726 dit_step3_xt 0.996784 dit_step4_vt 0.913730 dit_step4_xt 0.991901 dit_step5_vt 0.905087 dit_step5_xt 0.980221 dit_step6_vt 0.897286 dit_step6_xt 0.957169 dit_step7_vt 0.875796 dit_x0 0.920141 vae_audio 0.706843 vae_audio (STFT cosine) 0.918727 [Xl-turbo] Error growth GGML vs Python stage cos max_err mean_err mean_A std_A mean_B std_B dit_step0_xt 0.999842 0.163515 0.012590 -0.002921 0.978264 -0.002721 0.976534 dit_step1_xt 0.999523 0.281044 0.021715 -0.005783 0.951995 -0.005567 0.949004 dit_step2_xt 0.998706 0.445574 0.034240 -0.009497 0.923015 -0.009330 0.918931 dit_step3_xt 0.996784 0.677074 0.051270 -0.015067 0.892074 -0.014418 0.887241 dit_step4_xt 0.991901 1.129620 0.077853 -0.022620 0.863361 -0.021453 0.857796 dit_step5_xt 0.980221 1.729501 0.118951 -0.033829 0.848035 -0.032240 0.840033 dit_step6_xt 0.957169 2.506831 0.180784 -0.050709 0.873918 -0.049194 0.863964 [Xl-sft] steps=50, shift=1.0 | acestep-v15-xl-sft-Q5_K_M.gguf [GGML] Running acestep-v15-xl-sft-Q5_K_M.gguf... [GGML] Done, 131 dump files [Python] Initializing acestep-v15-xl-sft... Loading checkpoint shards: 0%| | 0/4 [00:00 2026-07-12 10:14:42.907 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:123 - ====================================================================== 2026-07-12 10:14:42.907 | 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 10:14:42.907 | INFO | acestep.core.generation.handler.conditioning_text:_prepare_text_conditioning_inputs:125 - ====================================================================== 2026-07-12 10:14:42.911 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:110 - [preprocess_batch] Inferring prompt embeddings... 2026-07-12 10:14:42.919 | INFO | acestep.core.generation.handler.conditioning_embed:preprocess_batch:113 - [preprocess_batch] Inferring lyric embeddings... 2026-07-12 10:14:42.919 | INFO | acestep.core.generation.handler.service_generate_execute:_execute_service_generate_diffusion:161 - [service_generate] Generating audio... (DiT backend: PyTorch (cuda)) 2026-07-12 10:14:42.926 | 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