enne2/qwentts.cpp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

tests logs

Browse Source
This commit is contained in:
Pascal
2026-05-11 13:38:21 +02:00
parent 530eed6ac5
commit 42510fff59
8 changed files with 1006 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/clone-CPU-BF16.log
+176
View File
@@ -0,0 +1,176 @@
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/qwentts.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/qwentts.cpp/build/libggml-vulkan.so
load_backend: loaded CPU backend from /mnt/workspace/qwentts.cpp/build/libggml-cpu-zen4.so
[Load] Talker backend: CPU (CPU threads: 16)
[GGUF] ../models/qwen-talker-1.7b-base-BF16.gguf: 480 tensors, data at offset 5954496
[WeightCtx] Loaded 316 tensors, 3322.0 MB into backend
[Talker] Loaded: 28 layers, hidden 2048, heads 16/8, head_dim 128, FFN 6144, RoPE theta 1000000, mrope sections [24,20,20] interleaved=0
[WeightCtx] Loaded 88 tensors, 334.1 MB into backend
[CodePredictor] Loaded: 5 layers, hidden 1024, heads 16/8, head_dim 128, FFN 3072, RoPE theta 1000000, 15 acoustic codebooks (vocab 2048 each), mtp_proj linear
[WeightCtx] Loaded 76 tensors, 34.9 MB into backend
[SpeakerEncoder] Loaded: enc_dim=2048 sr=24000 mel_dim=128 hidden=512 mfa=1536 asp_attn=128 se=128 scale=8
[GGUF] ../models/qwen-tokenizer-12hz-BF16.gguf: 398 tensors, data at offset 29760
[WeightCtx] Loaded 18 tensors, 33.0 MB into backend
[Quantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512
[WeightCtx] Loaded 93 tensors, 58.1 MB into backend
[Transformer] Loaded: 8 layers, hidden 512, heads 16/16, head_dim 64, FFN 1024, RoPE theta 10000, sliding window 72
[WeightCtx] Loaded 22 tensors, 48.1 MB into backend
[Upsample] Loaded: 2 blocks (2x ratio per block), channels 1024, dwconv kernel 7
[WeightCtx] Loaded 118 tensors, 142.9 MB into backend
[DAC] Loaded: 4 blocks (strides 8/5/4/3), 24 kHz mono out, weights 142.9 MB
[WeightCtx] Loaded 2 tensors, 3.0 MB into backend
[WeightCtx] Loaded 28 tensors, 24.1 MB into backend
[SEANet] Loaded: 4 stages (ratios 4/5/6/8), 64 -> 512 channels, weights 24.1 MB
[WeightCtx] Loaded 96 tensors, 48.1 MB into backend
[EncTransformer] Loaded: 8 layers, hidden 512, heads 8/8, head_dim 64, FFN 2048, RoPE theta 10000
[WeightCtx] Loaded 1 tensors, 2.0 MB into backend
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Ready: hop 1920 samples @ 24000 Hz mono, 16 codebooks @ 12.5 Hz
[KVCache] Allocated: 28 layers, 8 KV heads, head_dim 128, max_seq_len 4096 -> 896 MB
[KVCache] Allocated: 5 layers, 8 KV heads, head_dim 128, max_seq_len 16 -> 0 MB
[Pipeline] Loaded: arch=1b7 variant=base tokenizer=qwen3_tts_tokenizer_12hz codebooks=16 speaker_encoder=loaded speakers=0
[BPE] Loaded from GGUF: 151676 vocab, 151291 merges, eos_id=151643
[BPE] Registered 5 arch special tokens (total specials=6)
[WAV] Read buffer: 380672 samples, 22050 Hz, 1 ch, 16 bit
[Audio-Resample] 22050 Hz -> 24000 Hz, 380672 samples...
[Audio-Resample] Done: 380672 -> 414337 samples
[Debug] mel-spk: [1618, 128] first4: -6.330339 -6.170074 -6.011195 -6.243161
[Debug] mel-hann: [1024] first4: 0.000000 0.000009 0.000038 0.000085
[Debug] mel-basis: [128, 513] first4: 0.000000 0.033550 0.008569 0.000000
[Debug] mel-mag: [1618, 513] first4: 0.047604 0.042322 0.042187 0.066630
[Debug] spk-frontend: [1618, 512] first4: 0.000000 0.099754 0.858149 0.000000
[Debug] spk-block3: [1618, 512] first4: 1.958561 0.099754 1.238315 0.246145
[Debug] spk-mfa: [1618, 1536] first4: 0.000000 0.922110 0.609857 0.000000
[Debug] spk-asp: [1, 3072] first4: 0.067471 0.119239 0.240741 0.344303
[SpkExtract] Extracted 2048-dim embedding from ../examples/freeman.wav (414337 samples, padded 415105)
[WAV] Read buffer: 380672 samples, 22050 Hz, 1 ch, 16 bit
[Audio-Resample] 22050 Hz -> 24000 Hz, 380672 samples...
[Audio-Resample] Done: 380672 -> 414337 samples
[Debug] audio-input: [412800] first4: -0.000546 -0.001162 -0.000871 0.000340
[Debug] seanet-out: [430, 512] first4: -0.112500 13.760122 -0.329032 2.993861
[Debug] enc-transformer-out: [430, 512] first4: 0.526661 3.208130 -0.164395 0.605249
[Debug] codec-pre-fsq: [215, 512] first4: -3.826889 -6.990853 -0.834646 2.924343
[Debug] seanet-init: [412800, 64] first4: -0.015825 -0.054180 -0.117431 0.038554
[Debug] seanet-resnet0: [412800, 64] first4: 0.057527 -0.220565 -0.137270 -0.250936
[Debug] seanet-stage0: [103200, 128] first4: 0.233115 -0.323956 0.247371 0.205229
[Debug] seanet-stage1: [20640, 256] first4: -0.363234 0.317898 -0.574110 0.579073
[Debug] seanet-stage3: [430, 1024] first4: -2.301637 -4.978660 -6.998715 -9.010848
[Pipeline] ICL ref_codes: 215 frames at 12.5 Hz (412800 audio samples)
[Prompt] Built: 38 ids, N_text=30, N_instruct=0, T_ctx=225, hidden=2048, lang=english (id=2050), speaker=none (id=-1) ref_spk_emb=yes icl=yes
[Debug] prompt-ids: [38] first4: 151644.000000 77091.000000 198.000000 80.000000
[Debug] talker-input-embed: [225, 2048] first4: 0.022157 -0.009049 0.008161 -0.020214
[Debug] trailing-text-hidden: [1, 2048] first4: -0.005331 0.008751 -0.004225 0.001141
[Debug] tts-pad-embed: [2048] first4: -0.005331 0.008751 -0.004225 0.001141
[Debug] speaker-emb: [2048] first4: 0.098336 -0.023751 -0.011551 0.006713
[Debug] ref-codes: [16, 215] first4: 1221.000000 2042.000000 509.000000 1522.000000
[Debug] talker-hidden-prefill-l0: [225, 2048] first4: -0.461393 -0.069925 0.016141 -0.039265
[Debug] talker-hidden-prefill-l7: [225, 2048] first4: -1.523327 -0.014488 0.346884 1.775712
[Debug] talker-hidden-prefill-l14: [225, 2048] first4: -1.294868 0.182233 0.448534 1.742519
[Debug] talker-hidden-prefill-l21: [225, 2048] first4: -0.770371 0.163638 0.800043 1.111360
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 7.047174 -34.401340 -7.069329 28.628973
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.239853 -1.222670 -0.223573 0.888168
[Debug] talker-logits-prefill: [3072] first4: 1.343600 -7.634940 -6.992353 -10.519046
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Sample-CP] g=0 c=1408 u=-1.0000000000 subseq=1
[Sample-CP] g=1 c=1824 u=-1.0000000000 subseq=2
[Sample-CP] g=2 c=400 u=-1.0000000000 subseq=3
[Sample-CP] g=3 c=442 u=-1.0000000000 subseq=4
[Sample-CP] g=4 c=1506 u=-1.0000000000 subseq=5
[Sample-CP] g=5 c=1085 u=-1.0000000000 subseq=6
[Sample-CP] g=6 c=470 u=-1.0000000000 subseq=7
[Sample-CP] g=7 c=1713 u=-1.0000000000 subseq=8
[Sample-CP] g=8 c=843 u=-1.0000000000 subseq=9
[Sample-CP] g=9 c=391 u=-1.0000000000 subseq=10
[Sample-CP] g=10 c=829 u=-1.0000000000 subseq=11
[Sample-CP] g=11 c=411 u=-1.0000000000 subseq=12
[Sample-CP] g=12 c=866 u=-1.0000000000 subseq=13
[Sample-CP] g=13 c=122 u=-1.0000000000 subseq=14
[Sample-CP] g=14 c=365 u=-1.0000000000 subseq=15
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] next-emb-step0: [2048] first4: 0.073561 -0.034233 -0.008814 0.042004
[Debug] talker-hidden-step1: [2048] first4: 0.566472 2.581657 -0.032918 0.227096
[Sample] step=1 c0=1902 u=-1.0000000000 subseq=16
[Sample-CP] g=0 c=1924 u=-1.0000000000 subseq=17
[Sample-CP] g=1 c=2025 u=-1.0000000000 subseq=18
[Sample-CP] g=2 c=324 u=-1.0000000000 subseq=19
[Sample-CP] g=3 c=350 u=-1.0000000000 subseq=20
[Sample-CP] g=4 c=985 u=-1.0000000000 subseq=21
[Sample-CP] g=5 c=853 u=-1.0000000000 subseq=22
[Sample-CP] g=6 c=329 u=-1.0000000000 subseq=23
[Sample-CP] g=7 c=564 u=-1.0000000000 subseq=24
[Sample-CP] g=8 c=692 u=-1.0000000000 subseq=25
[Sample-CP] g=9 c=1702 u=-1.0000000000 subseq=26
[Sample-CP] g=10 c=437 u=-1.0000000000 subseq=27
[Sample-CP] g=11 c=729 u=-1.0000000000 subseq=28
[Sample-CP] g=12 c=344 u=-1.0000000000 subseq=29
[Sample-CP] g=13 c=85 u=-1.0000000000 subseq=30
[Sample-CP] g=14 c=1435 u=-1.0000000000 subseq=31
[Pipeline] Generated 8 frames
[Pipeline] Generated 16 frames
[Pipeline] Generated 24 frames
[Pipeline] Generated 32 frames
[Pipeline] Generated 40 frames
[Pipeline] Generated 48 frames
[Pipeline] Generated 56 frames
[Pipeline] Generated 64 frames
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] output-audio: [122880] first4: -0.001455 -0.002483 -0.002276 -0.001114
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
[Input] RefAudio: ../examples/freeman.wav
[Input] RefText: 213 chars: If you go into different cultures, they have different conce...
[Input] Lang: english Seed: 42 MaxNewTokens: 64
[Input] Mode: greedy ICL
[Python] Device: cuda
[Python] RefWav: 414337 samples 24000 Hz 17.26s
[Python] SpeakerEmb shape: (2048,) dtype: torch.float32
[Python] RefCodes shape: (16, 215) (K, T_codec)
[Python] InputIds shape: (1, 38)
[Python] RefIds shape: (1, 53)
[Python] Codes shape: (63, 16) (T_frames, num_code_groups)
[Python] Audio: 120960 samples 24000 Hz 5.04s -> python/clone/clone-python.wav
[Quant] BF16 -> ../models/qwen-talker-1.7b-base-BF16.gguf + ../models/qwen-tokenizer-12hz-BF16.gguf
[GGML] Cmd: ../build/qwen-tts --model ../models/qwen-talker-1.7b-base-BF16.gguf --codec ../models/qwen-tokenizer-12hz-BF16.gguf --seed --text [...] --ref-audio ../examples/freeman.wav --ref-text [...] --lang english --max-new 64 --dump cpp/clone -o cpp/clone/clone-cpp.wav --greedy
[GGML] Audio: 122880 samples 24000 Hz 5.12s -> cpp/clone/clone-cpp.wav
[Cossim] PromptIDs exact: 100.00% (38 values)
[Cossim] RefCodes exact: 86.77% (3440 values)
[Cossim] Embed cos: 0.978030 max: 1.8825e-01 mean: 6.5095e-03
[Cossim] TrailingText cos: 1.000000 max: 1.0431e-06 mean: 9.3616e-09
[Cossim] TTSPadEmbed cos: 1.000000 max: 1.0431e-06 mean: 9.3616e-09
[Cossim] L0 cos: 0.989590 max: 7.7399e-01 mean: 8.9716e-03
[Cossim] L7 cos: 0.999993 max: 1.2854e+01 mean: 2.3280e-02
[Cossim] L14 cos: 0.999976 max: 1.2858e+01 mean: 5.2397e-02
[Cossim] L21 cos: 0.999696 max: 1.7395e+01 mean: 2.0170e-01
[Cossim] L27 cos: 0.998838 max: 4.6131e+01 mean: 4.6407e-01
[Cossim] Final cos: 0.993945 max: 1.9927e+01 mean: 1.5086e-01
[Cossim] Logits cos: 0.999804 max: 5.7617e-01 mean: 7.8600e-02
[Cossim] NextEmbStep0 cos: 0.967766 max: 6.3431e-02 mean: 1.2632e-02
[Cossim] TalkerHiddenStep1 cos: 0.998228 max: 7.7749e-01 mean: 1.3022e-01
[Cossim] MelHann cos: 1.000000 max: 1.7881e-07 mean: 3.2043e-08
[Cossim] MelBasis cos: 1.000000 max: 1.7881e-07 mean: 3.9379e-10
[Cossim] MelMag cos: 1.000000 max: 3.1307e-05 mean: 5.6396e-07
[Cossim] MelSpk cos: 1.000000 max: 2.5492e-03 mean: 9.3754e-06
[Cossim] SeanetInit cos: 1.000000 max: 9.1240e-04 mean: 1.6087e-05
[Cossim] SeanetResnet0 cos: 0.999973 max: 1.9622e-03 mean: 1.4500e-04
[Cossim] SeanetStage0 cos: 0.999996 max: 3.9780e-02 mean: 3.7406e-04
[Cossim] SeanetStage1 cos: 0.999972 max: 7.4337e-01 mean: 2.1076e-03
[Cossim] SeanetStage3 cos: 0.999999 max: 9.8260e-01 mean: 2.5173e-02
[Cossim] SeanetOut cos: 0.999991 max: 8.8695e-02 mean: 9.9127e-03
[Cossim] EncTransformer cos: 0.999988 max: 9.6033e-01 mean: 6.5646e-03
[Cossim] CodecPreFSQ cos: 0.999991 max: 1.2993e-01 mean: 1.1939e-02
[Cossim] SpkFrontend cos: 0.999997 max: 2.5180e-02 mean: 9.4975e-04
[Cossim] SpkBlock3 cos: 0.999991 max: 1.0583e-01 mean: 4.2477e-03
[Cossim] SpkMFA cos: 0.999979 max: 6.8943e-02 mean: 1.0444e-03
[Cossim] SpkASP cos: 0.999999 max: 2.2756e-03 mean: 1.8859e-04
[Cossim] SpeakerEmb cos: 1.000000 max: 6.4747e-04 mean: 1.3471e-04
[Cossim] CodesFull exact: 3.77% (1008 values)
[Cossim] Audio cos: -0.002504
[Cossim] WAV stft_cos: 0.222491 samples: 120960
tests/clone-CPU-F32.log
+176
View File
@@ -0,0 +1,176 @@
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/qwentts.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/qwentts.cpp/build/libggml-vulkan.so
load_backend: loaded CPU backend from /mnt/workspace/qwentts.cpp/build/libggml-cpu-zen4.so
[Load] Talker backend: CPU (CPU threads: 16)
[GGUF] ../models/qwen-talker-1.7b-base-F32.gguf: 480 tensors, data at offset 5954432
[WeightCtx] Loaded 316 tensors, 6643.5 MB into backend
[Talker] Loaded: 28 layers, hidden 2048, heads 16/8, head_dim 128, FFN 6144, RoPE theta 1000000, mrope sections [24,20,20] interleaved=0
[WeightCtx] Loaded 88 tensors, 668.1 MB into backend
[CodePredictor] Loaded: 5 layers, hidden 1024, heads 16/8, head_dim 128, FFN 3072, RoPE theta 1000000, 15 acoustic codebooks (vocab 2048 each), mtp_proj linear
[WeightCtx] Loaded 76 tensors, 45.8 MB into backend
[SpeakerEncoder] Loaded: enc_dim=2048 sr=24000 mel_dim=128 hidden=512 mfa=1536 asp_attn=128 se=128 scale=8
[GGUF] ../models/qwen-tokenizer-12hz-F32.gguf: 398 tensors, data at offset 29664
[WeightCtx] Loaded 18 tensors, 33.0 MB into backend
[Quantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512
[WeightCtx] Loaded 93 tensors, 116.1 MB into backend
[Transformer] Loaded: 8 layers, hidden 512, heads 16/16, head_dim 64, FFN 1024, RoPE theta 10000, sliding window 72
[WeightCtx] Loaded 22 tensors, 80.1 MB into backend
[Upsample] Loaded: 2 blocks (2x ratio per block), channels 1024, dwconv kernel 7
[WeightCtx] Loaded 118 tensors, 142.9 MB into backend
[DAC] Loaded: 4 blocks (strides 8/5/4/3), 24 kHz mono out, weights 142.9 MB
[WeightCtx] Loaded 2 tensors, 3.0 MB into backend
[WeightCtx] Loaded 28 tensors, 24.1 MB into backend
[SEANet] Loaded: 4 stages (ratios 4/5/6/8), 64 -> 512 channels, weights 24.1 MB
[WeightCtx] Loaded 96 tensors, 96.1 MB into backend
[EncTransformer] Loaded: 8 layers, hidden 512, heads 8/8, head_dim 64, FFN 2048, RoPE theta 10000
[WeightCtx] Loaded 1 tensors, 2.0 MB into backend
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Ready: hop 1920 samples @ 24000 Hz mono, 16 codebooks @ 12.5 Hz
[KVCache] Allocated: 28 layers, 8 KV heads, head_dim 128, max_seq_len 4096 -> 896 MB
[KVCache] Allocated: 5 layers, 8 KV heads, head_dim 128, max_seq_len 16 -> 0 MB
[Pipeline] Loaded: arch=1b7 variant=base tokenizer=qwen3_tts_tokenizer_12hz codebooks=16 speaker_encoder=loaded speakers=0
[BPE] Loaded from GGUF: 151676 vocab, 151291 merges, eos_id=151643
[BPE] Registered 5 arch special tokens (total specials=6)
[WAV] Read buffer: 380672 samples, 22050 Hz, 1 ch, 16 bit
[Audio-Resample] 22050 Hz -> 24000 Hz, 380672 samples...
[Audio-Resample] Done: 380672 -> 414337 samples
[Debug] mel-spk: [1618, 128] first4: -6.330339 -6.170074 -6.011195 -6.243161
[Debug] mel-hann: [1024] first4: 0.000000 0.000009 0.000038 0.000085
[Debug] mel-basis: [128, 513] first4: 0.000000 0.033550 0.008569 0.000000
[Debug] mel-mag: [1618, 513] first4: 0.047604 0.042322 0.042187 0.066630
[Debug] spk-frontend: [1618, 512] first4: 0.000000 0.097904 0.854992 0.000000
[Debug] spk-block3: [1618, 512] first4: 1.955987 0.097904 1.233304 0.246811
[Debug] spk-mfa: [1618, 1536] first4: 0.000000 0.922734 0.614916 0.000000
[Debug] spk-asp: [1, 3072] first4: 0.067723 0.119136 0.240880 0.344137
[SpkExtract] Extracted 2048-dim embedding from ../examples/freeman.wav (414337 samples, padded 415105)
[WAV] Read buffer: 380672 samples, 22050 Hz, 1 ch, 16 bit
[Audio-Resample] 22050 Hz -> 24000 Hz, 380672 samples...
[Audio-Resample] Done: 380672 -> 414337 samples
[Debug] audio-input: [412800] first4: -0.000546 -0.001162 -0.000871 0.000340
[Debug] seanet-out: [430, 512] first4: -0.120292 13.742489 -0.348104 3.009387
[Debug] enc-transformer-out: [430, 512] first4: 0.526076 3.125370 -0.189241 0.630764
[Debug] codec-pre-fsq: [215, 512] first4: -3.819415 -6.973695 -0.843401 2.950197
[Debug] seanet-init: [412800, 64] first4: -0.015825 -0.054180 -0.117431 0.038554
[Debug] seanet-resnet0: [412800, 64] first4: 0.057635 -0.220722 -0.137297 -0.251265
[Debug] seanet-stage0: [103200, 128] first4: 0.233614 -0.324178 0.246648 0.204675
[Debug] seanet-stage1: [20640, 256] first4: -0.363436 0.316413 -0.575420 0.576034
[Debug] seanet-stage3: [430, 1024] first4: -2.315332 -4.954537 -7.026399 -8.983903
[Pipeline] ICL ref_codes: 215 frames at 12.5 Hz (412800 audio samples)
[Prompt] Built: 38 ids, N_text=30, N_instruct=0, T_ctx=225, hidden=2048, lang=english (id=2050), speaker=none (id=-1) ref_spk_emb=yes icl=yes
[Debug] prompt-ids: [38] first4: 151644.000000 77091.000000 198.000000 80.000000
[Debug] talker-input-embed: [225, 2048] first4: 0.022157 -0.009049 0.008161 -0.020214
[Debug] trailing-text-hidden: [1, 2048] first4: -0.005331 0.008751 -0.004225 0.001141
[Debug] tts-pad-embed: [2048] first4: -0.005331 0.008751 -0.004225 0.001141
[Debug] speaker-emb: [2048] first4: 0.098317 -0.023761 -0.011860 0.006760
[Debug] ref-codes: [16, 215] first4: 1221.000000 2042.000000 509.000000 1522.000000
[Debug] talker-hidden-prefill-l0: [225, 2048] first4: -0.460968 -0.069929 0.016046 -0.039092
[Debug] talker-hidden-prefill-l7: [225, 2048] first4: -1.525317 -0.007489 0.340217 1.777318
[Debug] talker-hidden-prefill-l14: [225, 2048] first4: -1.296772 0.189305 0.441262 1.743797
[Debug] talker-hidden-prefill-l21: [225, 2048] first4: -0.771753 0.171250 0.791419 1.111589
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 7.133091 -34.474182 -7.108906 28.674915
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.242892 -1.225839 -0.224931 0.890014
[Debug] talker-logits-prefill: [3072] first4: 1.330122 -7.768506 -6.935927 -10.338604
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Sample-CP] g=0 c=1408 u=-1.0000000000 subseq=1
[Sample-CP] g=1 c=1824 u=-1.0000000000 subseq=2
[Sample-CP] g=2 c=400 u=-1.0000000000 subseq=3
[Sample-CP] g=3 c=442 u=-1.0000000000 subseq=4
[Sample-CP] g=4 c=1506 u=-1.0000000000 subseq=5
[Sample-CP] g=5 c=1085 u=-1.0000000000 subseq=6
[Sample-CP] g=6 c=470 u=-1.0000000000 subseq=7
[Sample-CP] g=7 c=1713 u=-1.0000000000 subseq=8
[Sample-CP] g=8 c=843 u=-1.0000000000 subseq=9
[Sample-CP] g=9 c=391 u=-1.0000000000 subseq=10
[Sample-CP] g=10 c=829 u=-1.0000000000 subseq=11
[Sample-CP] g=11 c=411 u=-1.0000000000 subseq=12
[Sample-CP] g=12 c=866 u=-1.0000000000 subseq=13
[Sample-CP] g=13 c=122 u=-1.0000000000 subseq=14
[Sample-CP] g=14 c=365 u=-1.0000000000 subseq=15
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] next-emb-step0: [2048] first4: 0.073561 -0.034233 -0.008814 0.042004
[Debug] talker-hidden-step1: [2048] first4: 0.584089 2.683546 -0.065922 0.135429
[Sample] step=1 c0=1902 u=-1.0000000000 subseq=16
[Sample-CP] g=0 c=1924 u=-1.0000000000 subseq=17
[Sample-CP] g=1 c=2025 u=-1.0000000000 subseq=18
[Sample-CP] g=2 c=324 u=-1.0000000000 subseq=19
[Sample-CP] g=3 c=350 u=-1.0000000000 subseq=20
[Sample-CP] g=4 c=985 u=-1.0000000000 subseq=21
[Sample-CP] g=5 c=853 u=-1.0000000000 subseq=22
[Sample-CP] g=6 c=329 u=-1.0000000000 subseq=23
[Sample-CP] g=7 c=564 u=-1.0000000000 subseq=24
[Sample-CP] g=8 c=692 u=-1.0000000000 subseq=25
[Sample-CP] g=9 c=1702 u=-1.0000000000 subseq=26
[Sample-CP] g=10 c=437 u=-1.0000000000 subseq=27
[Sample-CP] g=11 c=729 u=-1.0000000000 subseq=28
[Sample-CP] g=12 c=344 u=-1.0000000000 subseq=29
[Sample-CP] g=13 c=85 u=-1.0000000000 subseq=30
[Sample-CP] g=14 c=1435 u=-1.0000000000 subseq=31
[Pipeline] Generated 8 frames
[Pipeline] Generated 16 frames
[Pipeline] Generated 24 frames
[Pipeline] Generated 32 frames
[Pipeline] Generated 40 frames
[Pipeline] Generated 48 frames
[Pipeline] Generated 56 frames
[Pipeline] Generated 64 frames
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] output-audio: [122880] first4: -0.001458 -0.002491 -0.002285 -0.001123
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
[Input] RefAudio: ../examples/freeman.wav
[Input] RefText: 213 chars: If you go into different cultures, they have different conce...
[Input] Lang: english Seed: 42 MaxNewTokens: 64
[Input] Mode: greedy ICL
[Python] Device: cuda
[Python] RefWav: 414337 samples 24000 Hz 17.26s
[Python] SpeakerEmb shape: (2048,) dtype: torch.float32
[Python] RefCodes shape: (16, 215) (K, T_codec)
[Python] InputIds shape: (1, 38)
[Python] RefIds shape: (1, 53)
[Python] Codes shape: (63, 16) (T_frames, num_code_groups)
[Python] Audio: 120960 samples 24000 Hz 5.04s -> python/clone/clone-python.wav
[Quant] F32 -> ../models/qwen-talker-1.7b-base-F32.gguf + ../models/qwen-tokenizer-12hz-F32.gguf
[GGML] Cmd: ../build/qwen-tts --model ../models/qwen-talker-1.7b-base-F32.gguf --codec ../models/qwen-tokenizer-12hz-F32.gguf --seed --text [...] --ref-audio ../examples/freeman.wav --ref-text [...] --lang english --max-new 64 --dump cpp/clone -o cpp/clone/clone-cpp.wav --greedy
[GGML] Audio: 122880 samples 24000 Hz 5.12s -> cpp/clone/clone-cpp.wav
[Cossim] PromptIDs exact: 100.00% (38 values)
[Cossim] RefCodes exact: 96.72% (3440 values)
[Cossim] Embed cos: 0.994609 max: 1.5292e-01 mean: 1.6562e-03
[Cossim] TrailingText cos: 1.000000 max: 1.0431e-06 mean: 9.3616e-09
[Cossim] TTSPadEmbed cos: 1.000000 max: 1.0431e-06 mean: 9.3616e-09
[Cossim] L0 cos: 0.997321 max: 7.2147e-01 mean: 2.5377e-03
[Cossim] L7 cos: 0.999998 max: 1.2205e+00 mean: 7.8975e-03
[Cossim] L14 cos: 0.999994 max: 2.2472e+00 mean: 2.0513e-02
[Cossim] L21 cos: 0.999920 max: 1.3672e+01 mean: 8.9238e-02
[Cossim] L27 cos: 0.999724 max: 2.5152e+01 mean: 2.0175e-01
[Cossim] Final cos: 0.998553 max: 1.1353e+01 mean: 6.5865e-02
[Cossim] Logits cos: 0.999966 max: 2.0577e-01 mean: 3.0349e-02
[Cossim] NextEmbStep0 cos: 0.967766 max: 6.3431e-02 mean: 1.2632e-02
[Cossim] TalkerHiddenStep1 cos: 0.999113 max: 4.3387e-01 mean: 9.2055e-02
[Cossim] MelHann cos: 1.000000 max: 1.7881e-07 mean: 3.2043e-08
[Cossim] MelBasis cos: 1.000000 max: 1.7881e-07 mean: 3.9379e-10
[Cossim] MelMag cos: 1.000000 max: 3.1307e-05 mean: 5.6396e-07
[Cossim] MelSpk cos: 1.000000 max: 2.5492e-03 mean: 9.3754e-06
[Cossim] SeanetInit cos: 1.000000 max: 2.2967e-04 mean: 2.9508e-06
[Cossim] SeanetResnet0 cos: 0.999998 max: 1.1875e-03 mean: 3.0708e-05
[Cossim] SeanetStage0 cos: 0.999998 max: 3.7565e-02 mean: 1.9554e-04
[Cossim] SeanetStage1 cos: 0.999976 max: 6.9249e-01 mean: 1.4980e-03
[Cossim] SeanetStage3 cos: 1.000000 max: 5.3660e-01 mean: 9.7786e-03
[Cossim] SeanetOut cos: 0.999998 max: 8.9135e-02 mean: 4.0736e-03
[Cossim] EncTransformer cos: 0.999999 max: 7.9834e-02 mean: 1.6466e-03
[Cossim] CodecPreFSQ cos: 0.999999 max: 1.1980e-01 mean: 4.2799e-03
[Cossim] SpkFrontend cos: 1.000000 max: 2.1482e-04 mean: 1.4394e-06
[Cossim] SpkBlock3 cos: 1.000000 max: 3.3045e-04 mean: 4.3737e-06
[Cossim] SpkMFA cos: 1.000000 max: 3.8040e-04 mean: 1.3130e-06
[Cossim] SpkASP cos: 1.000000 max: 8.1062e-06 mean: 2.0734e-07
[Cossim] SpeakerEmb cos: 1.000000 max: 1.3351e-05 mean: 2.5133e-07
[Cossim] CodesFull exact: 5.26% (1008 values)
[Cossim] Audio cos: 0.028471
[Cossim] WAV stft_cos: 0.257819 samples: 120960
tests/clone-CPU-Q4_K_M.log
+176
View File
@@ -0,0 +1,176 @@
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/qwentts.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/qwentts.cpp/build/libggml-vulkan.so
load_backend: loaded CPU backend from /mnt/workspace/qwentts.cpp/build/libggml-cpu-zen4.so
[Load] Talker backend: CPU (CPU threads: 16)
[GGUF] ../models/qwen-talker-1.7b-base-Q4_K_M.gguf: 480 tensors, data at offset 5954496
[WeightCtx] Loaded 316 tensors, 1012.7 MB into backend
[Talker] Loaded: 28 layers, hidden 2048, heads 16/8, head_dim 128, FFN 6144, RoPE theta 1000000, mrope sections [24,20,20] interleaved=0
[WeightCtx] Loaded 88 tensors, 109.5 MB into backend
[CodePredictor] Loaded: 5 layers, hidden 1024, heads 16/8, head_dim 128, FFN 3072, RoPE theta 1000000, 15 acoustic codebooks (vocab 2048 each), mtp_proj linear
[WeightCtx] Loaded 76 tensors, 34.9 MB into backend
[SpeakerEncoder] Loaded: enc_dim=2048 sr=24000 mel_dim=128 hidden=512 mfa=1536 asp_attn=128 se=128 scale=8
[GGUF] ../models/qwen-tokenizer-12hz-Q4_K_M.gguf: 398 tensors, data at offset 29760
[WeightCtx] Loaded 18 tensors, 33.0 MB into backend
[Quantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512
[WeightCtx] Loaded 93 tensors, 16.4 MB into backend
[Transformer] Loaded: 8 layers, hidden 512, heads 16/16, head_dim 64, FFN 1024, RoPE theta 10000, sliding window 72
[WeightCtx] Loaded 22 tensors, 25.1 MB into backend
[Upsample] Loaded: 2 blocks (2x ratio per block), channels 1024, dwconv kernel 7
[WeightCtx] Loaded 118 tensors, 142.9 MB into backend
[DAC] Loaded: 4 blocks (strides 8/5/4/3), 24 kHz mono out, weights 142.9 MB
[WeightCtx] Loaded 2 tensors, 3.0 MB into backend
[WeightCtx] Loaded 28 tensors, 24.1 MB into backend
[SEANet] Loaded: 4 stages (ratios 4/5/6/8), 64 -> 512 channels, weights 24.1 MB
[WeightCtx] Loaded 96 tensors, 13.6 MB into backend
[EncTransformer] Loaded: 8 layers, hidden 512, heads 8/8, head_dim 64, FFN 2048, RoPE theta 10000
[WeightCtx] Loaded 1 tensors, 2.0 MB into backend
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Ready: hop 1920 samples @ 24000 Hz mono, 16 codebooks @ 12.5 Hz
[KVCache] Allocated: 28 layers, 8 KV heads, head_dim 128, max_seq_len 4096 -> 896 MB
[KVCache] Allocated: 5 layers, 8 KV heads, head_dim 128, max_seq_len 16 -> 0 MB
[Pipeline] Loaded: arch=1b7 variant=base tokenizer=qwen3_tts_tokenizer_12hz codebooks=16 speaker_encoder=loaded speakers=0
[BPE] Loaded from GGUF: 151676 vocab, 151291 merges, eos_id=151643
[BPE] Registered 5 arch special tokens (total specials=6)
[WAV] Read buffer: 380672 samples, 22050 Hz, 1 ch, 16 bit
[Audio-Resample] 22050 Hz -> 24000 Hz, 380672 samples...
[Audio-Resample] Done: 380672 -> 414337 samples
[Debug] mel-spk: [1618, 128] first4: -6.330339 -6.170074 -6.011195 -6.243161
[Debug] mel-hann: [1024] first4: 0.000000 0.000009 0.000038 0.000085
[Debug] mel-basis: [128, 513] first4: 0.000000 0.033550 0.008569 0.000000
[Debug] mel-mag: [1618, 513] first4: 0.047604 0.042322 0.042187 0.066630
[Debug] spk-frontend: [1618, 512] first4: 0.000000 0.098112 0.854719 0.000000
[Debug] spk-block3: [1618, 512] first4: 1.956183 0.098112 1.233075 0.246215
[Debug] spk-mfa: [1618, 1536] first4: 0.000000 0.922732 0.615171 0.000000
[Debug] spk-asp: [1, 3072] first4: 0.067754 0.119138 0.240900 0.344101
[SpkExtract] Extracted 2048-dim embedding from ../examples/freeman.wav (414337 samples, padded 415105)
[WAV] Read buffer: 380672 samples, 22050 Hz, 1 ch, 16 bit
[Audio-Resample] 22050 Hz -> 24000 Hz, 380672 samples...
[Audio-Resample] Done: 380672 -> 414337 samples
[Debug] audio-input: [412800] first4: -0.000546 -0.001162 -0.000871 0.000340
[Debug] seanet-out: [430, 512] first4: -0.120292 13.742489 -0.348104 3.009387
[Debug] enc-transformer-out: [430, 512] first4: 0.867046 3.016766 0.059672 0.395116
[Debug] codec-pre-fsq: [215, 512] first4: -1.983214 -7.035533 -2.228207 2.551106
[Debug] seanet-init: [412800, 64] first4: -0.015825 -0.054180 -0.117431 0.038554
[Debug] seanet-resnet0: [412800, 64] first4: 0.057635 -0.220722 -0.137297 -0.251265
[Debug] seanet-stage0: [103200, 128] first4: 0.233614 -0.324178 0.246648 0.204675
[Debug] seanet-stage1: [20640, 256] first4: -0.363436 0.316413 -0.575420 0.576034
[Debug] seanet-stage3: [430, 1024] first4: -2.315332 -4.954537 -7.026399 -8.983903
[Pipeline] ICL ref_codes: 215 frames at 12.5 Hz (412800 audio samples)
[Prompt] Built: 38 ids, N_text=30, N_instruct=0, T_ctx=225, hidden=2048, lang=english (id=2050), speaker=none (id=-1) ref_spk_emb=yes icl=yes
[Debug] prompt-ids: [38] first4: 151644.000000 77091.000000 198.000000 80.000000
[Debug] talker-input-embed: [225, 2048] first4: 0.022105 -0.009079 0.008285 -0.019901
[Debug] trailing-text-hidden: [1, 2048] first4: -0.005388 0.008927 -0.004415 0.000679
[Debug] tts-pad-embed: [2048] first4: -0.005388 0.008927 -0.004415 0.000679
[Debug] speaker-emb: [2048] first4: 0.098343 -0.023770 -0.011812 0.006754
[Debug] ref-codes: [16, 215] first4: 1221.000000 2042.000000 696.000000 1522.000000
[Debug] talker-hidden-prefill-l0: [225, 2048] first4: -0.441821 -0.069394 0.003658 0.020266
[Debug] talker-hidden-prefill-l7: [225, 2048] first4: -8.877851 0.790778 0.485222 0.194165
[Debug] talker-hidden-prefill-l14: [225, 2048] first4: -8.792478 1.086515 0.586075 0.265361
[Debug] talker-hidden-prefill-l21: [225, 2048] first4: -8.716181 1.630248 1.123896 -0.509474
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 4.479937 -33.404484 -6.656628 23.418547
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.151891 -1.182686 -0.209713 0.723736
[Debug] talker-logits-prefill: [3072] first4: -0.049402 -4.614957 -7.224392 -10.561956
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Sample-CP] g=0 c=1408 u=-1.0000000000 subseq=1
[Sample-CP] g=1 c=1824 u=-1.0000000000 subseq=2
[Sample-CP] g=2 c=417 u=-1.0000000000 subseq=3
[Sample-CP] g=3 c=442 u=-1.0000000000 subseq=4
[Sample-CP] g=4 c=1412 u=-1.0000000000 subseq=5
[Sample-CP] g=5 c=1155 u=-1.0000000000 subseq=6
[Sample-CP] g=6 c=815 u=-1.0000000000 subseq=7
[Sample-CP] g=7 c=106 u=-1.0000000000 subseq=8
[Sample-CP] g=8 c=153 u=-1.0000000000 subseq=9
[Sample-CP] g=9 c=1776 u=-1.0000000000 subseq=10
[Sample-CP] g=10 c=1435 u=-1.0000000000 subseq=11
[Sample-CP] g=11 c=1728 u=-1.0000000000 subseq=12
[Sample-CP] g=12 c=1922 u=-1.0000000000 subseq=13
[Sample-CP] g=13 c=1022 u=-1.0000000000 subseq=14
[Sample-CP] g=14 c=302 u=-1.0000000000 subseq=15
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 417.000000
[Debug] next-emb-step0: [2048] first4: 0.033382 0.011384 -0.022313 0.047956
[Debug] talker-hidden-step1: [2048] first4: 0.727982 2.735055 1.014960 0.314809
[Sample] step=1 c0=1902 u=-1.0000000000 subseq=16
[Sample-CP] g=0 c=1924 u=-1.0000000000 subseq=17
[Sample-CP] g=1 c=2025 u=-1.0000000000 subseq=18
[Sample-CP] g=2 c=324 u=-1.0000000000 subseq=19
[Sample-CP] g=3 c=1210 u=-1.0000000000 subseq=20
[Sample-CP] g=4 c=985 u=-1.0000000000 subseq=21
[Sample-CP] g=5 c=1636 u=-1.0000000000 subseq=22
[Sample-CP] g=6 c=329 u=-1.0000000000 subseq=23
[Sample-CP] g=7 c=1058 u=-1.0000000000 subseq=24
[Sample-CP] g=8 c=2030 u=-1.0000000000 subseq=25
[Sample-CP] g=9 c=1702 u=-1.0000000000 subseq=26
[Sample-CP] g=10 c=437 u=-1.0000000000 subseq=27
[Sample-CP] g=11 c=706 u=-1.0000000000 subseq=28
[Sample-CP] g=12 c=1529 u=-1.0000000000 subseq=29
[Sample-CP] g=13 c=85 u=-1.0000000000 subseq=30
[Sample-CP] g=14 c=803 u=-1.0000000000 subseq=31
[Pipeline] Generated 8 frames
[Pipeline] Generated 16 frames
[Pipeline] Generated 24 frames
[Pipeline] Generated 32 frames
[Pipeline] Generated 40 frames
[Pipeline] Generated 48 frames
[Pipeline] Generated 56 frames
[Pipeline] Generated 64 frames
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 417.000000
[Debug] output-audio: [122880] first4: -0.001169 -0.002026 -0.001938 -0.000989
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
[Input] RefAudio: ../examples/freeman.wav
[Input] RefText: 213 chars: If you go into different cultures, they have different conce...
[Input] Lang: english Seed: 42 MaxNewTokens: 64
[Input] Mode: greedy ICL
[Python] Device: cuda
[Python] RefWav: 414337 samples 24000 Hz 17.26s
[Python] SpeakerEmb shape: (2048,) dtype: torch.float32
[Python] RefCodes shape: (16, 215) (K, T_codec)
[Python] InputIds shape: (1, 38)
[Python] RefIds shape: (1, 53)
[Python] Codes shape: (63, 16) (T_frames, num_code_groups)
[Python] Audio: 120960 samples 24000 Hz 5.04s -> python/clone/clone-python.wav
[Quant] Q4_K_M -> ../models/qwen-talker-1.7b-base-Q4_K_M.gguf + ../models/qwen-tokenizer-12hz-Q4_K_M.gguf
[GGML] Cmd: ../build/qwen-tts --model ../models/qwen-talker-1.7b-base-Q4_K_M.gguf --codec ../models/qwen-tokenizer-12hz-Q4_K_M.gguf --seed --text [...] --ref-audio ../examples/freeman.wav --ref-text [...] --lang english --max-new 64 --dump cpp/clone -o cpp/clone/clone-cpp.wav --greedy
[GGML] Audio: 122880 samples 24000 Hz 5.12s -> cpp/clone/clone-cpp.wav
[Cossim] PromptIDs exact: 100.00% (38 values)
[Cossim] RefCodes exact: 53.08% (3440 values)
[Cossim] Embed cos: 0.922425 max: 2.2247e-01 mean: 2.1113e-02
[Cossim] TrailingText cos: 0.999969 max: 1.4858e-03 mean: 2.2947e-04
[Cossim] TTSPadEmbed cos: 0.999969 max: 1.4858e-03 mean: 2.2947e-04
[Cossim] L0 cos: 0.963454 max: 9.8532e-01 mean: 2.7584e-02
[Cossim] L7 cos: 0.997433 max: 3.4714e+01 mean: 1.6242e-01
[Cossim] L14 cos: 0.997361 max: 3.4809e+01 mean: 2.4571e-01
[Cossim] L21 cos: 0.995869 max: 3.6073e+01 mean: 7.7571e-01
[Cossim] L27 cos: 0.945093 max: 2.1444e+03 mean: 2.0117e+00
[Cossim] Final cos: 0.954199 max: 5.7803e+01 mean: 5.5407e-01
[Cossim] Logits cos: 0.991186 max: 3.6044e+00 mean: 5.1607e-01
[Cossim] NextEmbStep0 cos: 0.828841 max: 1.4047e-01 mean: 2.9099e-02
[Cossim] TalkerHiddenStep1 cos: 0.963338 max: 3.9269e+00 mean: 5.8955e-01
[Cossim] MelHann cos: 1.000000 max: 1.7881e-07 mean: 3.2043e-08
[Cossim] MelBasis cos: 1.000000 max: 1.7881e-07 mean: 3.9379e-10
[Cossim] MelMag cos: 1.000000 max: 3.1307e-05 mean: 5.6396e-07
[Cossim] MelSpk cos: 1.000000 max: 2.5492e-03 mean: 9.3754e-06
[Cossim] SeanetInit cos: 1.000000 max: 2.2967e-04 mean: 2.9508e-06
[Cossim] SeanetResnet0 cos: 0.999998 max: 1.1875e-03 mean: 3.0708e-05
[Cossim] SeanetStage0 cos: 0.999998 max: 3.7565e-02 mean: 1.9554e-04
[Cossim] SeanetStage1 cos: 0.999976 max: 6.9249e-01 mean: 1.4980e-03
[Cossim] SeanetStage3 cos: 1.000000 max: 5.3660e-01 mean: 9.7786e-03
[Cossim] SeanetOut cos: 0.999998 max: 8.9135e-02 mean: 4.0736e-03
[Cossim] EncTransformer cos: 0.992676 max: 3.8309e+00 mean: 1.6672e-01
[Cossim] CodecPreFSQ cos: 0.996176 max: 2.1645e+00 mean: 2.4283e-01
[Cossim] SpkFrontend cos: 1.000000 max: 2.6605e-03 mean: 1.1964e-04
[Cossim] SpkBlock3 cos: 1.000000 max: 1.3163e-02 mean: 5.3422e-04
[Cossim] SpkMFA cos: 1.000000 max: 9.8418e-03 mean: 1.3212e-04
[Cossim] SpkASP cos: 1.000000 max: 2.6596e-04 mean: 2.3921e-05
[Cossim] SpeakerEmb cos: 1.000000 max: 2.3174e-04 mean: 1.6094e-05
[Cossim] CodesFull exact: 2.08% (1008 values)
[Cossim] Audio cos: -0.004291
[Cossim] WAV stft_cos: 0.191699 samples: 120960
tests/clone-CPU-Q8_0.log
+176
View File
@@ -0,0 +1,176 @@
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/qwentts.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/qwentts.cpp/build/libggml-vulkan.so
load_backend: loaded CPU backend from /mnt/workspace/qwentts.cpp/build/libggml-cpu-zen4.so
[Load] Talker backend: CPU (CPU threads: 16)
[GGUF] ../models/qwen-talker-1.7b-base-Q8_0.gguf: 480 tensors, data at offset 5954496
[WeightCtx] Loaded 316 tensors, 1765.0 MB into backend
[Talker] Loaded: 28 layers, hidden 2048, heads 16/8, head_dim 128, FFN 6144, RoPE theta 1000000, mrope sections [24,20,20] interleaved=0
[WeightCtx] Loaded 88 tensors, 177.5 MB into backend
[CodePredictor] Loaded: 5 layers, hidden 1024, heads 16/8, head_dim 128, FFN 3072, RoPE theta 1000000, 15 acoustic codebooks (vocab 2048 each), mtp_proj linear
[WeightCtx] Loaded 76 tensors, 34.9 MB into backend
[SpeakerEncoder] Loaded: enc_dim=2048 sr=24000 mel_dim=128 hidden=512 mfa=1536 asp_attn=128 se=128 scale=8
[GGUF] ../models/qwen-tokenizer-12hz-Q8_0.gguf: 398 tensors, data at offset 29760
[WeightCtx] Loaded 18 tensors, 33.0 MB into backend
[Quantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512
[WeightCtx] Loaded 93 tensors, 30.9 MB into backend
[Transformer] Loaded: 8 layers, hidden 512, heads 16/16, head_dim 64, FFN 1024, RoPE theta 10000, sliding window 72
[WeightCtx] Loaded 22 tensors, 33.1 MB into backend
[Upsample] Loaded: 2 blocks (2x ratio per block), channels 1024, dwconv kernel 7
[WeightCtx] Loaded 118 tensors, 142.9 MB into backend
[DAC] Loaded: 4 blocks (strides 8/5/4/3), 24 kHz mono out, weights 142.9 MB
[WeightCtx] Loaded 2 tensors, 3.0 MB into backend
[WeightCtx] Loaded 28 tensors, 24.1 MB into backend
[SEANet] Loaded: 4 stages (ratios 4/5/6/8), 64 -> 512 channels, weights 24.1 MB
[WeightCtx] Loaded 96 tensors, 25.6 MB into backend
[EncTransformer] Loaded: 8 layers, hidden 512, heads 8/8, head_dim 64, FFN 2048, RoPE theta 10000
[WeightCtx] Loaded 1 tensors, 2.0 MB into backend
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Ready: hop 1920 samples @ 24000 Hz mono, 16 codebooks @ 12.5 Hz
[KVCache] Allocated: 28 layers, 8 KV heads, head_dim 128, max_seq_len 4096 -> 896 MB
[KVCache] Allocated: 5 layers, 8 KV heads, head_dim 128, max_seq_len 16 -> 0 MB
[Pipeline] Loaded: arch=1b7 variant=base tokenizer=qwen3_tts_tokenizer_12hz codebooks=16 speaker_encoder=loaded speakers=0
[BPE] Loaded from GGUF: 151676 vocab, 151291 merges, eos_id=151643
[BPE] Registered 5 arch special tokens (total specials=6)
[WAV] Read buffer: 380672 samples, 22050 Hz, 1 ch, 16 bit
[Audio-Resample] 22050 Hz -> 24000 Hz, 380672 samples...
[Audio-Resample] Done: 380672 -> 414337 samples
[Debug] mel-spk: [1618, 128] first4: -6.330339 -6.170074 -6.011195 -6.243161
[Debug] mel-hann: [1024] first4: 0.000000 0.000009 0.000038 0.000085
[Debug] mel-basis: [128, 513] first4: 0.000000 0.033550 0.008569 0.000000
[Debug] mel-mag: [1618, 513] first4: 0.047604 0.042322 0.042187 0.066630
[Debug] spk-frontend: [1618, 512] first4: 0.000000 0.098112 0.854719 0.000000
[Debug] spk-block3: [1618, 512] first4: 1.956183 0.098112 1.233075 0.246215
[Debug] spk-mfa: [1618, 1536] first4: 0.000000 0.922732 0.615171 0.000000
[Debug] spk-asp: [1, 3072] first4: 0.067754 0.119138 0.240900 0.344101
[SpkExtract] Extracted 2048-dim embedding from ../examples/freeman.wav (414337 samples, padded 415105)
[WAV] Read buffer: 380672 samples, 22050 Hz, 1 ch, 16 bit
[Audio-Resample] 22050 Hz -> 24000 Hz, 380672 samples...
[Audio-Resample] Done: 380672 -> 414337 samples
[Debug] audio-input: [412800] first4: -0.000546 -0.001162 -0.000871 0.000340
[Debug] seanet-out: [430, 512] first4: -0.120292 13.742489 -0.348104 3.009387
[Debug] enc-transformer-out: [430, 512] first4: 0.716872 2.959070 -0.221450 0.852651
[Debug] codec-pre-fsq: [215, 512] first4: -3.691730 -6.839433 -0.608027 2.856996
[Debug] seanet-init: [412800, 64] first4: -0.015825 -0.054180 -0.117431 0.038554
[Debug] seanet-resnet0: [412800, 64] first4: 0.057635 -0.220722 -0.137297 -0.251265
[Debug] seanet-stage0: [103200, 128] first4: 0.233614 -0.324178 0.246648 0.204675
[Debug] seanet-stage1: [20640, 256] first4: -0.363436 0.316413 -0.575420 0.576034
[Debug] seanet-stage3: [430, 1024] first4: -2.315332 -4.954537 -7.026399 -8.983903
[Pipeline] ICL ref_codes: 215 frames at 12.5 Hz (412800 audio samples)
[Prompt] Built: 38 ids, N_text=30, N_instruct=0, T_ctx=225, hidden=2048, lang=english (id=2050), speaker=none (id=-1) ref_spk_emb=yes icl=yes
[Debug] prompt-ids: [38] first4: 151644.000000 77091.000000 198.000000 80.000000
[Debug] talker-input-embed: [225, 2048] first4: 0.021979 -0.009018 0.008292 -0.020170
[Debug] trailing-text-hidden: [1, 2048] first4: -0.005270 0.008716 -0.004355 0.001036
[Debug] tts-pad-embed: [2048] first4: -0.005270 0.008716 -0.004355 0.001036
[Debug] speaker-emb: [2048] first4: 0.098343 -0.023770 -0.011812 0.006754
[Debug] ref-codes: [16, 215] first4: 1221.000000 2042.000000 509.000000 1522.000000
[Debug] talker-hidden-prefill-l0: [225, 2048] first4: -0.468042 -0.069505 0.014260 -0.037197
[Debug] talker-hidden-prefill-l7: [225, 2048] first4: -2.188824 0.154519 0.209425 1.591020
[Debug] talker-hidden-prefill-l14: [225, 2048] first4: -1.960747 0.344378 0.321142 1.566153
[Debug] talker-hidden-prefill-l21: [225, 2048] first4: -1.448369 0.302417 0.673400 0.974501
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 6.696815 -34.333122 -7.302668 28.583275
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.227877 -1.219968 -0.230900 0.886548
[Debug] talker-logits-prefill: [3072] first4: 1.200291 -7.925805 -6.852166 -10.537476
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Sample-CP] g=0 c=1408 u=-1.0000000000 subseq=1
[Sample-CP] g=1 c=1824 u=-1.0000000000 subseq=2
[Sample-CP] g=2 c=400 u=-1.0000000000 subseq=3
[Sample-CP] g=3 c=550 u=-1.0000000000 subseq=4
[Sample-CP] g=4 c=1506 u=-1.0000000000 subseq=5
[Sample-CP] g=5 c=1192 u=-1.0000000000 subseq=6
[Sample-CP] g=6 c=1162 u=-1.0000000000 subseq=7
[Sample-CP] g=7 c=853 u=-1.0000000000 subseq=8
[Sample-CP] g=8 c=692 u=-1.0000000000 subseq=9
[Sample-CP] g=9 c=18 u=-1.0000000000 subseq=10
[Sample-CP] g=10 c=1502 u=-1.0000000000 subseq=11
[Sample-CP] g=11 c=1702 u=-1.0000000000 subseq=12
[Sample-CP] g=12 c=1860 u=-1.0000000000 subseq=13
[Sample-CP] g=13 c=1008 u=-1.0000000000 subseq=14
[Sample-CP] g=14 c=265 u=-1.0000000000 subseq=15
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] next-emb-step0: [2048] first4: 0.033802 -0.026094 0.058828 0.088771
[Debug] talker-hidden-step1: [2048] first4: 0.394369 2.910929 -0.614081 0.194712
[Sample] step=1 c0=1902 u=-1.0000000000 subseq=16
[Sample-CP] g=0 c=1924 u=-1.0000000000 subseq=17
[Sample-CP] g=1 c=2025 u=-1.0000000000 subseq=18
[Sample-CP] g=2 c=324 u=-1.0000000000 subseq=19
[Sample-CP] g=3 c=350 u=-1.0000000000 subseq=20
[Sample-CP] g=4 c=985 u=-1.0000000000 subseq=21
[Sample-CP] g=5 c=853 u=-1.0000000000 subseq=22
[Sample-CP] g=6 c=329 u=-1.0000000000 subseq=23
[Sample-CP] g=7 c=564 u=-1.0000000000 subseq=24
[Sample-CP] g=8 c=692 u=-1.0000000000 subseq=25
[Sample-CP] g=9 c=1702 u=-1.0000000000 subseq=26
[Sample-CP] g=10 c=437 u=-1.0000000000 subseq=27
[Sample-CP] g=11 c=729 u=-1.0000000000 subseq=28
[Sample-CP] g=12 c=344 u=-1.0000000000 subseq=29
[Sample-CP] g=13 c=85 u=-1.0000000000 subseq=30
[Sample-CP] g=14 c=1668 u=-1.0000000000 subseq=31
[Pipeline] Generated 8 frames
[Pipeline] Generated 16 frames
[Pipeline] Generated 24 frames
[Pipeline] Generated 32 frames
[Pipeline] Generated 40 frames
[Pipeline] Generated 48 frames
[Pipeline] Generated 56 frames
[Pipeline] Generated 64 frames
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] output-audio: [122880] first4: -0.001312 -0.002237 -0.002049 -0.000977
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
[Input] RefAudio: ../examples/freeman.wav
[Input] RefText: 213 chars: If you go into different cultures, they have different conce...
[Input] Lang: english Seed: 42 MaxNewTokens: 64
[Input] Mode: greedy ICL
[Python] Device: cuda
[Python] RefWav: 414337 samples 24000 Hz 17.26s
[Python] SpeakerEmb shape: (2048,) dtype: torch.float32
[Python] RefCodes shape: (16, 215) (K, T_codec)
[Python] InputIds shape: (1, 38)
[Python] RefIds shape: (1, 53)
[Python] Codes shape: (63, 16) (T_frames, num_code_groups)
[Python] Audio: 120960 samples 24000 Hz 5.04s -> python/clone/clone-python.wav
[Quant] Q8_0 -> ../models/qwen-talker-1.7b-base-Q8_0.gguf + ../models/qwen-tokenizer-12hz-Q8_0.gguf
[GGML] Cmd: ../build/qwen-tts --model ../models/qwen-talker-1.7b-base-Q8_0.gguf --codec ../models/qwen-tokenizer-12hz-Q8_0.gguf --seed --text [...] --ref-audio ../examples/freeman.wav --ref-text [...] --lang english --max-new 64 --dump cpp/clone -o cpp/clone/clone-cpp.wav --greedy
[GGML] Audio: 122880 samples 24000 Hz 5.12s -> cpp/clone/clone-cpp.wav
[Cossim] PromptIDs exact: 100.00% (38 values)
[Cossim] RefCodes exact: 90.90% (3440 values)
[Cossim] Embed cos: 0.984762 max: 1.6358e-01 mean: 5.1389e-03
[Cossim] TrailingText cos: 0.999999 max: 2.1233e-04 mean: 4.3863e-05
[Cossim] TTSPadEmbed cos: 0.999999 max: 2.1233e-04 mean: 4.3863e-05
[Cossim] L0 cos: 0.992177 max: 7.7281e-01 mean: 7.3119e-03
[Cossim] L7 cos: 0.999983 max: 2.5666e+00 mean: 2.6623e-02
[Cossim] L14 cos: 0.999969 max: 2.5645e+00 mean: 5.2457e-02
[Cossim] L21 cos: 0.999711 max: 1.7818e+01 mean: 2.0121e-01
[Cossim] L27 cos: 0.998660 max: 1.1899e+02 mean: 4.8494e-01
[Cossim] Final cos: 0.994302 max: 1.6731e+01 mean: 1.4969e-01
[Cossim] Logits cos: 0.999849 max: 6.0111e-01 mean: 6.8404e-02
[Cossim] NextEmbStep0 cos: 0.825702 max: 2.4852e-01 mean: 2.9797e-02
[Cossim] TalkerHiddenStep1 cos: 0.988163 max: 2.2708e+00 mean: 3.3062e-01
[Cossim] MelHann cos: 1.000000 max: 1.7881e-07 mean: 3.2043e-08
[Cossim] MelBasis cos: 1.000000 max: 1.7881e-07 mean: 3.9379e-10
[Cossim] MelMag cos: 1.000000 max: 3.1307e-05 mean: 5.6396e-07
[Cossim] MelSpk cos: 1.000000 max: 2.5492e-03 mean: 9.3754e-06
[Cossim] SeanetInit cos: 1.000000 max: 2.2967e-04 mean: 2.9508e-06
[Cossim] SeanetResnet0 cos: 0.999998 max: 1.1875e-03 mean: 3.0708e-05
[Cossim] SeanetStage0 cos: 0.999998 max: 3.7565e-02 mean: 1.9554e-04
[Cossim] SeanetStage1 cos: 0.999976 max: 6.9249e-01 mean: 1.4980e-03
[Cossim] SeanetStage3 cos: 1.000000 max: 5.3660e-01 mean: 9.7786e-03
[Cossim] SeanetOut cos: 0.999998 max: 8.9135e-02 mean: 4.0736e-03
[Cossim] EncTransformer cos: 0.999877 max: 8.5370e-01 mean: 2.1630e-02
[Cossim] CodecPreFSQ cos: 0.999942 max: 4.8146e-01 mean: 3.0311e-02
[Cossim] SpkFrontend cos: 1.000000 max: 2.6605e-03 mean: 1.1964e-04
[Cossim] SpkBlock3 cos: 1.000000 max: 1.3163e-02 mean: 5.3422e-04
[Cossim] SpkMFA cos: 1.000000 max: 9.8418e-03 mean: 1.3212e-04
[Cossim] SpkASP cos: 1.000000 max: 2.6596e-04 mean: 2.3921e-05
[Cossim] SpeakerEmb cos: 1.000000 max: 2.3174e-04 mean: 1.6094e-05
[Cossim] CodesFull exact: 2.68% (1008 values)
[Cossim] Audio cos: -0.002931
[Cossim] WAV stft_cos: 0.145943 samples: 120960
tests/clone-CUDA0-BF16.log
+141 -1
View File
@@ -18,7 +18,112 @@ load_backend: loaded CPU backend from /mnt/workspace/qwentts.cpp/build/libggml-c
[Quantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512
[WeightCtx] Loaded 93 tensors, 58.1 MB into backend
[Transformer] Loaded: 8 layers, hidden 512, heads 16/16, head_dim 64, FFN 1024, RoPE theta 10000, sliding window 72
[CausalTransConv] FATAL: 'tok_dec.upsample.0.conv.weight' expected F32 or F16, got type 30
[WeightCtx] Loaded 22 tensors, 48.1 MB into backend
[Upsample] Loaded: 2 blocks (2x ratio per block), channels 1024, dwconv kernel 7
[WeightCtx] Loaded 118 tensors, 142.9 MB into backend
[DAC] Loaded: 4 blocks (strides 8/5/4/3), 24 kHz mono out, weights 142.9 MB
[WeightCtx] Loaded 2 tensors, 3.0 MB into backend
[WeightCtx] Loaded 28 tensors, 24.1 MB into backend
[SEANet] Loaded: 4 stages (ratios 4/5/6/8), 64 -> 512 channels, weights 24.1 MB
[WeightCtx] Loaded 96 tensors, 48.1 MB into backend
[EncTransformer] Loaded: 8 layers, hidden 512, heads 8/8, head_dim 64, FFN 2048, RoPE theta 10000
[WeightCtx] Loaded 1 tensors, 2.0 MB into backend
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Ready: hop 1920 samples @ 24000 Hz mono, 16 codebooks @ 12.5 Hz
[KVCache] Allocated: 28 layers, 8 KV heads, head_dim 128, max_seq_len 4096 -> 896 MB
[KVCache] Allocated: 5 layers, 8 KV heads, head_dim 128, max_seq_len 16 -> 0 MB
[Pipeline] Loaded: arch=1b7 variant=base tokenizer=qwen3_tts_tokenizer_12hz codebooks=16 speaker_encoder=loaded speakers=0
[BPE] Loaded from GGUF: 151676 vocab, 151291 merges, eos_id=151643
[BPE] Registered 5 arch special tokens (total specials=6)
[WAV] Read buffer: 380672 samples, 22050 Hz, 1 ch, 16 bit
[Audio-Resample] 22050 Hz -> 24000 Hz, 380672 samples...
[Audio-Resample] Done: 380672 -> 414337 samples
[Debug] mel-spk: [1618, 128] first4: -6.330339 -6.170074 -6.011195 -6.243161
[Debug] mel-hann: [1024] first4: 0.000000 0.000009 0.000038 0.000085
[Debug] mel-basis: [128, 513] first4: 0.000000 0.033550 0.008569 0.000000
[Debug] mel-mag: [1618, 513] first4: 0.047604 0.042322 0.042187 0.066630
[Debug] spk-frontend: [1618, 512] first4: 0.000000 0.101562 0.858398 0.000000
[Debug] spk-block3: [1618, 512] first4: 1.958879 0.101562 1.239357 0.246141
[Debug] spk-mfa: [1618, 1536] first4: 0.000000 0.919922 0.609741 0.000000
[Debug] spk-asp: [1, 3072] first4: 0.067580 0.118872 0.240696 0.344030
[SpkExtract] Extracted 2048-dim embedding from ../examples/freeman.wav (414337 samples, padded 415105)
[WAV] Read buffer: 380672 samples, 22050 Hz, 1 ch, 16 bit
[Audio-Resample] 22050 Hz -> 24000 Hz, 380672 samples...
[Audio-Resample] Done: 380672 -> 414337 samples
[Debug] audio-input: [412800] first4: -0.000546 -0.001162 -0.000871 0.000340
[Debug] seanet-out: [430, 512] first4: -0.111631 13.770183 -0.302467 2.984188
[Debug] enc-transformer-out: [430, 512] first4: 0.531195 3.238480 -0.146394 0.612285
[Debug] codec-pre-fsq: [215, 512] first4: -3.886719 -7.015625 -0.871094 2.914062
[Debug] seanet-init: [412800, 64] first4: -0.015825 -0.054180 -0.117431 0.038554
[Debug] seanet-resnet0: [412800, 64] first4: 0.057569 -0.220525 -0.137337 -0.251168
[Debug] seanet-stage0: [103200, 128] first4: 0.233175 -0.323887 0.247025 0.205248
[Debug] seanet-stage1: [20640, 256] first4: -0.363801 0.317300 -0.571584 0.594021
[Debug] seanet-stage3: [430, 1024] first4: -2.271604 -4.929065 -6.964391 -9.027892
[Pipeline] ICL ref_codes: 215 frames at 12.5 Hz (412800 audio samples)
[Prompt] Built: 38 ids, N_text=30, N_instruct=0, T_ctx=225, hidden=2048, lang=english (id=2050), speaker=none (id=-1) ref_spk_emb=yes icl=yes
[Debug] prompt-ids: [38] first4: 151644.000000 77091.000000 198.000000 80.000000
[Debug] talker-input-embed: [225, 2048] first4: 0.022157 -0.009049 0.008161 -0.020214
[Debug] trailing-text-hidden: [1, 2048] first4: -0.005331 0.008751 -0.004225 0.001141
[Debug] tts-pad-embed: [2048] first4: -0.005331 0.008751 -0.004225 0.001141
[Debug] speaker-emb: [2048] first4: 0.098059 -0.023872 -0.011379 0.006934
[Debug] ref-codes: [16, 215] first4: 1221.000000 2042.000000 509.000000 1522.000000
[Debug] talker-hidden-prefill-l0: [225, 2048] first4: -0.461242 -0.070023 0.016279 -0.039013
[Debug] talker-hidden-prefill-l7: [225, 2048] first4: -1.526536 -0.010379 0.339523 1.777291
[Debug] talker-hidden-prefill-l14: [225, 2048] first4: -1.298100 0.186848 0.440622 1.744122
[Debug] talker-hidden-prefill-l21: [225, 2048] first4: -0.770047 0.168826 0.790043 1.108689
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 7.103000 -34.495621 -7.125972 28.862335
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.241600 -1.225247 -0.225222 0.894842
[Debug] talker-logits-prefill: [3072] first4: 0.847656 -7.875000 -6.593750 -10.375000
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Sample-CP] g=0 c=1408 u=-1.0000000000 subseq=1
[Sample-CP] g=1 c=1824 u=-1.0000000000 subseq=2
[Sample-CP] g=2 c=400 u=-1.0000000000 subseq=3
[Sample-CP] g=3 c=240 u=-1.0000000000 subseq=4
[Sample-CP] g=4 c=2010 u=-1.0000000000 subseq=5
[Sample-CP] g=5 c=196 u=-1.0000000000 subseq=6
[Sample-CP] g=6 c=580 u=-1.0000000000 subseq=7
[Sample-CP] g=7 c=1713 u=-1.0000000000 subseq=8
[Sample-CP] g=8 c=1221 u=-1.0000000000 subseq=9
[Sample-CP] g=9 c=1796 u=-1.0000000000 subseq=10
[Sample-CP] g=10 c=1228 u=-1.0000000000 subseq=11
[Sample-CP] g=11 c=2015 u=-1.0000000000 subseq=12
[Sample-CP] g=12 c=143 u=-1.0000000000 subseq=13
[Sample-CP] g=13 c=1062 u=-1.0000000000 subseq=14
[Sample-CP] g=14 c=1094 u=-1.0000000000 subseq=15
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] next-emb-step0: [2048] first4: 0.010340 -0.009316 0.074921 0.036954
[Debug] talker-hidden-step1: [2048] first4: 0.028097 2.930421 -0.771587 0.041852
[Sample] step=1 c0=1902 u=-1.0000000000 subseq=16
[Sample-CP] g=0 c=1924 u=-1.0000000000 subseq=17
[Sample-CP] g=1 c=2025 u=-1.0000000000 subseq=18
[Sample-CP] g=2 c=324 u=-1.0000000000 subseq=19
[Sample-CP] g=3 c=350 u=-1.0000000000 subseq=20
[Sample-CP] g=4 c=985 u=-1.0000000000 subseq=21
[Sample-CP] g=5 c=853 u=-1.0000000000 subseq=22
[Sample-CP] g=6 c=329 u=-1.0000000000 subseq=23
[Sample-CP] g=7 c=564 u=-1.0000000000 subseq=24
[Sample-CP] g=8 c=692 u=-1.0000000000 subseq=25
[Sample-CP] g=9 c=1702 u=-1.0000000000 subseq=26
[Sample-CP] g=10 c=437 u=-1.0000000000 subseq=27
[Sample-CP] g=11 c=729 u=-1.0000000000 subseq=28
[Sample-CP] g=12 c=344 u=-1.0000000000 subseq=29
[Sample-CP] g=13 c=85 u=-1.0000000000 subseq=30
[Sample-CP] g=14 c=1668 u=-1.0000000000 subseq=31
[Pipeline] Generated 8 frames
[Pipeline] Generated 16 frames
[Pipeline] Generated 24 frames
[Pipeline] Generated 32 frames
[Pipeline] Generated 40 frames
[Pipeline] Generated 48 frames
[Pipeline] Generated 56 frames
[Pipeline] Generated 64 frames
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] output-audio: [122880] first4: -0.001310 -0.002270 -0.002123 -0.000994
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
[Input] RefAudio: ../examples/freeman.wav
[Input] RefText: 213 chars: If you go into different cultures, they have different conce...
@@ -34,3 +139,38 @@ load_backend: loaded CPU backend from /mnt/workspace/qwentts.cpp/build/libggml-c
[Python] Audio: 120960 samples 24000 Hz 5.04s -> python/clone/clone-python.wav
[Quant] BF16 -> ../models/qwen-talker-1.7b-base-BF16.gguf + ../models/qwen-tokenizer-12hz-BF16.gguf
[GGML] Cmd: ../build/qwen-tts --model ../models/qwen-talker-1.7b-base-BF16.gguf --codec ../models/qwen-tokenizer-12hz-BF16.gguf --seed --text [...] --ref-audio ../examples/freeman.wav --ref-text [...] --lang english --max-new 64 --dump cpp/clone -o cpp/clone/clone-cpp.wav --greedy
[GGML] Audio: 122880 samples 24000 Hz 5.12s -> cpp/clone/clone-cpp.wav
[Cossim] PromptIDs exact: 100.00% (38 values)
[Cossim] RefCodes exact: 78.52% (3440 values)
[Cossim] Embed cos: 0.965579 max: 1.9437e-01 mean: 9.8515e-03
[Cossim] TrailingText cos: 1.000000 max: 1.0431e-06 mean: 9.3616e-09
[Cossim] TTSPadEmbed cos: 1.000000 max: 1.0431e-06 mean: 9.3616e-09
[Cossim] L0 cos: 0.984331 max: 8.0261e-01 mean: 1.3008e-02
[Cossim] L7 cos: 0.999981 max: 4.7529e+01 mean: 3.1958e-02
[Cossim] L14 cos: 0.999956 max: 4.7530e+01 mean: 7.0140e-02
[Cossim] L21 cos: 0.999568 max: 4.7582e+01 mean: 2.6741e-01
[Cossim] L27 cos: 0.998450 max: 4.7913e+01 mean: 6.0463e-01
[Cossim] Final cos: 0.992009 max: 1.5263e+01 mean: 1.9648e-01
[Cossim] Logits cos: 0.999456 max: 1.0356e+00 mean: 1.3013e-01
[Cossim] NextEmbStep0 cos: 0.829096 max: 1.5468e-01 mean: 2.9154e-02
[Cossim] TalkerHiddenStep1 cos: 0.992008 max: 2.2751e+00 mean: 2.7376e-01
[Cossim] MelHann cos: 1.000000 max: 1.7881e-07 mean: 3.2043e-08
[Cossim] MelBasis cos: 1.000000 max: 1.7881e-07 mean: 3.9379e-10
[Cossim] MelMag cos: 1.000000 max: 3.1307e-05 mean: 5.6396e-07
[Cossim] MelSpk cos: 1.000000 max: 2.5492e-03 mean: 9.3754e-06
[Cossim] SeanetInit cos: 1.000000 max: 9.1669e-04 mean: 1.6221e-05
[Cossim] SeanetResnet0 cos: 0.999961 max: 2.2789e-03 mean: 1.6912e-04
[Cossim] SeanetStage0 cos: 0.999983 max: 5.9753e-02 mean: 7.5125e-04
[Cossim] SeanetStage1 cos: 0.999862 max: 7.2425e-01 mean: 5.3037e-03
[Cossim] SeanetStage3 cos: 0.999994 max: 2.1089e+00 mean: 4.7947e-02
[Cossim] SeanetOut cos: 0.999953 max: 3.2791e-01 mean: 2.0392e-02
[Cossim] EncTransformer cos: 0.999956 max: 1.1787e+00 mean: 1.1794e-02
[Cossim] CodecPreFSQ cos: 0.999960 max: 4.7962e-01 mean: 2.3320e-02
[Cossim] SpkFrontend cos: 0.999995 max: 3.5096e-02 mean: 1.0924e-03
[Cossim] SpkBlock3 cos: 0.999987 max: 1.3367e-01 mean: 4.9918e-03
[Cossim] SpkMFA cos: 0.999973 max: 7.5789e-02 mean: 1.1867e-03
[Cossim] SpkASP cos: 0.999999 max: 3.9836e-03 mean: 2.4063e-04
[Cossim] SpeakerEmb cos: 1.000000 max: 1.7091e-03 mean: 2.1396e-04
[Cossim] CodesFull exact: 4.76% (1008 values)
[Cossim] Audio cos: 0.062271
[Cossim] WAV stft_cos: 0.382412 samples: 120960
tests/clone-CUDA0-F32.log
+10 -10
View File
@@ -20,15 +20,15 @@ load_backend: loaded CPU backend from /mnt/workspace/qwentts.cpp/build/libggml-c
[Transformer] Loaded: 8 layers, hidden 512, heads 16/16, head_dim 64, FFN 1024, RoPE theta 10000, sliding window 72
[WeightCtx] Loaded 22 tensors, 80.1 MB into backend
[Upsample] Loaded: 2 blocks (2x ratio per block), channels 1024, dwconv kernel 7
[WeightCtx] Loaded 118 tensors, 199.8 MB into backend
[DAC] Loaded: 4 blocks (strides 8/5/4/3), 24 kHz mono out, weights 199.8 MB
[WeightCtx] Loaded 2 tensors, 6.0 MB into backend
[WeightCtx] Loaded 28 tensors, 48.2 MB into backend
[SEANet] Loaded: 4 stages (ratios 4/5/6/8), 64 -> 512 channels, weights 48.2 MB
[WeightCtx] Loaded 118 tensors, 142.9 MB into backend
[DAC] Loaded: 4 blocks (strides 8/5/4/3), 24 kHz mono out, weights 142.9 MB
[WeightCtx] Loaded 2 tensors, 3.0 MB into backend
[WeightCtx] Loaded 28 tensors, 24.1 MB into backend
[SEANet] Loaded: 4 stages (ratios 4/5/6/8), 64 -> 512 channels, weights 24.1 MB
[WeightCtx] Loaded 96 tensors, 96.1 MB into backend
[EncTransformer] Loaded: 8 layers, hidden 512, heads 8/8, head_dim 64, FFN 2048, RoPE theta 10000
[WeightCtx] Loaded 1 tensors, 4.0 MB into backend
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 4.0 MB
[WeightCtx] Loaded 1 tensors, 2.0 MB into backend
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Ready: hop 1920 samples @ 24000 Hz mono, 16 codebooks @ 12.5 Hz
@@ -121,7 +121,7 @@ load_backend: loaded CPU backend from /mnt/workspace/qwentts.cpp/build/libggml-c
[Pipeline] Generated 64 frames
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] output-audio: [122880] first4: -0.001264 -0.002193 -0.002047 -0.000957
[Debug] output-audio: [122880] first4: -0.001265 -0.002192 -0.002047 -0.000956
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
@@ -172,5 +172,5 @@ load_backend: loaded CPU backend from /mnt/workspace/qwentts.cpp/build/libggml-c
[Cossim] SpkASP cos: 1.000000 max: 7.9870e-06 mean: 2.2269e-07
[Cossim] SpeakerEmb cos: 1.000000 max: 1.4305e-05 mean: 2.5463e-07
[Cossim] CodesFull exact: 3.27% (1008 values)
[Cossim] Audio cos: 0.053928
[Cossim] WAV stft_cos: 0.375522 samples: 120960
[Cossim] Audio cos: 0.053931
[Cossim] WAV stft_cos: 0.375521 samples: 120960
tests/clone-Vulkan0-BF16.log
+141 -1
View File
@@ -18,7 +18,112 @@ load_backend: loaded CPU backend from /mnt/workspace/qwentts.cpp/build/libggml-c
[Quantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512
[WeightCtx] Loaded 93 tensors, 58.1 MB into backend
[Transformer] Loaded: 8 layers, hidden 512, heads 16/16, head_dim 64, FFN 1024, RoPE theta 10000, sliding window 72
[CausalTransConv] FATAL: 'tok_dec.upsample.0.conv.weight' expected F32 or F16, got type 30
[WeightCtx] Loaded 22 tensors, 48.1 MB into backend
[Upsample] Loaded: 2 blocks (2x ratio per block), channels 1024, dwconv kernel 7
[WeightCtx] Loaded 118 tensors, 142.9 MB into backend
[DAC] Loaded: 4 blocks (strides 8/5/4/3), 24 kHz mono out, weights 142.9 MB
[WeightCtx] Loaded 2 tensors, 3.0 MB into backend
[WeightCtx] Loaded 28 tensors, 24.1 MB into backend
[SEANet] Loaded: 4 stages (ratios 4/5/6/8), 64 -> 512 channels, weights 24.1 MB
[WeightCtx] Loaded 96 tensors, 48.1 MB into backend
[EncTransformer] Loaded: 8 layers, hidden 512, heads 8/8, head_dim 64, FFN 2048, RoPE theta 10000
[WeightCtx] Loaded 1 tensors, 2.0 MB into backend
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Ready: hop 1920 samples @ 24000 Hz mono, 16 codebooks @ 12.5 Hz
[KVCache] Allocated: 28 layers, 8 KV heads, head_dim 128, max_seq_len 4096 -> 896 MB
[KVCache] Allocated: 5 layers, 8 KV heads, head_dim 128, max_seq_len 16 -> 0 MB
[Pipeline] Loaded: arch=1b7 variant=base tokenizer=qwen3_tts_tokenizer_12hz codebooks=16 speaker_encoder=loaded speakers=0
[BPE] Loaded from GGUF: 151676 vocab, 151291 merges, eos_id=151643
[BPE] Registered 5 arch special tokens (total specials=6)
[WAV] Read buffer: 380672 samples, 22050 Hz, 1 ch, 16 bit
[Audio-Resample] 22050 Hz -> 24000 Hz, 380672 samples...
[Audio-Resample] Done: 380672 -> 414337 samples
[Debug] mel-spk: [1618, 128] first4: -6.330339 -6.170074 -6.011195 -6.243161
[Debug] mel-hann: [1024] first4: 0.000000 0.000009 0.000038 0.000085
[Debug] mel-basis: [128, 513] first4: 0.000000 0.033550 0.008569 0.000000
[Debug] mel-mag: [1618, 513] first4: 0.047604 0.042322 0.042187 0.066630
[Debug] spk-frontend: [1618, 512] first4: 0.000000 0.099754 0.858150 0.000000
[Debug] spk-block3: [1618, 512] first4: 1.962623 0.099754 1.239126 0.246363
[Debug] spk-mfa: [1618, 1536] first4: 0.000000 0.923592 0.609606 0.000000
[Debug] spk-asp: [1, 3072] first4: 0.067512 0.118912 0.240832 0.344075
[SpkExtract] Extracted 2048-dim embedding from ../examples/freeman.wav (414337 samples, padded 415105)
[WAV] Read buffer: 380672 samples, 22050 Hz, 1 ch, 16 bit
[Audio-Resample] 22050 Hz -> 24000 Hz, 380672 samples...
[Audio-Resample] Done: 380672 -> 414337 samples
[Debug] audio-input: [412800] first4: -0.000546 -0.001162 -0.000871 0.000340
[Debug] seanet-out: [430, 512] first4: -0.110593 13.765666 -0.317055 2.981106
[Debug] enc-transformer-out: [430, 512] first4: 0.536260 3.195749 -0.156796 0.610111
[Debug] codec-pre-fsq: [215, 512] first4: -3.871765 -7.009766 -0.896362 2.946289
[Debug] seanet-init: [412800, 64] first4: -0.015825 -0.054180 -0.117431 0.038554
[Debug] seanet-resnet0: [412800, 64] first4: 0.057569 -0.220647 -0.137337 -0.251168
[Debug] seanet-stage0: [103200, 128] first4: 0.233175 -0.323887 0.247269 0.205126
[Debug] seanet-stage1: [20640, 256] first4: -0.363801 0.318033 -0.571584 0.592068
[Debug] seanet-stage3: [430, 1024] first4: -2.265745 -4.951282 -6.929723 -8.998595
[Pipeline] ICL ref_codes: 215 frames at 12.5 Hz (412800 audio samples)
[Prompt] Built: 38 ids, N_text=30, N_instruct=0, T_ctx=225, hidden=2048, lang=english (id=2050), speaker=none (id=-1) ref_spk_emb=yes icl=yes
[Debug] prompt-ids: [38] first4: 151644.000000 77091.000000 198.000000 80.000000
[Debug] talker-input-embed: [225, 2048] first4: 0.022157 -0.009049 0.008161 -0.020214
[Debug] trailing-text-hidden: [1, 2048] first4: -0.005331 0.008751 -0.004225 0.001141
[Debug] tts-pad-embed: [2048] first4: -0.005331 0.008751 -0.004225 0.001141
[Debug] speaker-emb: [2048] first4: 0.098348 -0.023729 -0.011668 0.006747
[Debug] ref-codes: [16, 215] first4: 1221.000000 2042.000000 509.000000 1522.000000
[Debug] talker-hidden-prefill-l0: [225, 2048] first4: -0.461532 -0.069882 0.015988 -0.039215
[Debug] talker-hidden-prefill-l7: [225, 2048] first4: -1.523297 -0.014717 0.347053 1.776380
[Debug] talker-hidden-prefill-l14: [225, 2048] first4: -1.294979 0.182117 0.448301 1.743050
[Debug] talker-hidden-prefill-l21: [225, 2048] first4: -0.769804 0.163546 0.799334 1.111618
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 7.050038 -34.400974 -7.071052 28.632160
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.239952 -1.222665 -0.223629 0.888272
[Debug] talker-logits-prefill: [3072] first4: 0.900955 -7.879465 -6.792802 -10.430634
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Sample-CP] g=0 c=1408 u=-1.0000000000 subseq=1
[Sample-CP] g=1 c=1203 u=-1.0000000000 subseq=2
[Sample-CP] g=2 c=203 u=-1.0000000000 subseq=3
[Sample-CP] g=3 c=447 u=-1.0000000000 subseq=4
[Sample-CP] g=4 c=645 u=-1.0000000000 subseq=5
[Sample-CP] g=5 c=580 u=-1.0000000000 subseq=6
[Sample-CP] g=6 c=905 u=-1.0000000000 subseq=7
[Sample-CP] g=7 c=432 u=-1.0000000000 subseq=8
[Sample-CP] g=8 c=1962 u=-1.0000000000 subseq=9
[Sample-CP] g=9 c=346 u=-1.0000000000 subseq=10
[Sample-CP] g=10 c=93 u=-1.0000000000 subseq=11
[Sample-CP] g=11 c=1546 u=-1.0000000000 subseq=12
[Sample-CP] g=12 c=936 u=-1.0000000000 subseq=13
[Sample-CP] g=13 c=2002 u=-1.0000000000 subseq=14
[Sample-CP] g=14 c=906 u=-1.0000000000 subseq=15
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1203.000000 203.000000
[Debug] next-emb-step0: [2048] first4: 0.056839 -0.027733 0.030860 -0.040129
[Debug] talker-hidden-step1: [2048] first4: -0.632528 3.517379 -1.484728 0.181401
[Sample] step=1 c0=1570 u=-1.0000000000 subseq=16
[Sample-CP] g=0 c=682 u=-1.0000000000 subseq=17
[Sample-CP] g=1 c=814 u=-1.0000000000 subseq=18
[Sample-CP] g=2 c=367 u=-1.0000000000 subseq=19
[Sample-CP] g=3 c=1906 u=-1.0000000000 subseq=20
[Sample-CP] g=4 c=645 u=-1.0000000000 subseq=21
[Sample-CP] g=5 c=267 u=-1.0000000000 subseq=22
[Sample-CP] g=6 c=91 u=-1.0000000000 subseq=23
[Sample-CP] g=7 c=1573 u=-1.0000000000 subseq=24
[Sample-CP] g=8 c=162 u=-1.0000000000 subseq=25
[Sample-CP] g=9 c=765 u=-1.0000000000 subseq=26
[Sample-CP] g=10 c=1767 u=-1.0000000000 subseq=27
[Sample-CP] g=11 c=855 u=-1.0000000000 subseq=28
[Sample-CP] g=12 c=1456 u=-1.0000000000 subseq=29
[Sample-CP] g=13 c=1185 u=-1.0000000000 subseq=30
[Sample-CP] g=14 c=108 u=-1.0000000000 subseq=31
[Pipeline] Generated 8 frames
[Pipeline] Generated 16 frames
[Pipeline] Generated 24 frames
[Pipeline] Generated 32 frames
[Pipeline] Generated 40 frames
[Pipeline] Generated 48 frames
[Pipeline] Generated 56 frames
[Pipeline] Generated 64 frames
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1203.000000 203.000000
[Debug] output-audio: [122880] first4: -0.001460 -0.002505 -0.002506 -0.001414
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
[Input] RefAudio: ../examples/freeman.wav
[Input] RefText: 213 chars: If you go into different cultures, they have different conce...
@@ -34,3 +139,38 @@ load_backend: loaded CPU backend from /mnt/workspace/qwentts.cpp/build/libggml-c
[Python] Audio: 120960 samples 24000 Hz 5.04s -> python/clone/clone-python.wav
[Quant] BF16 -> ../models/qwen-talker-1.7b-base-BF16.gguf + ../models/qwen-tokenizer-12hz-BF16.gguf
[GGML] Cmd: ../build/qwen-tts --model ../models/qwen-talker-1.7b-base-BF16.gguf --codec ../models/qwen-tokenizer-12hz-BF16.gguf --seed --text [...] --ref-audio ../examples/freeman.wav --ref-text [...] --lang english --max-new 64 --dump cpp/clone -o cpp/clone/clone-cpp.wav --greedy
[GGML] Audio: 122880 samples 24000 Hz 5.12s -> cpp/clone/clone-cpp.wav
[Cossim] PromptIDs exact: 100.00% (38 values)
[Cossim] RefCodes exact: 76.83% (3440 values)
[Cossim] Embed cos: 0.962709 max: 1.9437e-01 mean: 1.0686e-02
[Cossim] TrailingText cos: 1.000000 max: 1.0431e-06 mean: 9.3616e-09
[Cossim] TTSPadEmbed cos: 1.000000 max: 1.0431e-06 mean: 9.3616e-09
[Cossim] L0 cos: 0.983223 max: 7.7067e-01 mean: 1.3991e-02
[Cossim] L7 cos: 0.999989 max: 1.2859e+01 mean: 3.3234e-02
[Cossim] L14 cos: 0.999964 max: 1.2860e+01 mean: 7.1854e-02
[Cossim] L21 cos: 0.999532 max: 1.6955e+01 mean: 2.8042e-01
[Cossim] L27 cos: 0.998217 max: 4.0720e+01 mean: 6.4082e-01
[Cossim] Final cos: 0.990703 max: 1.5827e+01 mean: 2.0875e-01
[Cossim] Logits cos: 0.999316 max: 1.1188e+00 mean: 1.3424e-01
[Cossim] NextEmbStep0 cos: 0.541941 max: 2.4936e-01 mean: 5.0259e-02
[Cossim] TalkerHiddenStep1 cos: 0.520135 max: 2.0678e+01 mean: 1.9063e+00
[Cossim] MelHann cos: 1.000000 max: 1.7881e-07 mean: 3.2043e-08
[Cossim] MelBasis cos: 1.000000 max: 1.7881e-07 mean: 3.9379e-10
[Cossim] MelMag cos: 1.000000 max: 3.1307e-05 mean: 5.6396e-07
[Cossim] MelSpk cos: 1.000000 max: 2.5492e-03 mean: 9.3754e-06
[Cossim] SeanetInit cos: 1.000000 max: 9.1669e-04 mean: 1.6221e-05
[Cossim] SeanetResnet0 cos: 0.999961 max: 2.1108e-03 mean: 1.6910e-04
[Cossim] SeanetStage0 cos: 0.999983 max: 4.4128e-02 mean: 7.5091e-04
[Cossim] SeanetStage1 cos: 0.999862 max: 7.2962e-01 mean: 5.2996e-03
[Cossim] SeanetStage3 cos: 0.999994 max: 2.1304e+00 mean: 4.7641e-02
[Cossim] SeanetOut cos: 0.999953 max: 3.4988e-01 mean: 2.0273e-02
[Cossim] EncTransformer cos: 0.999963 max: 8.5211e-01 mean: 1.0424e-02
[Cossim] CodecPreFSQ cos: 0.999962 max: 4.6668e-01 mean: 2.2088e-02
[Cossim] SpkFrontend cos: 0.999997 max: 2.5180e-02 mean: 9.4975e-04
[Cossim] SpkBlock3 cos: 0.999991 max: 1.0191e-01 mean: 4.1523e-03
[Cossim] SpkMFA cos: 0.999979 max: 7.0797e-02 mean: 1.0370e-03
[Cossim] SpkASP cos: 1.000000 max: 2.3381e-03 mean: 1.5046e-04
[Cossim] SpeakerEmb cos: 1.000000 max: 6.5649e-04 mean: 1.2850e-04
[Cossim] CodesFull exact: 0.30% (1008 values)
[Cossim] Audio cos: -0.005066
[Cossim] WAV stft_cos: 0.148098 samples: 120960
tests/clone-Vulkan0-F32.log
+10 -10
View File
@@ -20,15 +20,15 @@ load_backend: loaded CPU backend from /mnt/workspace/qwentts.cpp/build/libggml-c
[Transformer] Loaded: 8 layers, hidden 512, heads 16/16, head_dim 64, FFN 1024, RoPE theta 10000, sliding window 72
[WeightCtx] Loaded 22 tensors, 80.1 MB into backend
[Upsample] Loaded: 2 blocks (2x ratio per block), channels 1024, dwconv kernel 7
[WeightCtx] Loaded 118 tensors, 199.8 MB into backend
[DAC] Loaded: 4 blocks (strides 8/5/4/3), 24 kHz mono out, weights 199.8 MB
[WeightCtx] Loaded 2 tensors, 6.0 MB into backend
[WeightCtx] Loaded 28 tensors, 48.2 MB into backend
[SEANet] Loaded: 4 stages (ratios 4/5/6/8), 64 -> 512 channels, weights 48.2 MB
[WeightCtx] Loaded 118 tensors, 142.9 MB into backend
[DAC] Loaded: 4 blocks (strides 8/5/4/3), 24 kHz mono out, weights 142.9 MB
[WeightCtx] Loaded 2 tensors, 3.0 MB into backend
[WeightCtx] Loaded 28 tensors, 24.1 MB into backend
[SEANet] Loaded: 4 stages (ratios 4/5/6/8), 64 -> 512 channels, weights 24.1 MB
[WeightCtx] Loaded 96 tensors, 96.1 MB into backend
[EncTransformer] Loaded: 8 layers, hidden 512, heads 8/8, head_dim 64, FFN 2048, RoPE theta 10000
[WeightCtx] Loaded 1 tensors, 4.0 MB into backend
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 4.0 MB
[WeightCtx] Loaded 1 tensors, 2.0 MB into backend
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Ready: hop 1920 samples @ 24000 Hz mono, 16 codebooks @ 12.5 Hz
@@ -121,7 +121,7 @@ load_backend: loaded CPU backend from /mnt/workspace/qwentts.cpp/build/libggml-c
[Pipeline] Generated 64 frames
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1203.000000 203.000000
[Debug] output-audio: [122880] first4: -0.001441 -0.002431 -0.002447 -0.001548
[Debug] output-audio: [122880] first4: -0.001441 -0.002432 -0.002447 -0.001549
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
@@ -172,5 +172,5 @@ load_backend: loaded CPU backend from /mnt/workspace/qwentts.cpp/build/libggml-c
[Cossim] SpkASP cos: 1.000000 max: 2.2933e-04 mean: 1.7414e-05
[Cossim] SpeakerEmb cos: 1.000000 max: 7.9088e-05 mean: 1.3439e-05
[Cossim] CodesFull exact: 0.20% (1008 values)
[Cossim] Audio cos: -0.005102
[Cossim] WAV stft_cos: 0.156900 samples: 120960
[Cossim] Audio cos: -0.005021
[Cossim] WAV stft_cos: 0.156799 samples: 120960