| { |
| "metadata": { |
| "GGUF.tensor_count": 238, |
| "GGUF.kv_count": 34, |
| "general.architecture": "mossdual_istft", |
| "general.name": "ZONOS3 Mini MR-iSTFT decoder (keep2_plain, step 50000)", |
| "mossdual_istft.sample_rate": 44100, |
| "mossdual_istft.internal_sample_rate": 44160, |
| "mossdual_istft.frame_rate": 15, |
| "mossdual_istft.hop": 2940, |
| "mossdual_istft.internal_hop": 2944, |
| "mossdual_istft.n_codebooks": 10, |
| "mossdual_istft.codebook_size": 4096, |
| "mossdual_istft.latent_dim": 1024, |
| "mossdual_istft.semantic_dim": 768, |
| "mossdual_istft.sem_head.depth": 4, |
| "mossdual_istft.head.depth": 8, |
| "mossdual_istft.head.hidden": 2048, |
| "mossdual_istft.head.hop": 184, |
| "mossdual_istft.head.fir_kernel": 127, |
| "mossdual_istft.head.step": 50000, |
| "mossdual_istft.head.n_fft": [ |
| 736, |
| 2208 |
| ], |
| "mossdual_istft.head.max_log_magnitude": 4.605170249938965, |
| "mossdual_istft.headroom_db": 3.0, |
| "mossdual_istft.head.variant": "keep2_plain", |
| "mossdual_istft.ln_eps": 9.999999747378752e-06, |
| "mossdual_istft.convnext_eps": 9.999999974752427e-07, |
| "mossdual_istft.stage.in_dim": [ |
| 1024, |
| 640, |
| 384, |
| 384, |
| 384 |
| ], |
| "mossdual_istft.stage.out_dim": [ |
| 1280, |
| 768, |
| 768, |
| 768, |
| 0 |
| ], |
| "mossdual_istft.stage.d_model": [ |
| 1280, |
| 768, |
| 768, |
| 768, |
| 768 |
| ], |
| "mossdual_istft.stage.n_head": [ |
| 20, |
| 12, |
| 12, |
| 12, |
| 12 |
| ], |
| "mossdual_istft.stage.n_ff": [ |
| 5120, |
| 3072, |
| 3072, |
| 3072, |
| 3072 |
| ], |
| "mossdual_istft.stage.context": [ |
| 150, |
| 300, |
| 480, |
| 480, |
| 480 |
| ], |
| "mossdual_istft.stage.n_layer": [ |
| 2, |
| 2, |
| 2, |
| 2, |
| 2 |
| ], |
| "mossdual_istft.stage.patch": [ |
| 2, |
| 2, |
| 2, |
| 2, |
| 1 |
| ], |
| "mossdual_istft.resample.orig": 736, |
| "mossdual_istft.resample.new": 735, |
| "mossdual_istft.resample.width": 17 |
| }, |
| "tensors": { |
| "sem.table": { |
| "shape": [ |
| 4096, |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.0.table": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.0.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.1.table": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.2.table": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.3.table": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.3.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.4.table": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.4.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.5.table": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.5.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.6.table": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.6.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.7.table": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.7.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.8.table": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.8.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "res.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.0.dw.weight": { |
| "shape": [ |
| 768, |
| 7 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.0.dw.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.0.norm.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.0.norm.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.0.pw1.weight": { |
| "shape": [ |
| 2048, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "sem_head.0.pw1.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.0.pw2.weight": { |
| "shape": [ |
| 768, |
| 2048 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "sem_head.0.pw2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.1.dw.weight": { |
| "shape": [ |
| 768, |
| 7 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.1.dw.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.1.norm.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.1.norm.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.1.pw1.weight": { |
| "shape": [ |
| 2048, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "sem_head.1.pw1.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.1.pw2.weight": { |
| "shape": [ |
| 768, |
| 2048 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "sem_head.1.pw2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.2.dw.weight": { |
| "shape": [ |
| 768, |
| 7 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.2.dw.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.2.norm.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.2.norm.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.2.pw1.weight": { |
| "shape": [ |
| 2048, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "sem_head.2.pw1.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.2.pw2.weight": { |
| "shape": [ |
| 768, |
| 2048 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "sem_head.2.pw2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.3.dw.weight": { |
| "shape": [ |
| 768, |
| 7 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.3.dw.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.3.norm.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.3.norm.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.3.pw1.weight": { |
| "shape": [ |
| 2048, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "sem_head.3.pw1.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.3.pw2.weight": { |
| "shape": [ |
| 768, |
| 2048 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "sem_head.3.pw2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "sem_head.out.weight": { |
| "shape": [ |
| 1024, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "sem_head.out.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.0.in.weight": { |
| "shape": [ |
| 1280, |
| 1024 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.0.out.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.0.l0.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.0.l0.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.0.l0.qkv.weight": { |
| "shape": [ |
| 3840, |
| 1280 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.0.l0.wo.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.0.l0.attn_scale": { |
| "shape": [ |
| 1280 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.0.l0.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.0.l0.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.0.l0.ffn1.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.0.l0.ffn2.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.0.l0.ffn_scale": { |
| "shape": [ |
| 1280 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.0.l1.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.0.l1.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.0.l1.qkv.weight": { |
| "shape": [ |
| 3840, |
| 1280 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.0.l1.wo.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.0.l1.attn_scale": { |
| "shape": [ |
| 1280 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.0.l1.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.0.l1.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.0.l1.ffn1.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.0.l1.ffn2.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.0.l1.ffn_scale": { |
| "shape": [ |
| 1280 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.1.in.weight": { |
| "shape": [ |
| 768, |
| 640 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.1.out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.1.l0.norm1.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.1.l0.norm1.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.1.l0.qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.1.l0.wo.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.1.l0.attn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.1.l0.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.1.l0.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.1.l0.ffn1.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.1.l0.ffn2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.1.l0.ffn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.1.l1.norm1.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.1.l1.norm1.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.1.l1.qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.1.l1.wo.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.1.l1.attn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.1.l1.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.1.l1.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.1.l1.ffn1.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.1.l1.ffn2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.1.l1.ffn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.2.in.weight": { |
| "shape": [ |
| 768, |
| 384 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.2.out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.2.l0.norm1.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.2.l0.norm1.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.2.l0.qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.2.l0.wo.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.2.l0.attn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.2.l0.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.2.l0.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.2.l0.ffn1.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.2.l0.ffn2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.2.l0.ffn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.2.l1.norm1.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.2.l1.norm1.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.2.l1.qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.2.l1.wo.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.2.l1.attn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.2.l1.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.2.l1.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.2.l1.ffn1.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.2.l1.ffn2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.2.l1.ffn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.3.in.weight": { |
| "shape": [ |
| 768, |
| 384 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.3.out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.3.l0.norm1.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.3.l0.norm1.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.3.l0.qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.3.l0.wo.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.3.l0.attn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.3.l0.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.3.l0.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.3.l0.ffn1.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.3.l0.ffn2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.3.l0.ffn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.3.l1.norm1.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.3.l1.norm1.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.3.l1.qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.3.l1.wo.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.3.l1.attn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.3.l1.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.3.l1.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.3.l1.ffn1.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.3.l1.ffn2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.3.l1.ffn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.4.in.weight": { |
| "shape": [ |
| 768, |
| 384 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.4.l0.norm1.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.4.l0.norm1.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.4.l0.qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.4.l0.wo.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.4.l0.attn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.4.l0.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.4.l0.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.4.l0.ffn1.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.4.l0.ffn2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.4.l0.ffn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.4.l1.norm1.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.4.l1.norm1.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.4.l1.qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.4.l1.wo.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.4.l1.attn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.4.l1.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.4.l1.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "stage.4.l1.ffn1.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.4.l1.ffn2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "stage.4.l1.ffn_scale": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.0.dw.weight": { |
| "shape": [ |
| 768, |
| 7 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.0.dw.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.0.norm.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.0.norm.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.0.pw1.weight": { |
| "shape": [ |
| 2048, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.0.pw1.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.0.pw2.weight": { |
| "shape": [ |
| 768, |
| 2048 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.0.pw2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.1.dw.weight": { |
| "shape": [ |
| 768, |
| 7 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.1.dw.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.1.norm.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.1.norm.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.1.pw1.weight": { |
| "shape": [ |
| 2048, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.1.pw1.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.1.pw2.weight": { |
| "shape": [ |
| 768, |
| 2048 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.1.pw2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.2.dw.weight": { |
| "shape": [ |
| 768, |
| 7 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.2.dw.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.2.norm.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.2.norm.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.2.pw1.weight": { |
| "shape": [ |
| 2048, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.2.pw1.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.2.pw2.weight": { |
| "shape": [ |
| 768, |
| 2048 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.2.pw2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.3.dw.weight": { |
| "shape": [ |
| 768, |
| 7 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.3.dw.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.3.norm.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.3.norm.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.3.pw1.weight": { |
| "shape": [ |
| 2048, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.3.pw1.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.3.pw2.weight": { |
| "shape": [ |
| 768, |
| 2048 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.3.pw2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.4.dw.weight": { |
| "shape": [ |
| 768, |
| 7 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.4.dw.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.4.norm.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.4.norm.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.4.pw1.weight": { |
| "shape": [ |
| 2048, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.4.pw1.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.4.pw2.weight": { |
| "shape": [ |
| 768, |
| 2048 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.4.pw2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.5.dw.weight": { |
| "shape": [ |
| 768, |
| 7 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.5.dw.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.5.norm.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.5.norm.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.5.pw1.weight": { |
| "shape": [ |
| 2048, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.5.pw1.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.5.pw2.weight": { |
| "shape": [ |
| 768, |
| 2048 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.5.pw2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.6.dw.weight": { |
| "shape": [ |
| 768, |
| 7 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.6.dw.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.6.norm.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.6.norm.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.6.pw1.weight": { |
| "shape": [ |
| 2048, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.6.pw1.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.6.pw2.weight": { |
| "shape": [ |
| 768, |
| 2048 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.6.pw2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.7.dw.weight": { |
| "shape": [ |
| 768, |
| 7 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.7.dw.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.7.norm.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.7.norm.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.7.pw1.weight": { |
| "shape": [ |
| 2048, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.7.pw1.bias": { |
| "shape": [ |
| 2048 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.7.pw2.weight": { |
| "shape": [ |
| 768, |
| 2048 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.7.pw2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.norm.weight": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.norm.bias": { |
| "shape": [ |
| 768 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.out.0.weight": { |
| "shape": [ |
| 738, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.out.0.bias": { |
| "shape": [ |
| 738 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.out.1.weight": { |
| "shape": [ |
| 2210, |
| 768 |
| ], |
| "matrix": true, |
| "dtype": "float16", |
| "bits": null |
| }, |
| "head.out.1.bias": { |
| "shape": [ |
| 2210 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.mixer.weight": { |
| "shape": [ |
| 2, |
| 127 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.window.0": { |
| "shape": [ |
| 736 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "head.window.1": { |
| "shape": [ |
| 2208 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| }, |
| "resample.kernel": { |
| "shape": [ |
| 735, |
| 770 |
| ], |
| "matrix": false, |
| "dtype": "float32" |
| } |
| } |
| } |
|
|
