| { |
| "artifact_format": "orbitquant-v1", |
| "source_model_id": "Tongyi-MAI/Z-Image-Turbo", |
| "source_revision": "f332072aa78be7aecdf3ee76d5c247082da564a6", |
| "source_license": "apache-2.0", |
| "quant_method": "orbitquant", |
| "paper": "https://arxiv.org/abs/2607.02461", |
| "weight_bits": 4, |
| "activation_bits": 4, |
| "rotation": "rpbh", |
| "rotation_seed": 0, |
| "block_size": "paper", |
| "block_size_policy": "largest_power_of_two_dividing_dim", |
| "codebook": "lloyd_max", |
| "codebook_version": 2, |
| "row_norm_dtype": "bfloat16", |
| "runtime_mode": "auto_fused", |
| "activation_kernel_backend": "auto", |
| "activation_eps": 1e-10, |
| "adaln_group_size": 64, |
| "quantization_device": "cuda", |
| "weight_quantization_backend": "triton_cuda", |
| "quantization_staging_mode": "streaming", |
| "target_policy": "z_image", |
| "adaln_policy": "int4_rtn_group64_bf16_activation", |
| "quantized_modules": [ |
| "noise_refiner.0.attention.to_q", |
| "noise_refiner.0.attention.to_k", |
| "noise_refiner.0.attention.to_v", |
| "noise_refiner.0.attention.to_out.0", |
| "noise_refiner.0.feed_forward.w1", |
| "noise_refiner.0.feed_forward.w2", |
| "noise_refiner.0.feed_forward.w3", |
| "noise_refiner.1.attention.to_q", |
| "noise_refiner.1.attention.to_k", |
| "noise_refiner.1.attention.to_v", |
| "noise_refiner.1.attention.to_out.0", |
| "noise_refiner.1.feed_forward.w1", |
| "noise_refiner.1.feed_forward.w2", |
| "noise_refiner.1.feed_forward.w3", |
| "context_refiner.0.attention.to_q", |
| "context_refiner.0.attention.to_k", |
| "context_refiner.0.attention.to_v", |
| "context_refiner.0.attention.to_out.0", |
| "context_refiner.0.feed_forward.w1", |
| "context_refiner.0.feed_forward.w2", |
| "context_refiner.0.feed_forward.w3", |
| "context_refiner.1.attention.to_q", |
| "context_refiner.1.attention.to_k", |
| "context_refiner.1.attention.to_v", |
| "context_refiner.1.attention.to_out.0", |
| "context_refiner.1.feed_forward.w1", |
| "context_refiner.1.feed_forward.w2", |
| "context_refiner.1.feed_forward.w3", |
| "layers.0.attention.to_q", |
| "layers.0.attention.to_k", |
| "layers.0.attention.to_v", |
| "layers.0.attention.to_out.0", |
| "layers.0.feed_forward.w1", |
| "layers.0.feed_forward.w2", |
| "layers.0.feed_forward.w3", |
| "layers.1.attention.to_q", |
| "layers.1.attention.to_k", |
| "layers.1.attention.to_v", |
| "layers.1.attention.to_out.0", |
| "layers.1.feed_forward.w1", |
| "layers.1.feed_forward.w2", |
| "layers.1.feed_forward.w3", |
| "layers.2.attention.to_q", |
| "layers.2.attention.to_k", |
| "layers.2.attention.to_v", |
| "layers.2.attention.to_out.0", |
| "layers.2.feed_forward.w1", |
| "layers.2.feed_forward.w2", |
| "layers.2.feed_forward.w3", |
| "layers.3.attention.to_q", |
| "layers.3.attention.to_k", |
| "layers.3.attention.to_v", |
| "layers.3.attention.to_out.0", |
| "layers.3.feed_forward.w1", |
| "layers.3.feed_forward.w2", |
| "layers.3.feed_forward.w3", |
| "layers.4.attention.to_q", |
| "layers.4.attention.to_k", |
| "layers.4.attention.to_v", |
| "layers.4.attention.to_out.0", |
| "layers.4.feed_forward.w1", |
| "layers.4.feed_forward.w2", |
| "layers.4.feed_forward.w3", |
| "layers.5.attention.to_q", |
| "layers.5.attention.to_k", |
| "layers.5.attention.to_v", |
| "layers.5.attention.to_out.0", |
| "layers.5.feed_forward.w1", |
| "layers.5.feed_forward.w2", |
| "layers.5.feed_forward.w3", |
| "layers.6.attention.to_q", |
| "layers.6.attention.to_k", |
| "layers.6.attention.to_v", |
| "layers.6.attention.to_out.0", |
| "layers.6.feed_forward.w1", |
| "layers.6.feed_forward.w2", |
| "layers.6.feed_forward.w3", |
| "layers.7.attention.to_q", |
| "layers.7.attention.to_k", |
| "layers.7.attention.to_v", |
| "layers.7.attention.to_out.0", |
| "layers.7.feed_forward.w1", |
| "layers.7.feed_forward.w2", |
| "layers.7.feed_forward.w3", |
| "layers.8.attention.to_q", |
| "layers.8.attention.to_k", |
| "layers.8.attention.to_v", |
| "layers.8.attention.to_out.0", |
| "layers.8.feed_forward.w1", |
| "layers.8.feed_forward.w2", |
| "layers.8.feed_forward.w3", |
| "layers.9.attention.to_q", |
| "layers.9.attention.to_k", |
| "layers.9.attention.to_v", |
| "layers.9.attention.to_out.0", |
| "layers.9.feed_forward.w1", |
| "layers.9.feed_forward.w2", |
| "layers.9.feed_forward.w3", |
| "layers.10.attention.to_q", |
| "layers.10.attention.to_k", |
| "layers.10.attention.to_v", |
| "layers.10.attention.to_out.0", |
| "layers.10.feed_forward.w1", |
| "layers.10.feed_forward.w2", |
| "layers.10.feed_forward.w3", |
| "layers.11.attention.to_q", |
| "layers.11.attention.to_k", |
| "layers.11.attention.to_v", |
| "layers.11.attention.to_out.0", |
| "layers.11.feed_forward.w1", |
| "layers.11.feed_forward.w2", |
| "layers.11.feed_forward.w3", |
| "layers.12.attention.to_q", |
| "layers.12.attention.to_k", |
| "layers.12.attention.to_v", |
| "layers.12.attention.to_out.0", |
| "layers.12.feed_forward.w1", |
| "layers.12.feed_forward.w2", |
| "layers.12.feed_forward.w3", |
| "layers.13.attention.to_q", |
| "layers.13.attention.to_k", |
| "layers.13.attention.to_v", |
| "layers.13.attention.to_out.0", |
| "layers.13.feed_forward.w1", |
| "layers.13.feed_forward.w2", |
| "layers.13.feed_forward.w3", |
| "layers.14.attention.to_q", |
| "layers.14.attention.to_k", |
| "layers.14.attention.to_v", |
| "layers.14.attention.to_out.0", |
| "layers.14.feed_forward.w1", |
| "layers.14.feed_forward.w2", |
| "layers.14.feed_forward.w3", |
| "layers.15.attention.to_q", |
| "layers.15.attention.to_k", |
| "layers.15.attention.to_v", |
| "layers.15.attention.to_out.0", |
| "layers.15.feed_forward.w1", |
| "layers.15.feed_forward.w2", |
| "layers.15.feed_forward.w3", |
| "layers.16.attention.to_q", |
| "layers.16.attention.to_k", |
| "layers.16.attention.to_v", |
| "layers.16.attention.to_out.0", |
| "layers.16.feed_forward.w1", |
| "layers.16.feed_forward.w2", |
| "layers.16.feed_forward.w3", |
| "layers.17.attention.to_q", |
| "layers.17.attention.to_k", |
| "layers.17.attention.to_v", |
| "layers.17.attention.to_out.0", |
| "layers.17.feed_forward.w1", |
| "layers.17.feed_forward.w2", |
| "layers.17.feed_forward.w3", |
| "layers.18.attention.to_q", |
| "layers.18.attention.to_k", |
| "layers.18.attention.to_v", |
| "layers.18.attention.to_out.0", |
| "layers.18.feed_forward.w1", |
| "layers.18.feed_forward.w2", |
| "layers.18.feed_forward.w3", |
| "layers.19.attention.to_q", |
| "layers.19.attention.to_k", |
| "layers.19.attention.to_v", |
| "layers.19.attention.to_out.0", |
| "layers.19.feed_forward.w1", |
| "layers.19.feed_forward.w2", |
| "layers.19.feed_forward.w3", |
| "layers.20.attention.to_q", |
| "layers.20.attention.to_k", |
| "layers.20.attention.to_v", |
| "layers.20.attention.to_out.0", |
| "layers.20.feed_forward.w1", |
| "layers.20.feed_forward.w2", |
| "layers.20.feed_forward.w3", |
| "layers.21.attention.to_q", |
| "layers.21.attention.to_k", |
| "layers.21.attention.to_v", |
| "layers.21.attention.to_out.0", |
| "layers.21.feed_forward.w1", |
| "layers.21.feed_forward.w2", |
| "layers.21.feed_forward.w3", |
| "layers.22.attention.to_q", |
| "layers.22.attention.to_k", |
| "layers.22.attention.to_v", |
| "layers.22.attention.to_out.0", |
| "layers.22.feed_forward.w1", |
| "layers.22.feed_forward.w2", |
| "layers.22.feed_forward.w3", |
| "layers.23.attention.to_q", |
| "layers.23.attention.to_k", |
| "layers.23.attention.to_v", |
| "layers.23.attention.to_out.0", |
| "layers.23.feed_forward.w1", |
| "layers.23.feed_forward.w2", |
| "layers.23.feed_forward.w3", |
| "layers.24.attention.to_q", |
| "layers.24.attention.to_k", |
| "layers.24.attention.to_v", |
| "layers.24.attention.to_out.0", |
| "layers.24.feed_forward.w1", |
| "layers.24.feed_forward.w2", |
| "layers.24.feed_forward.w3", |
| "layers.25.attention.to_q", |
| "layers.25.attention.to_k", |
| "layers.25.attention.to_v", |
| "layers.25.attention.to_out.0", |
| "layers.25.feed_forward.w1", |
| "layers.25.feed_forward.w2", |
| "layers.25.feed_forward.w3", |
| "layers.26.attention.to_q", |
| "layers.26.attention.to_k", |
| "layers.26.attention.to_v", |
| "layers.26.attention.to_out.0", |
| "layers.26.feed_forward.w1", |
| "layers.26.feed_forward.w2", |
| "layers.26.feed_forward.w3", |
| "layers.27.attention.to_q", |
| "layers.27.attention.to_k", |
| "layers.27.attention.to_v", |
| "layers.27.attention.to_out.0", |
| "layers.27.feed_forward.w1", |
| "layers.27.feed_forward.w2", |
| "layers.27.feed_forward.w3", |
| "layers.28.attention.to_q", |
| "layers.28.attention.to_k", |
| "layers.28.attention.to_v", |
| "layers.28.attention.to_out.0", |
| "layers.28.feed_forward.w1", |
| "layers.28.feed_forward.w2", |
| "layers.28.feed_forward.w3", |
| "layers.29.attention.to_q", |
| "layers.29.attention.to_k", |
| "layers.29.attention.to_v", |
| "layers.29.attention.to_out.0", |
| "layers.29.feed_forward.w1", |
| "layers.29.feed_forward.w2", |
| "layers.29.feed_forward.w3" |
| ], |
| "adaln_modules": [ |
| "noise_refiner.0.adaLN_modulation.0", |
| "noise_refiner.1.adaLN_modulation.0", |
| "layers.0.adaLN_modulation.0", |
| "layers.1.adaLN_modulation.0", |
| "layers.2.adaLN_modulation.0", |
| "layers.3.adaLN_modulation.0", |
| "layers.4.adaLN_modulation.0", |
| "layers.5.adaLN_modulation.0", |
| "layers.6.adaLN_modulation.0", |
| "layers.7.adaLN_modulation.0", |
| "layers.8.adaLN_modulation.0", |
| "layers.9.adaLN_modulation.0", |
| "layers.10.adaLN_modulation.0", |
| "layers.11.adaLN_modulation.0", |
| "layers.12.adaLN_modulation.0", |
| "layers.13.adaLN_modulation.0", |
| "layers.14.adaLN_modulation.0", |
| "layers.15.adaLN_modulation.0", |
| "layers.16.adaLN_modulation.0", |
| "layers.17.adaLN_modulation.0", |
| "layers.18.adaLN_modulation.0", |
| "layers.19.adaLN_modulation.0", |
| "layers.20.adaLN_modulation.0", |
| "layers.21.adaLN_modulation.0", |
| "layers.22.adaLN_modulation.0", |
| "layers.23.adaLN_modulation.0", |
| "layers.24.adaLN_modulation.0", |
| "layers.25.adaLN_modulation.0", |
| "layers.26.adaLN_modulation.0", |
| "layers.27.adaLN_modulation.0", |
| "layers.28.adaLN_modulation.0", |
| "layers.29.adaLN_modulation.0" |
| ], |
| "skipped_modules": [ |
| "all_x_embedder.2-1", |
| "all_final_layer.2-1.linear", |
| "all_final_layer.2-1.adaLN_modulation.1", |
| "t_embedder.mlp.0", |
| "t_embedder.mlp.2", |
| "cap_embedder.1" |
| ], |
| "module_shapes": { |
| "x_pad_token": [ |
| 1, |
| 3840 |
| ], |
| "cap_pad_token": [ |
| 1, |
| 3840 |
| ], |
| "all_x_embedder.2-1.weight": [ |
| 3840, |
| 64 |
| ], |
| "all_x_embedder.2-1.bias": [ |
| 3840 |
| ], |
| "all_final_layer.2-1.linear.weight": [ |
| 64, |
| 3840 |
| ], |
| "all_final_layer.2-1.linear.bias": [ |
| 64 |
| ], |
| "all_final_layer.2-1.adaLN_modulation.1.weight": [ |
| 3840, |
| 256 |
| ], |
| "all_final_layer.2-1.adaLN_modulation.1.bias": [ |
| 3840 |
| ], |
| "noise_refiner.0.attention.norm_q.weight": [ |
| 128 |
| ], |
| "noise_refiner.0.attention.norm_k.weight": [ |
| 128 |
| ], |
| "noise_refiner.0.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "noise_refiner.0.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "noise_refiner.0.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "noise_refiner.0.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "noise_refiner.0.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "noise_refiner.0.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "noise_refiner.0.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "noise_refiner.0.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "noise_refiner.0.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "noise_refiner.0.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "noise_refiner.0.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "noise_refiner.0.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "noise_refiner.0.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "noise_refiner.0.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "noise_refiner.0.attention_norm1.weight": [ |
| 3840 |
| ], |
| "noise_refiner.0.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "noise_refiner.0.attention_norm2.weight": [ |
| 3840 |
| ], |
| "noise_refiner.0.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "noise_refiner.0.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "noise_refiner.0.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "noise_refiner.0.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "noise_refiner.1.attention.norm_q.weight": [ |
| 128 |
| ], |
| "noise_refiner.1.attention.norm_k.weight": [ |
| 128 |
| ], |
| "noise_refiner.1.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "noise_refiner.1.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "noise_refiner.1.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "noise_refiner.1.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "noise_refiner.1.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "noise_refiner.1.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "noise_refiner.1.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "noise_refiner.1.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "noise_refiner.1.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "noise_refiner.1.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "noise_refiner.1.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "noise_refiner.1.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "noise_refiner.1.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "noise_refiner.1.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "noise_refiner.1.attention_norm1.weight": [ |
| 3840 |
| ], |
| "noise_refiner.1.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "noise_refiner.1.attention_norm2.weight": [ |
| 3840 |
| ], |
| "noise_refiner.1.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "noise_refiner.1.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "noise_refiner.1.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "noise_refiner.1.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "context_refiner.0.attention.norm_q.weight": [ |
| 128 |
| ], |
| "context_refiner.0.attention.norm_k.weight": [ |
| 128 |
| ], |
| "context_refiner.0.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "context_refiner.0.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "context_refiner.0.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "context_refiner.0.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "context_refiner.0.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "context_refiner.0.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "context_refiner.0.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "context_refiner.0.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "context_refiner.0.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "context_refiner.0.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "context_refiner.0.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "context_refiner.0.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "context_refiner.0.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "context_refiner.0.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "context_refiner.0.attention_norm1.weight": [ |
| 3840 |
| ], |
| "context_refiner.0.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "context_refiner.0.attention_norm2.weight": [ |
| 3840 |
| ], |
| "context_refiner.0.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "context_refiner.1.attention.norm_q.weight": [ |
| 128 |
| ], |
| "context_refiner.1.attention.norm_k.weight": [ |
| 128 |
| ], |
| "context_refiner.1.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "context_refiner.1.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "context_refiner.1.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "context_refiner.1.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "context_refiner.1.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "context_refiner.1.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "context_refiner.1.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "context_refiner.1.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "context_refiner.1.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "context_refiner.1.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "context_refiner.1.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "context_refiner.1.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "context_refiner.1.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "context_refiner.1.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "context_refiner.1.attention_norm1.weight": [ |
| 3840 |
| ], |
| "context_refiner.1.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "context_refiner.1.attention_norm2.weight": [ |
| 3840 |
| ], |
| "context_refiner.1.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "t_embedder.mlp.0.weight": [ |
| 1024, |
| 256 |
| ], |
| "t_embedder.mlp.0.bias": [ |
| 1024 |
| ], |
| "t_embedder.mlp.2.weight": [ |
| 256, |
| 1024 |
| ], |
| "t_embedder.mlp.2.bias": [ |
| 256 |
| ], |
| "cap_embedder.0.weight": [ |
| 2560 |
| ], |
| "cap_embedder.1.weight": [ |
| 3840, |
| 2560 |
| ], |
| "cap_embedder.1.bias": [ |
| 3840 |
| ], |
| "layers.0.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.0.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.0.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.0.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.0.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.0.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.0.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.0.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.0.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.0.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.0.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.0.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.0.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.0.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.0.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.0.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.0.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.0.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.0.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.0.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.0.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.0.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.0.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.1.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.1.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.1.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.1.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.1.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.1.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.1.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.1.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.1.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.1.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.1.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.1.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.1.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.1.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.1.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.1.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.1.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.1.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.1.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.1.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.1.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.1.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.1.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.2.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.2.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.2.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.2.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.2.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.2.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.2.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.2.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.2.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.2.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.2.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.2.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.2.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.2.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.2.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.2.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.2.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.2.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.2.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.2.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.2.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.2.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.2.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.3.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.3.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.3.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.3.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.3.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.3.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.3.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.3.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.3.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.3.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.3.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.3.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.3.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.3.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.3.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.3.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.3.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.3.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.3.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.3.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.3.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.3.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.3.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.4.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.4.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.4.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.4.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.4.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.4.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.4.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.4.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.4.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.4.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.4.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.4.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.4.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.4.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.4.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.4.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.4.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.4.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.4.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.4.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.4.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.4.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.4.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.5.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.5.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.5.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.5.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.5.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.5.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.5.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.5.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.5.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.5.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.5.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.5.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.5.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.5.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.5.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.5.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.5.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.5.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.5.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.5.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.5.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.5.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.5.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.6.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.6.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.6.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.6.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.6.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.6.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.6.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.6.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.6.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.6.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.6.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.6.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.6.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.6.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.6.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.6.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.6.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.6.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.6.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.6.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.6.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.6.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.6.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.7.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.7.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.7.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.7.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.7.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.7.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.7.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.7.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.7.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.7.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.7.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.7.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.7.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.7.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.7.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.7.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.7.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.7.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.7.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.7.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.7.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.7.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.7.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.8.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.8.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.8.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.8.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.8.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.8.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.8.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.8.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.8.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.8.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.8.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.8.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.8.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.8.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.8.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.8.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.8.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.8.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.8.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.8.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.8.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.8.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.8.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.9.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.9.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.9.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.9.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.9.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.9.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.9.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.9.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.9.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.9.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.9.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.9.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.9.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.9.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.9.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.9.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.9.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.9.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.9.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.9.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.9.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.9.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.9.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.10.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.10.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.10.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.10.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.10.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.10.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.10.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.10.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.10.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.10.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.10.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.10.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.10.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.10.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.10.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.10.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.10.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.10.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.10.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.10.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.10.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.10.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.10.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.11.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.11.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.11.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.11.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.11.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.11.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.11.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.11.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.11.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.11.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.11.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.11.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.11.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.11.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.11.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.11.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.11.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.11.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.11.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.11.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.11.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.11.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.11.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.12.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.12.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.12.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.12.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.12.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.12.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.12.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.12.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.12.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.12.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.12.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.12.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.12.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.12.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.12.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.12.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.12.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.12.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.12.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.12.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.12.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.12.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.12.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.13.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.13.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.13.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.13.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.13.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.13.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.13.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.13.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.13.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.13.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.13.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.13.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.13.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.13.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.13.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.13.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.13.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.13.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.13.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.13.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.13.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.13.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.13.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.14.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.14.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.14.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.14.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.14.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.14.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.14.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.14.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.14.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.14.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.14.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.14.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.14.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.14.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.14.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.14.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.14.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.14.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.14.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.14.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.14.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.14.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.14.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.15.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.15.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.15.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.15.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.15.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.15.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.15.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.15.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.15.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.15.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.15.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.15.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.15.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.15.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.15.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.15.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.15.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.15.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.15.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.15.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.15.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.15.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.15.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.16.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.16.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.16.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.16.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.16.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.16.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.16.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.16.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.16.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.16.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.16.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.16.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.16.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.16.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.16.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.16.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.16.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.16.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.16.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.16.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.16.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.16.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.16.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.17.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.17.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.17.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.17.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.17.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.17.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.17.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.17.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.17.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.17.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.17.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.17.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.17.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.17.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.17.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.17.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.17.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.17.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.17.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.17.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.17.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.17.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.17.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.18.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.18.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.18.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.18.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.18.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.18.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.18.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.18.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.18.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.18.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.18.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.18.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.18.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.18.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.18.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.18.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.18.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.18.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.18.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.18.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.18.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.18.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.18.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.19.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.19.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.19.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.19.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.19.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.19.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.19.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.19.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.19.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.19.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.19.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.19.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.19.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.19.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.19.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.19.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.19.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.19.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.19.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.19.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.19.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.19.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.19.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.20.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.20.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.20.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.20.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.20.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.20.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.20.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.20.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.20.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.20.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.20.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.20.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.20.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.20.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.20.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.20.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.20.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.20.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.20.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.20.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.20.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.20.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.20.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.21.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.21.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.21.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.21.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.21.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.21.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.21.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.21.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.21.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.21.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.21.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.21.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.21.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.21.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.21.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.21.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.21.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.21.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.21.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.21.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.21.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.21.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.21.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.22.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.22.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.22.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.22.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.22.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.22.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.22.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.22.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.22.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.22.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.22.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.22.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.22.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.22.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.22.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.22.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.22.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.22.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.22.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.22.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.22.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.22.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.22.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.23.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.23.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.23.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.23.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.23.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.23.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.23.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.23.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.23.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.23.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.23.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.23.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.23.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.23.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.23.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.23.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.23.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.23.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.23.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.23.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.23.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.23.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.23.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.24.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.24.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.24.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.24.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.24.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.24.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.24.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.24.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.24.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.24.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.24.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.24.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.24.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.24.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.24.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.24.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.24.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.24.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.24.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.24.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.24.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.24.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.24.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.25.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.25.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.25.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.25.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.25.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.25.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.25.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.25.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.25.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.25.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.25.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.25.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.25.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.25.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.25.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.25.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.25.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.25.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.25.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.25.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.25.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.25.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.25.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.26.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.26.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.26.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.26.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.26.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.26.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.26.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.26.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.26.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.26.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.26.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.26.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.26.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.26.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.26.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.26.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.26.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.26.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.26.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.26.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.26.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.26.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.26.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.27.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.27.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.27.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.27.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.27.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.27.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.27.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.27.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.27.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.27.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.27.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.27.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.27.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.27.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.27.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.27.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.27.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.27.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.27.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.27.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.27.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.27.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.27.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.28.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.28.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.28.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.28.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.28.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.28.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.28.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.28.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.28.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.28.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.28.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.28.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.28.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.28.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.28.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.28.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.28.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.28.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.28.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.28.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.28.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.28.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.28.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ], |
| "layers.29.attention.norm_q.weight": [ |
| 128 |
| ], |
| "layers.29.attention.norm_k.weight": [ |
| 128 |
| ], |
| "layers.29.attention.to_q.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.29.attention.to_q.row_norms": [ |
| 3840 |
| ], |
| "layers.29.attention.to_k.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.29.attention.to_k.row_norms": [ |
| 3840 |
| ], |
| "layers.29.attention.to_v.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.29.attention.to_v.row_norms": [ |
| 3840 |
| ], |
| "layers.29.attention.to_out.0.packed_weight_indices": [ |
| 7372800 |
| ], |
| "layers.29.attention.to_out.0.row_norms": [ |
| 3840 |
| ], |
| "layers.29.feed_forward.w1.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.29.feed_forward.w1.row_norms": [ |
| 10240 |
| ], |
| "layers.29.feed_forward.w2.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.29.feed_forward.w2.row_norms": [ |
| 3840 |
| ], |
| "layers.29.feed_forward.w3.packed_weight_indices": [ |
| 19660800 |
| ], |
| "layers.29.feed_forward.w3.row_norms": [ |
| 10240 |
| ], |
| "layers.29.attention_norm1.weight": [ |
| 3840 |
| ], |
| "layers.29.ffn_norm1.weight": [ |
| 3840 |
| ], |
| "layers.29.attention_norm2.weight": [ |
| 3840 |
| ], |
| "layers.29.ffn_norm2.weight": [ |
| 3840 |
| ], |
| "layers.29.adaLN_modulation.0.bias": [ |
| 15360 |
| ], |
| "layers.29.adaLN_modulation.0.packed_weight": [ |
| 1966080 |
| ], |
| "layers.29.adaLN_modulation.0.scales": [ |
| 15360, |
| 4 |
| ] |
| }, |
| "checksums": { |
| "assets/image_generation_comparison_matrix.webp": "0b036acba67d919d8bea0a073838344ea9c600d8d77702c904b86efc544f726c", |
| "benchmark/summary.json": "b8b7c5a6a34ccb0fe3099ed0ba578505924ade93ba1fee46dc9b75b36169c6e0", |
| "model.safetensors": "677ef7a12e36dd92731ba9973b1b31eff3e3f03ff432a33ed6cbd127dddf16eb", |
| "model_index.json": "aa71fb7274ee17e5e94ff2434ec0aaa586bc74d09f077b5740ce4fb1a0af17c9", |
| "orbitquant_codebooks.safetensors": "80b16b36294630315e4c94a85e35b90d10b67e502cc1cd70bc7151fa44ff7a12", |
| "orbitquant_rotations.safetensors": "e1724c9394bf83ea05da6b457b8072d2bcb3d5a533dafce916da8deb517d3825", |
| "prompts.json": "0aee42e64abd66ff9293ec140df94e62bf7836eb96a21528a728282a0278c6ed", |
| "quantization_config.json": "afa4f2ede51d2f322a89ef593fbf50a6bdd9192bcfe160f6c5c77b4fe0117a3a" |
| } |
| } |
|
|
