diff --git "a/image-large/sparseStructureFlow/manifest.json" "b/image-large/sparseStructureFlow/manifest.json" new file mode 100644--- /dev/null +++ "b/image-large/sparseStructureFlow/manifest.json" @@ -0,0 +1,10104 @@ +{ + "schema": "trellis-webgpu-pack", + "version": 1, + "stage": "sparseStructureFlow", + "upstreamRevision": "25e0d31ffbebe4b5a97464dd851910efc3002d96", + "shards": [ + { + "file": "stage-000.bin", + "bytes": 66705408, + "sha256": "f0784b06e2b1de485881fb7382f4fb5f12ad1afb3bbf5a57cdbba9701b0a84b2" + }, + { + "file": "stage-001.bin", + "bytes": 64538624, + "sha256": "44b727fea9979fd8f0d93a9f8edb3e8c37d380bcabd5fb4bd6cdc05903fa84e7" + }, + { + "file": "stage-002.bin", + "bytes": 65628160, + "sha256": "2569e48e2700140e674038592de2f95440f1ed4f7897305300fe1cfd5f7b7e19" + }, + { + "file": "stage-003.bin", + "bytes": 65617920, + "sha256": "3913321cb4f6dd6fc47619fc5c16a1369d28a5342fde66afa6f55d570468966e" + }, + { + "file": "stage-004.bin", + "bytes": 62349312, + "sha256": "187edbe3a8d1258a2b2f15307a0c19a60608c9111aa98e13764935e7ba645f4f" + }, + { + "file": "stage-005.bin", + "bytes": 66693120, + "sha256": "a1b1300cd56ac0379cbde82c8e5cf7ee17b81610db5a9bc36a6a8d0c9a36f7a7" + }, + { + "file": "stage-006.bin", + "bytes": 64538624, + "sha256": "521267c28bfa739213e478c017db00eee2b42572c8b3f9c1e55808b65d74dd45" + }, + { + "file": "stage-007.bin", + "bytes": 65628160, + "sha256": "0a1e8e93c5f80670cc0bca860d26a4d92e82c307707a41ef1bc0ced957b5e56f" + }, + { + "file": "stage-008.bin", + "bytes": 65617920, + "sha256": "810a6c3f441ec717975ad7173e269178020f91ca307e703cb0d01f454a1c886e" + }, + { + "file": "stage-009.bin", + "bytes": 62349312, + "sha256": "215a57b9568b49fd7b049780ddd563bb109fb72dc8754c7798420e03bb744657" + }, + { + "file": "stage-010.bin", + "bytes": 66693120, + "sha256": "407fa67941be2024845c160fe3ccbe8c684cfe906e53b2c62704b97012a80898" + }, + { + "file": "stage-011.bin", + "bytes": 64538624, + "sha256": "3fb0f00dbe8b4a45adcab7b36f298d1f975fef4f9249ac35bc4c8f1b37bd800f" + }, + { + "file": "stage-012.bin", + "bytes": 65628160, + "sha256": "ea04edc07730f3dbd4f0df87a07d0027a57d7ed7513475f96878c90fc34d8f75" + }, + { + "file": "stage-013.bin", + "bytes": 26177792, + "sha256": "cb7118a3f8a9dbbd2b7a4a4f83ef707ac2dbea1ce3f6dedb51457d4e857d969a" + } + ], + "tensors": { + "blocks.0.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 0, + "bytes": 12288 + } + }, + "blocks.0.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 0, + "offset": 12288, + "bytes": 6291456 + }, + "base_scales": { + "shard": 0, + "offset": 6303744, + "bytes": 196608 + }, + "residual_data": { + "shard": 0, + "offset": 6500352, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 0, + "offset": 9646080, + "bytes": 196608 + } + }, + "blocks.0.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 9842688, + "bytes": 4096 + } + }, + "blocks.0.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 0, + "offset": 9846784, + "bytes": 2097152 + }, + "base_scales": { + "shard": 0, + "offset": 11943936, + "bytes": 65536 + }, + "residual_data": { + "shard": 0, + "offset": 12009472, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 0, + "offset": 13058048, + "bytes": 65536 + } + }, + "blocks.0.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 13123584, + "bytes": 2048 + } + }, + "blocks.0.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 0, + "offset": 13125632, + "bytes": 1048576 + }, + "base_scales": { + "shard": 0, + "offset": 14174208, + "bytes": 32768 + }, + "residual_data": { + "shard": 0, + "offset": 14206976, + "bytes": 524288 + }, + "residual_scales": { + "shard": 0, + "offset": 14731264, + "bytes": 32768 + } + }, + "blocks.0.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 14764032, + "bytes": 2048 + } + }, + "blocks.0.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 0, + "offset": 14766080, + "bytes": 1048576 + }, + "base_scales": { + "shard": 0, + "offset": 15814656, + "bytes": 32768 + }, + "residual_data": { + "shard": 0, + "offset": 15847424, + "bytes": 524288 + }, + "residual_scales": { + "shard": 0, + "offset": 16371712, + "bytes": 32768 + } + }, + "blocks.0.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 16404480, + "bytes": 8192 + } + }, + "blocks.0.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 0, + "offset": 16412672, + "bytes": 4194304 + }, + "base_scales": { + "shard": 0, + "offset": 20606976, + "bytes": 131072 + }, + "residual_data": { + "shard": 0, + "offset": 20738048, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 0, + "offset": 22835200, + "bytes": 131072 + } + }, + "blocks.0.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 22966272, + "bytes": 2048 + } + }, + "blocks.0.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 0, + "offset": 22968320, + "bytes": 4194304 + }, + "base_scales": { + "shard": 0, + "offset": 27162624, + "bytes": 131072 + }, + "residual_data": { + "shard": 0, + "offset": 27293696, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 0, + "offset": 29390848, + "bytes": 131072 + } + }, + "blocks.0.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 29521920, + "bytes": 2048 + } + }, + "blocks.0.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 29523968, + "bytes": 2048 + } + }, + "blocks.0.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 29526016, + "bytes": 2048 + } + }, + "blocks.0.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 29528064, + "bytes": 2048 + } + }, + "blocks.0.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 29530112, + "bytes": 2048 + } + }, + "blocks.0.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 0, + "offset": 29532160, + "bytes": 1048576 + }, + "base_scales": { + "shard": 0, + "offset": 30580736, + "bytes": 32768 + }, + "residual_data": { + "shard": 0, + "offset": 30613504, + "bytes": 524288 + }, + "residual_scales": { + "shard": 0, + "offset": 31137792, + "bytes": 32768 + } + }, + "blocks.0.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 31170560, + "bytes": 6144 + } + }, + "blocks.0.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 0, + "offset": 31176704, + "bytes": 3145728 + }, + "base_scales": { + "shard": 0, + "offset": 34322432, + "bytes": 98304 + }, + "residual_data": { + "shard": 0, + "offset": 34420736, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 0, + "offset": 35993600, + "bytes": 98304 + } + }, + "blocks.1.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 36091904, + "bytes": 12288 + } + }, + "blocks.1.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 0, + "offset": 36104192, + "bytes": 6291456 + }, + "base_scales": { + "shard": 0, + "offset": 42395648, + "bytes": 196608 + }, + "residual_data": { + "shard": 0, + "offset": 42592256, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 0, + "offset": 45737984, + "bytes": 196608 + } + }, + "blocks.1.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 45934592, + "bytes": 4096 + } + }, + "blocks.1.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 0, + "offset": 45938688, + "bytes": 2097152 + }, + "base_scales": { + "shard": 0, + "offset": 48035840, + "bytes": 65536 + }, + "residual_data": { + "shard": 0, + "offset": 48101376, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 0, + "offset": 49149952, + "bytes": 65536 + } + }, + "blocks.1.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 49215488, + "bytes": 2048 + } + }, + "blocks.1.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 0, + "offset": 49217536, + "bytes": 1048576 + }, + "base_scales": { + "shard": 0, + "offset": 50266112, + "bytes": 32768 + }, + "residual_data": { + "shard": 0, + "offset": 50298880, + "bytes": 524288 + }, + "residual_scales": { + "shard": 0, + "offset": 50823168, + "bytes": 32768 + } + }, + "blocks.1.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 50855936, + "bytes": 2048 + } + }, + "blocks.1.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 0, + "offset": 50857984, + "bytes": 1048576 + }, + "base_scales": { + "shard": 0, + "offset": 51906560, + "bytes": 32768 + }, + "residual_data": { + "shard": 0, + "offset": 51939328, + "bytes": 524288 + }, + "residual_scales": { + "shard": 0, + "offset": 52463616, + "bytes": 32768 + } + }, + "blocks.1.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 52496384, + "bytes": 8192 + } + }, + "blocks.1.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 0, + "offset": 52504576, + "bytes": 4194304 + }, + "base_scales": { + "shard": 0, + "offset": 56698880, + "bytes": 131072 + }, + "residual_data": { + "shard": 0, + "offset": 56829952, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 0, + "offset": 58927104, + "bytes": 131072 + } + }, + "blocks.1.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 59058176, + "bytes": 2048 + } + }, + "blocks.1.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 0, + "offset": 59060224, + "bytes": 4194304 + }, + "base_scales": { + "shard": 0, + "offset": 63254528, + "bytes": 131072 + }, + "residual_data": { + "shard": 0, + "offset": 63385600, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 0, + "offset": 65482752, + "bytes": 131072 + } + }, + "blocks.1.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 65613824, + "bytes": 2048 + } + }, + "blocks.1.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 65615872, + "bytes": 2048 + } + }, + "blocks.1.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 65617920, + "bytes": 2048 + } + }, + "blocks.1.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 65619968, + "bytes": 2048 + } + }, + "blocks.1.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 0, + "offset": 65622016, + "bytes": 2048 + } + }, + "blocks.1.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 0, + "offset": 65624064, + "bytes": 1048576 + }, + "base_scales": { + "shard": 0, + "offset": 66672640, + "bytes": 32768 + }, + "residual_data": { + "shard": 1, + "offset": 0, + "bytes": 524288 + }, + "residual_scales": { + "shard": 1, + "offset": 524288, + "bytes": 32768 + } + }, + "blocks.1.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 557056, + "bytes": 6144 + } + }, + "blocks.1.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 1, + "offset": 563200, + "bytes": 3145728 + }, + "base_scales": { + "shard": 1, + "offset": 3708928, + "bytes": 98304 + }, + "residual_data": { + "shard": 1, + "offset": 3807232, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 1, + "offset": 5380096, + "bytes": 98304 + } + }, + "blocks.10.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 5478400, + "bytes": 12288 + } + }, + "blocks.10.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 1, + "offset": 5490688, + "bytes": 6291456 + }, + "base_scales": { + "shard": 1, + "offset": 11782144, + "bytes": 196608 + }, + "residual_data": { + "shard": 1, + "offset": 11978752, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 1, + "offset": 15124480, + "bytes": 196608 + } + }, + "blocks.10.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 15321088, + "bytes": 4096 + } + }, + "blocks.10.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 1, + "offset": 15325184, + "bytes": 2097152 + }, + "base_scales": { + "shard": 1, + "offset": 17422336, + "bytes": 65536 + }, + "residual_data": { + "shard": 1, + "offset": 17487872, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 1, + "offset": 18536448, + "bytes": 65536 + } + }, + "blocks.10.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 18601984, + "bytes": 2048 + } + }, + "blocks.10.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 1, + "offset": 18604032, + "bytes": 1048576 + }, + "base_scales": { + "shard": 1, + "offset": 19652608, + "bytes": 32768 + }, + "residual_data": { + "shard": 1, + "offset": 19685376, + "bytes": 524288 + }, + "residual_scales": { + "shard": 1, + "offset": 20209664, + "bytes": 32768 + } + }, + "blocks.10.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 20242432, + "bytes": 2048 + } + }, + "blocks.10.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 1, + "offset": 20244480, + "bytes": 1048576 + }, + "base_scales": { + "shard": 1, + "offset": 21293056, + "bytes": 32768 + }, + "residual_data": { + "shard": 1, + "offset": 21325824, + "bytes": 524288 + }, + "residual_scales": { + "shard": 1, + "offset": 21850112, + "bytes": 32768 + } + }, + "blocks.10.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 21882880, + "bytes": 8192 + } + }, + "blocks.10.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 1, + "offset": 21891072, + "bytes": 4194304 + }, + "base_scales": { + "shard": 1, + "offset": 26085376, + "bytes": 131072 + }, + "residual_data": { + "shard": 1, + "offset": 26216448, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 1, + "offset": 28313600, + "bytes": 131072 + } + }, + "blocks.10.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 28444672, + "bytes": 2048 + } + }, + "blocks.10.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 1, + "offset": 28446720, + "bytes": 4194304 + }, + "base_scales": { + "shard": 1, + "offset": 32641024, + "bytes": 131072 + }, + "residual_data": { + "shard": 1, + "offset": 32772096, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 1, + "offset": 34869248, + "bytes": 131072 + } + }, + "blocks.10.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 35000320, + "bytes": 2048 + } + }, + "blocks.10.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 35002368, + "bytes": 2048 + } + }, + "blocks.10.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 35004416, + "bytes": 2048 + } + }, + "blocks.10.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 35006464, + "bytes": 2048 + } + }, + "blocks.10.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 35008512, + "bytes": 2048 + } + }, + "blocks.10.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 1, + "offset": 35010560, + "bytes": 1048576 + }, + "base_scales": { + "shard": 1, + "offset": 36059136, + "bytes": 32768 + }, + "residual_data": { + "shard": 1, + "offset": 36091904, + "bytes": 524288 + }, + "residual_scales": { + "shard": 1, + "offset": 36616192, + "bytes": 32768 + } + }, + "blocks.10.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 36648960, + "bytes": 6144 + } + }, + "blocks.10.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 1, + "offset": 36655104, + "bytes": 3145728 + }, + "base_scales": { + "shard": 1, + "offset": 39800832, + "bytes": 98304 + }, + "residual_data": { + "shard": 1, + "offset": 39899136, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 1, + "offset": 41472000, + "bytes": 98304 + } + }, + "blocks.11.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 41570304, + "bytes": 12288 + } + }, + "blocks.11.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 1, + "offset": 41582592, + "bytes": 6291456 + }, + "base_scales": { + "shard": 1, + "offset": 47874048, + "bytes": 196608 + }, + "residual_data": { + "shard": 1, + "offset": 48070656, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 1, + "offset": 51216384, + "bytes": 196608 + } + }, + "blocks.11.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 51412992, + "bytes": 4096 + } + }, + "blocks.11.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 1, + "offset": 51417088, + "bytes": 2097152 + }, + "base_scales": { + "shard": 1, + "offset": 53514240, + "bytes": 65536 + }, + "residual_data": { + "shard": 1, + "offset": 53579776, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 1, + "offset": 54628352, + "bytes": 65536 + } + }, + "blocks.11.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 54693888, + "bytes": 2048 + } + }, + "blocks.11.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 1, + "offset": 54695936, + "bytes": 1048576 + }, + "base_scales": { + "shard": 1, + "offset": 55744512, + "bytes": 32768 + }, + "residual_data": { + "shard": 1, + "offset": 55777280, + "bytes": 524288 + }, + "residual_scales": { + "shard": 1, + "offset": 56301568, + "bytes": 32768 + } + }, + "blocks.11.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 56334336, + "bytes": 2048 + } + }, + "blocks.11.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 1, + "offset": 56336384, + "bytes": 1048576 + }, + "base_scales": { + "shard": 1, + "offset": 57384960, + "bytes": 32768 + }, + "residual_data": { + "shard": 1, + "offset": 57417728, + "bytes": 524288 + }, + "residual_scales": { + "shard": 1, + "offset": 57942016, + "bytes": 32768 + } + }, + "blocks.11.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 57974784, + "bytes": 8192 + } + }, + "blocks.11.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 1, + "offset": 57982976, + "bytes": 4194304 + }, + "base_scales": { + "shard": 1, + "offset": 62177280, + "bytes": 131072 + }, + "residual_data": { + "shard": 1, + "offset": 62308352, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 1, + "offset": 64405504, + "bytes": 131072 + } + }, + "blocks.11.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 1, + "offset": 64536576, + "bytes": 2048 + } + }, + "blocks.11.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 2, + "offset": 0, + "bytes": 4194304 + }, + "base_scales": { + "shard": 2, + "offset": 4194304, + "bytes": 131072 + }, + "residual_data": { + "shard": 2, + "offset": 4325376, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 2, + "offset": 6422528, + "bytes": 131072 + } + }, + "blocks.11.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 6553600, + "bytes": 2048 + } + }, + "blocks.11.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 6555648, + "bytes": 2048 + } + }, + "blocks.11.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 6557696, + "bytes": 2048 + } + }, + "blocks.11.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 6559744, + "bytes": 2048 + } + }, + "blocks.11.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 6561792, + "bytes": 2048 + } + }, + "blocks.11.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 2, + "offset": 6563840, + "bytes": 1048576 + }, + "base_scales": { + "shard": 2, + "offset": 7612416, + "bytes": 32768 + }, + "residual_data": { + "shard": 2, + "offset": 7645184, + "bytes": 524288 + }, + "residual_scales": { + "shard": 2, + "offset": 8169472, + "bytes": 32768 + } + }, + "blocks.11.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 8202240, + "bytes": 6144 + } + }, + "blocks.11.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 2, + "offset": 8208384, + "bytes": 3145728 + }, + "base_scales": { + "shard": 2, + "offset": 11354112, + "bytes": 98304 + }, + "residual_data": { + "shard": 2, + "offset": 11452416, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 2, + "offset": 13025280, + "bytes": 98304 + } + }, + "blocks.12.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 13123584, + "bytes": 12288 + } + }, + "blocks.12.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 2, + "offset": 13135872, + "bytes": 6291456 + }, + "base_scales": { + "shard": 2, + "offset": 19427328, + "bytes": 196608 + }, + "residual_data": { + "shard": 2, + "offset": 19623936, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 2, + "offset": 22769664, + "bytes": 196608 + } + }, + "blocks.12.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 22966272, + "bytes": 4096 + } + }, + "blocks.12.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 2, + "offset": 22970368, + "bytes": 2097152 + }, + "base_scales": { + "shard": 2, + "offset": 25067520, + "bytes": 65536 + }, + "residual_data": { + "shard": 2, + "offset": 25133056, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 2, + "offset": 26181632, + "bytes": 65536 + } + }, + "blocks.12.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 26247168, + "bytes": 2048 + } + }, + "blocks.12.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 2, + "offset": 26249216, + "bytes": 1048576 + }, + "base_scales": { + "shard": 2, + "offset": 27297792, + "bytes": 32768 + }, + "residual_data": { + "shard": 2, + "offset": 27330560, + "bytes": 524288 + }, + "residual_scales": { + "shard": 2, + "offset": 27854848, + "bytes": 32768 + } + }, + "blocks.12.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 27887616, + "bytes": 2048 + } + }, + "blocks.12.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 2, + "offset": 27889664, + "bytes": 1048576 + }, + "base_scales": { + "shard": 2, + "offset": 28938240, + "bytes": 32768 + }, + "residual_data": { + "shard": 2, + "offset": 28971008, + "bytes": 524288 + }, + "residual_scales": { + "shard": 2, + "offset": 29495296, + "bytes": 32768 + } + }, + "blocks.12.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 29528064, + "bytes": 8192 + } + }, + "blocks.12.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 2, + "offset": 29536256, + "bytes": 4194304 + }, + "base_scales": { + "shard": 2, + "offset": 33730560, + "bytes": 131072 + }, + "residual_data": { + "shard": 2, + "offset": 33861632, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 2, + "offset": 35958784, + "bytes": 131072 + } + }, + "blocks.12.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 36089856, + "bytes": 2048 + } + }, + "blocks.12.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 2, + "offset": 36091904, + "bytes": 4194304 + }, + "base_scales": { + "shard": 2, + "offset": 40286208, + "bytes": 131072 + }, + "residual_data": { + "shard": 2, + "offset": 40417280, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 2, + "offset": 42514432, + "bytes": 131072 + } + }, + "blocks.12.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 42645504, + "bytes": 2048 + } + }, + "blocks.12.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 42647552, + "bytes": 2048 + } + }, + "blocks.12.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 42649600, + "bytes": 2048 + } + }, + "blocks.12.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 42651648, + "bytes": 2048 + } + }, + "blocks.12.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 42653696, + "bytes": 2048 + } + }, + "blocks.12.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 2, + "offset": 42655744, + "bytes": 1048576 + }, + "base_scales": { + "shard": 2, + "offset": 43704320, + "bytes": 32768 + }, + "residual_data": { + "shard": 2, + "offset": 43737088, + "bytes": 524288 + }, + "residual_scales": { + "shard": 2, + "offset": 44261376, + "bytes": 32768 + } + }, + "blocks.12.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 44294144, + "bytes": 6144 + } + }, + "blocks.12.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 2, + "offset": 44300288, + "bytes": 3145728 + }, + "base_scales": { + "shard": 2, + "offset": 47446016, + "bytes": 98304 + }, + "residual_data": { + "shard": 2, + "offset": 47544320, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 2, + "offset": 49117184, + "bytes": 98304 + } + }, + "blocks.13.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 49215488, + "bytes": 12288 + } + }, + "blocks.13.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 2, + "offset": 49227776, + "bytes": 6291456 + }, + "base_scales": { + "shard": 2, + "offset": 55519232, + "bytes": 196608 + }, + "residual_data": { + "shard": 2, + "offset": 55715840, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 2, + "offset": 58861568, + "bytes": 196608 + } + }, + "blocks.13.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 59058176, + "bytes": 4096 + } + }, + "blocks.13.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 2, + "offset": 59062272, + "bytes": 2097152 + }, + "base_scales": { + "shard": 2, + "offset": 61159424, + "bytes": 65536 + }, + "residual_data": { + "shard": 2, + "offset": 61224960, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 2, + "offset": 62273536, + "bytes": 65536 + } + }, + "blocks.13.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 62339072, + "bytes": 2048 + } + }, + "blocks.13.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 2, + "offset": 62341120, + "bytes": 1048576 + }, + "base_scales": { + "shard": 2, + "offset": 63389696, + "bytes": 32768 + }, + "residual_data": { + "shard": 2, + "offset": 63422464, + "bytes": 524288 + }, + "residual_scales": { + "shard": 2, + "offset": 63946752, + "bytes": 32768 + } + }, + "blocks.13.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 63979520, + "bytes": 2048 + } + }, + "blocks.13.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 2, + "offset": 63981568, + "bytes": 1048576 + }, + "base_scales": { + "shard": 2, + "offset": 65030144, + "bytes": 32768 + }, + "residual_data": { + "shard": 2, + "offset": 65062912, + "bytes": 524288 + }, + "residual_scales": { + "shard": 2, + "offset": 65587200, + "bytes": 32768 + } + }, + "blocks.13.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 2, + "offset": 65619968, + "bytes": 8192 + } + }, + "blocks.13.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 3, + "offset": 0, + "bytes": 4194304 + }, + "base_scales": { + "shard": 3, + "offset": 4194304, + "bytes": 131072 + }, + "residual_data": { + "shard": 3, + "offset": 4325376, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 3, + "offset": 6422528, + "bytes": 131072 + } + }, + "blocks.13.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 6553600, + "bytes": 2048 + } + }, + "blocks.13.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 3, + "offset": 6555648, + "bytes": 4194304 + }, + "base_scales": { + "shard": 3, + "offset": 10749952, + "bytes": 131072 + }, + "residual_data": { + "shard": 3, + "offset": 10881024, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 3, + "offset": 12978176, + "bytes": 131072 + } + }, + "blocks.13.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 13109248, + "bytes": 2048 + } + }, + "blocks.13.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 13111296, + "bytes": 2048 + } + }, + "blocks.13.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 13113344, + "bytes": 2048 + } + }, + "blocks.13.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 13115392, + "bytes": 2048 + } + }, + "blocks.13.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 13117440, + "bytes": 2048 + } + }, + "blocks.13.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 3, + "offset": 13119488, + "bytes": 1048576 + }, + "base_scales": { + "shard": 3, + "offset": 14168064, + "bytes": 32768 + }, + "residual_data": { + "shard": 3, + "offset": 14200832, + "bytes": 524288 + }, + "residual_scales": { + "shard": 3, + "offset": 14725120, + "bytes": 32768 + } + }, + "blocks.13.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 14757888, + "bytes": 6144 + } + }, + "blocks.13.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 3, + "offset": 14764032, + "bytes": 3145728 + }, + "base_scales": { + "shard": 3, + "offset": 17909760, + "bytes": 98304 + }, + "residual_data": { + "shard": 3, + "offset": 18008064, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 3, + "offset": 19580928, + "bytes": 98304 + } + }, + "blocks.14.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 19679232, + "bytes": 12288 + } + }, + "blocks.14.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 3, + "offset": 19691520, + "bytes": 6291456 + }, + "base_scales": { + "shard": 3, + "offset": 25982976, + "bytes": 196608 + }, + "residual_data": { + "shard": 3, + "offset": 26179584, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 3, + "offset": 29325312, + "bytes": 196608 + } + }, + "blocks.14.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 29521920, + "bytes": 4096 + } + }, + "blocks.14.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 3, + "offset": 29526016, + "bytes": 2097152 + }, + "base_scales": { + "shard": 3, + "offset": 31623168, + "bytes": 65536 + }, + "residual_data": { + "shard": 3, + "offset": 31688704, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 3, + "offset": 32737280, + "bytes": 65536 + } + }, + "blocks.14.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 32802816, + "bytes": 2048 + } + }, + "blocks.14.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 3, + "offset": 32804864, + "bytes": 1048576 + }, + "base_scales": { + "shard": 3, + "offset": 33853440, + "bytes": 32768 + }, + "residual_data": { + "shard": 3, + "offset": 33886208, + "bytes": 524288 + }, + "residual_scales": { + "shard": 3, + "offset": 34410496, + "bytes": 32768 + } + }, + "blocks.14.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 34443264, + "bytes": 2048 + } + }, + "blocks.14.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 3, + "offset": 34445312, + "bytes": 1048576 + }, + "base_scales": { + "shard": 3, + "offset": 35493888, + "bytes": 32768 + }, + "residual_data": { + "shard": 3, + "offset": 35526656, + "bytes": 524288 + }, + "residual_scales": { + "shard": 3, + "offset": 36050944, + "bytes": 32768 + } + }, + "blocks.14.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 36083712, + "bytes": 8192 + } + }, + "blocks.14.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 3, + "offset": 36091904, + "bytes": 4194304 + }, + "base_scales": { + "shard": 3, + "offset": 40286208, + "bytes": 131072 + }, + "residual_data": { + "shard": 3, + "offset": 40417280, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 3, + "offset": 42514432, + "bytes": 131072 + } + }, + "blocks.14.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 42645504, + "bytes": 2048 + } + }, + "blocks.14.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 3, + "offset": 42647552, + "bytes": 4194304 + }, + "base_scales": { + "shard": 3, + "offset": 46841856, + "bytes": 131072 + }, + "residual_data": { + "shard": 3, + "offset": 46972928, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 3, + "offset": 49070080, + "bytes": 131072 + } + }, + "blocks.14.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 49201152, + "bytes": 2048 + } + }, + "blocks.14.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 49203200, + "bytes": 2048 + } + }, + "blocks.14.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 49205248, + "bytes": 2048 + } + }, + "blocks.14.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 49207296, + "bytes": 2048 + } + }, + "blocks.14.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 49209344, + "bytes": 2048 + } + }, + "blocks.14.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 3, + "offset": 49211392, + "bytes": 1048576 + }, + "base_scales": { + "shard": 3, + "offset": 50259968, + "bytes": 32768 + }, + "residual_data": { + "shard": 3, + "offset": 50292736, + "bytes": 524288 + }, + "residual_scales": { + "shard": 3, + "offset": 50817024, + "bytes": 32768 + } + }, + "blocks.14.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 50849792, + "bytes": 6144 + } + }, + "blocks.14.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 3, + "offset": 50855936, + "bytes": 3145728 + }, + "base_scales": { + "shard": 3, + "offset": 54001664, + "bytes": 98304 + }, + "residual_data": { + "shard": 3, + "offset": 54099968, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 3, + "offset": 55672832, + "bytes": 98304 + } + }, + "blocks.15.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 55771136, + "bytes": 12288 + } + }, + "blocks.15.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 3, + "offset": 55783424, + "bytes": 6291456 + }, + "base_scales": { + "shard": 3, + "offset": 62074880, + "bytes": 196608 + }, + "residual_data": { + "shard": 3, + "offset": 62271488, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 3, + "offset": 65417216, + "bytes": 196608 + } + }, + "blocks.15.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 3, + "offset": 65613824, + "bytes": 4096 + } + }, + "blocks.15.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 4, + "offset": 0, + "bytes": 2097152 + }, + "base_scales": { + "shard": 4, + "offset": 2097152, + "bytes": 65536 + }, + "residual_data": { + "shard": 4, + "offset": 2162688, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 4, + "offset": 3211264, + "bytes": 65536 + } + }, + "blocks.15.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 3276800, + "bytes": 2048 + } + }, + "blocks.15.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 4, + "offset": 3278848, + "bytes": 1048576 + }, + "base_scales": { + "shard": 4, + "offset": 4327424, + "bytes": 32768 + }, + "residual_data": { + "shard": 4, + "offset": 4360192, + "bytes": 524288 + }, + "residual_scales": { + "shard": 4, + "offset": 4884480, + "bytes": 32768 + } + }, + "blocks.15.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 4917248, + "bytes": 2048 + } + }, + "blocks.15.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 4, + "offset": 4919296, + "bytes": 1048576 + }, + "base_scales": { + "shard": 4, + "offset": 5967872, + "bytes": 32768 + }, + "residual_data": { + "shard": 4, + "offset": 6000640, + "bytes": 524288 + }, + "residual_scales": { + "shard": 4, + "offset": 6524928, + "bytes": 32768 + } + }, + "blocks.15.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 6557696, + "bytes": 8192 + } + }, + "blocks.15.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 4, + "offset": 6565888, + "bytes": 4194304 + }, + "base_scales": { + "shard": 4, + "offset": 10760192, + "bytes": 131072 + }, + "residual_data": { + "shard": 4, + "offset": 10891264, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 4, + "offset": 12988416, + "bytes": 131072 + } + }, + "blocks.15.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 13119488, + "bytes": 2048 + } + }, + "blocks.15.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 4, + "offset": 13121536, + "bytes": 4194304 + }, + "base_scales": { + "shard": 4, + "offset": 17315840, + "bytes": 131072 + }, + "residual_data": { + "shard": 4, + "offset": 17446912, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 4, + "offset": 19544064, + "bytes": 131072 + } + }, + "blocks.15.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 19675136, + "bytes": 2048 + } + }, + "blocks.15.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 19677184, + "bytes": 2048 + } + }, + "blocks.15.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 19679232, + "bytes": 2048 + } + }, + "blocks.15.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 19681280, + "bytes": 2048 + } + }, + "blocks.15.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 19683328, + "bytes": 2048 + } + }, + "blocks.15.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 4, + "offset": 19685376, + "bytes": 1048576 + }, + "base_scales": { + "shard": 4, + "offset": 20733952, + "bytes": 32768 + }, + "residual_data": { + "shard": 4, + "offset": 20766720, + "bytes": 524288 + }, + "residual_scales": { + "shard": 4, + "offset": 21291008, + "bytes": 32768 + } + }, + "blocks.15.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 21323776, + "bytes": 6144 + } + }, + "blocks.15.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 4, + "offset": 21329920, + "bytes": 3145728 + }, + "base_scales": { + "shard": 4, + "offset": 24475648, + "bytes": 98304 + }, + "residual_data": { + "shard": 4, + "offset": 24573952, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 4, + "offset": 26146816, + "bytes": 98304 + } + }, + "blocks.16.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 26245120, + "bytes": 12288 + } + }, + "blocks.16.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 4, + "offset": 26257408, + "bytes": 6291456 + }, + "base_scales": { + "shard": 4, + "offset": 32548864, + "bytes": 196608 + }, + "residual_data": { + "shard": 4, + "offset": 32745472, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 4, + "offset": 35891200, + "bytes": 196608 + } + }, + "blocks.16.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 36087808, + "bytes": 4096 + } + }, + "blocks.16.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 4, + "offset": 36091904, + "bytes": 2097152 + }, + "base_scales": { + "shard": 4, + "offset": 38189056, + "bytes": 65536 + }, + "residual_data": { + "shard": 4, + "offset": 38254592, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 4, + "offset": 39303168, + "bytes": 65536 + } + }, + "blocks.16.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 39368704, + "bytes": 2048 + } + }, + "blocks.16.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 4, + "offset": 39370752, + "bytes": 1048576 + }, + "base_scales": { + "shard": 4, + "offset": 40419328, + "bytes": 32768 + }, + "residual_data": { + "shard": 4, + "offset": 40452096, + "bytes": 524288 + }, + "residual_scales": { + "shard": 4, + "offset": 40976384, + "bytes": 32768 + } + }, + "blocks.16.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 41009152, + "bytes": 2048 + } + }, + "blocks.16.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 4, + "offset": 41011200, + "bytes": 1048576 + }, + "base_scales": { + "shard": 4, + "offset": 42059776, + "bytes": 32768 + }, + "residual_data": { + "shard": 4, + "offset": 42092544, + "bytes": 524288 + }, + "residual_scales": { + "shard": 4, + "offset": 42616832, + "bytes": 32768 + } + }, + "blocks.16.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 42649600, + "bytes": 8192 + } + }, + "blocks.16.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 4, + "offset": 42657792, + "bytes": 4194304 + }, + "base_scales": { + "shard": 4, + "offset": 46852096, + "bytes": 131072 + }, + "residual_data": { + "shard": 4, + "offset": 46983168, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 4, + "offset": 49080320, + "bytes": 131072 + } + }, + "blocks.16.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 49211392, + "bytes": 2048 + } + }, + "blocks.16.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 4, + "offset": 49213440, + "bytes": 4194304 + }, + "base_scales": { + "shard": 4, + "offset": 53407744, + "bytes": 131072 + }, + "residual_data": { + "shard": 4, + "offset": 53538816, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 4, + "offset": 55635968, + "bytes": 131072 + } + }, + "blocks.16.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 55767040, + "bytes": 2048 + } + }, + "blocks.16.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 55769088, + "bytes": 2048 + } + }, + "blocks.16.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 55771136, + "bytes": 2048 + } + }, + "blocks.16.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 55773184, + "bytes": 2048 + } + }, + "blocks.16.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 55775232, + "bytes": 2048 + } + }, + "blocks.16.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 4, + "offset": 55777280, + "bytes": 1048576 + }, + "base_scales": { + "shard": 4, + "offset": 56825856, + "bytes": 32768 + }, + "residual_data": { + "shard": 4, + "offset": 56858624, + "bytes": 524288 + }, + "residual_scales": { + "shard": 4, + "offset": 57382912, + "bytes": 32768 + } + }, + "blocks.16.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 57415680, + "bytes": 6144 + } + }, + "blocks.16.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 4, + "offset": 57421824, + "bytes": 3145728 + }, + "base_scales": { + "shard": 4, + "offset": 60567552, + "bytes": 98304 + }, + "residual_data": { + "shard": 4, + "offset": 60665856, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 4, + "offset": 62238720, + "bytes": 98304 + } + }, + "blocks.17.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 4, + "offset": 62337024, + "bytes": 12288 + } + }, + "blocks.17.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 5, + "offset": 0, + "bytes": 6291456 + }, + "base_scales": { + "shard": 5, + "offset": 6291456, + "bytes": 196608 + }, + "residual_data": { + "shard": 5, + "offset": 6488064, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 5, + "offset": 9633792, + "bytes": 196608 + } + }, + "blocks.17.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 9830400, + "bytes": 4096 + } + }, + "blocks.17.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 5, + "offset": 9834496, + "bytes": 2097152 + }, + "base_scales": { + "shard": 5, + "offset": 11931648, + "bytes": 65536 + }, + "residual_data": { + "shard": 5, + "offset": 11997184, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 5, + "offset": 13045760, + "bytes": 65536 + } + }, + "blocks.17.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 13111296, + "bytes": 2048 + } + }, + "blocks.17.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 5, + "offset": 13113344, + "bytes": 1048576 + }, + "base_scales": { + "shard": 5, + "offset": 14161920, + "bytes": 32768 + }, + "residual_data": { + "shard": 5, + "offset": 14194688, + "bytes": 524288 + }, + "residual_scales": { + "shard": 5, + "offset": 14718976, + "bytes": 32768 + } + }, + "blocks.17.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 14751744, + "bytes": 2048 + } + }, + "blocks.17.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 5, + "offset": 14753792, + "bytes": 1048576 + }, + "base_scales": { + "shard": 5, + "offset": 15802368, + "bytes": 32768 + }, + "residual_data": { + "shard": 5, + "offset": 15835136, + "bytes": 524288 + }, + "residual_scales": { + "shard": 5, + "offset": 16359424, + "bytes": 32768 + } + }, + "blocks.17.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 16392192, + "bytes": 8192 + } + }, + "blocks.17.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 5, + "offset": 16400384, + "bytes": 4194304 + }, + "base_scales": { + "shard": 5, + "offset": 20594688, + "bytes": 131072 + }, + "residual_data": { + "shard": 5, + "offset": 20725760, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 5, + "offset": 22822912, + "bytes": 131072 + } + }, + "blocks.17.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 22953984, + "bytes": 2048 + } + }, + "blocks.17.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 5, + "offset": 22956032, + "bytes": 4194304 + }, + "base_scales": { + "shard": 5, + "offset": 27150336, + "bytes": 131072 + }, + "residual_data": { + "shard": 5, + "offset": 27281408, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 5, + "offset": 29378560, + "bytes": 131072 + } + }, + "blocks.17.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 29509632, + "bytes": 2048 + } + }, + "blocks.17.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 29511680, + "bytes": 2048 + } + }, + "blocks.17.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 29513728, + "bytes": 2048 + } + }, + "blocks.17.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 29515776, + "bytes": 2048 + } + }, + "blocks.17.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 29517824, + "bytes": 2048 + } + }, + "blocks.17.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 5, + "offset": 29519872, + "bytes": 1048576 + }, + "base_scales": { + "shard": 5, + "offset": 30568448, + "bytes": 32768 + }, + "residual_data": { + "shard": 5, + "offset": 30601216, + "bytes": 524288 + }, + "residual_scales": { + "shard": 5, + "offset": 31125504, + "bytes": 32768 + } + }, + "blocks.17.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 31158272, + "bytes": 6144 + } + }, + "blocks.17.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 5, + "offset": 31164416, + "bytes": 3145728 + }, + "base_scales": { + "shard": 5, + "offset": 34310144, + "bytes": 98304 + }, + "residual_data": { + "shard": 5, + "offset": 34408448, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 5, + "offset": 35981312, + "bytes": 98304 + } + }, + "blocks.18.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 36079616, + "bytes": 12288 + } + }, + "blocks.18.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 5, + "offset": 36091904, + "bytes": 6291456 + }, + "base_scales": { + "shard": 5, + "offset": 42383360, + "bytes": 196608 + }, + "residual_data": { + "shard": 5, + "offset": 42579968, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 5, + "offset": 45725696, + "bytes": 196608 + } + }, + "blocks.18.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 45922304, + "bytes": 4096 + } + }, + "blocks.18.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 5, + "offset": 45926400, + "bytes": 2097152 + }, + "base_scales": { + "shard": 5, + "offset": 48023552, + "bytes": 65536 + }, + "residual_data": { + "shard": 5, + "offset": 48089088, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 5, + "offset": 49137664, + "bytes": 65536 + } + }, + "blocks.18.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 49203200, + "bytes": 2048 + } + }, + "blocks.18.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 5, + "offset": 49205248, + "bytes": 1048576 + }, + "base_scales": { + "shard": 5, + "offset": 50253824, + "bytes": 32768 + }, + "residual_data": { + "shard": 5, + "offset": 50286592, + "bytes": 524288 + }, + "residual_scales": { + "shard": 5, + "offset": 50810880, + "bytes": 32768 + } + }, + "blocks.18.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 50843648, + "bytes": 2048 + } + }, + "blocks.18.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 5, + "offset": 50845696, + "bytes": 1048576 + }, + "base_scales": { + "shard": 5, + "offset": 51894272, + "bytes": 32768 + }, + "residual_data": { + "shard": 5, + "offset": 51927040, + "bytes": 524288 + }, + "residual_scales": { + "shard": 5, + "offset": 52451328, + "bytes": 32768 + } + }, + "blocks.18.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 52484096, + "bytes": 8192 + } + }, + "blocks.18.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 5, + "offset": 52492288, + "bytes": 4194304 + }, + "base_scales": { + "shard": 5, + "offset": 56686592, + "bytes": 131072 + }, + "residual_data": { + "shard": 5, + "offset": 56817664, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 5, + "offset": 58914816, + "bytes": 131072 + } + }, + "blocks.18.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 59045888, + "bytes": 2048 + } + }, + "blocks.18.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 5, + "offset": 59047936, + "bytes": 4194304 + }, + "base_scales": { + "shard": 5, + "offset": 63242240, + "bytes": 131072 + }, + "residual_data": { + "shard": 5, + "offset": 63373312, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 5, + "offset": 65470464, + "bytes": 131072 + } + }, + "blocks.18.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 65601536, + "bytes": 2048 + } + }, + "blocks.18.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 65603584, + "bytes": 2048 + } + }, + "blocks.18.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 65605632, + "bytes": 2048 + } + }, + "blocks.18.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 65607680, + "bytes": 2048 + } + }, + "blocks.18.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 5, + "offset": 65609728, + "bytes": 2048 + } + }, + "blocks.18.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 5, + "offset": 65611776, + "bytes": 1048576 + }, + "base_scales": { + "shard": 5, + "offset": 66660352, + "bytes": 32768 + }, + "residual_data": { + "shard": 6, + "offset": 0, + "bytes": 524288 + }, + "residual_scales": { + "shard": 6, + "offset": 524288, + "bytes": 32768 + } + }, + "blocks.18.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 557056, + "bytes": 6144 + } + }, + "blocks.18.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 6, + "offset": 563200, + "bytes": 3145728 + }, + "base_scales": { + "shard": 6, + "offset": 3708928, + "bytes": 98304 + }, + "residual_data": { + "shard": 6, + "offset": 3807232, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 6, + "offset": 5380096, + "bytes": 98304 + } + }, + "blocks.19.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 5478400, + "bytes": 12288 + } + }, + "blocks.19.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 6, + "offset": 5490688, + "bytes": 6291456 + }, + "base_scales": { + "shard": 6, + "offset": 11782144, + "bytes": 196608 + }, + "residual_data": { + "shard": 6, + "offset": 11978752, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 6, + "offset": 15124480, + "bytes": 196608 + } + }, + "blocks.19.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 15321088, + "bytes": 4096 + } + }, + "blocks.19.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 6, + "offset": 15325184, + "bytes": 2097152 + }, + "base_scales": { + "shard": 6, + "offset": 17422336, + "bytes": 65536 + }, + "residual_data": { + "shard": 6, + "offset": 17487872, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 6, + "offset": 18536448, + "bytes": 65536 + } + }, + "blocks.19.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 18601984, + "bytes": 2048 + } + }, + "blocks.19.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 6, + "offset": 18604032, + "bytes": 1048576 + }, + "base_scales": { + "shard": 6, + "offset": 19652608, + "bytes": 32768 + }, + "residual_data": { + "shard": 6, + "offset": 19685376, + "bytes": 524288 + }, + "residual_scales": { + "shard": 6, + "offset": 20209664, + "bytes": 32768 + } + }, + "blocks.19.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 20242432, + "bytes": 2048 + } + }, + "blocks.19.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 6, + "offset": 20244480, + "bytes": 1048576 + }, + "base_scales": { + "shard": 6, + "offset": 21293056, + "bytes": 32768 + }, + "residual_data": { + "shard": 6, + "offset": 21325824, + "bytes": 524288 + }, + "residual_scales": { + "shard": 6, + "offset": 21850112, + "bytes": 32768 + } + }, + "blocks.19.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 21882880, + "bytes": 8192 + } + }, + "blocks.19.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 6, + "offset": 21891072, + "bytes": 4194304 + }, + "base_scales": { + "shard": 6, + "offset": 26085376, + "bytes": 131072 + }, + "residual_data": { + "shard": 6, + "offset": 26216448, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 6, + "offset": 28313600, + "bytes": 131072 + } + }, + "blocks.19.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 28444672, + "bytes": 2048 + } + }, + "blocks.19.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 6, + "offset": 28446720, + "bytes": 4194304 + }, + "base_scales": { + "shard": 6, + "offset": 32641024, + "bytes": 131072 + }, + "residual_data": { + "shard": 6, + "offset": 32772096, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 6, + "offset": 34869248, + "bytes": 131072 + } + }, + "blocks.19.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 35000320, + "bytes": 2048 + } + }, + "blocks.19.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 35002368, + "bytes": 2048 + } + }, + "blocks.19.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 35004416, + "bytes": 2048 + } + }, + "blocks.19.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 35006464, + "bytes": 2048 + } + }, + "blocks.19.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 35008512, + "bytes": 2048 + } + }, + "blocks.19.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 6, + "offset": 35010560, + "bytes": 1048576 + }, + "base_scales": { + "shard": 6, + "offset": 36059136, + "bytes": 32768 + }, + "residual_data": { + "shard": 6, + "offset": 36091904, + "bytes": 524288 + }, + "residual_scales": { + "shard": 6, + "offset": 36616192, + "bytes": 32768 + } + }, + "blocks.19.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 36648960, + "bytes": 6144 + } + }, + "blocks.19.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 6, + "offset": 36655104, + "bytes": 3145728 + }, + "base_scales": { + "shard": 6, + "offset": 39800832, + "bytes": 98304 + }, + "residual_data": { + "shard": 6, + "offset": 39899136, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 6, + "offset": 41472000, + "bytes": 98304 + } + }, + "blocks.2.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 41570304, + "bytes": 12288 + } + }, + "blocks.2.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 6, + "offset": 41582592, + "bytes": 6291456 + }, + "base_scales": { + "shard": 6, + "offset": 47874048, + "bytes": 196608 + }, + "residual_data": { + "shard": 6, + "offset": 48070656, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 6, + "offset": 51216384, + "bytes": 196608 + } + }, + "blocks.2.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 51412992, + "bytes": 4096 + } + }, + "blocks.2.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 6, + "offset": 51417088, + "bytes": 2097152 + }, + "base_scales": { + "shard": 6, + "offset": 53514240, + "bytes": 65536 + }, + "residual_data": { + "shard": 6, + "offset": 53579776, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 6, + "offset": 54628352, + "bytes": 65536 + } + }, + "blocks.2.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 54693888, + "bytes": 2048 + } + }, + "blocks.2.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 6, + "offset": 54695936, + "bytes": 1048576 + }, + "base_scales": { + "shard": 6, + "offset": 55744512, + "bytes": 32768 + }, + "residual_data": { + "shard": 6, + "offset": 55777280, + "bytes": 524288 + }, + "residual_scales": { + "shard": 6, + "offset": 56301568, + "bytes": 32768 + } + }, + "blocks.2.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 56334336, + "bytes": 2048 + } + }, + "blocks.2.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 6, + "offset": 56336384, + "bytes": 1048576 + }, + "base_scales": { + "shard": 6, + "offset": 57384960, + "bytes": 32768 + }, + "residual_data": { + "shard": 6, + "offset": 57417728, + "bytes": 524288 + }, + "residual_scales": { + "shard": 6, + "offset": 57942016, + "bytes": 32768 + } + }, + "blocks.2.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 57974784, + "bytes": 8192 + } + }, + "blocks.2.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 6, + "offset": 57982976, + "bytes": 4194304 + }, + "base_scales": { + "shard": 6, + "offset": 62177280, + "bytes": 131072 + }, + "residual_data": { + "shard": 6, + "offset": 62308352, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 6, + "offset": 64405504, + "bytes": 131072 + } + }, + "blocks.2.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 6, + "offset": 64536576, + "bytes": 2048 + } + }, + "blocks.2.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 7, + "offset": 0, + "bytes": 4194304 + }, + "base_scales": { + "shard": 7, + "offset": 4194304, + "bytes": 131072 + }, + "residual_data": { + "shard": 7, + "offset": 4325376, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 7, + "offset": 6422528, + "bytes": 131072 + } + }, + "blocks.2.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 6553600, + "bytes": 2048 + } + }, + "blocks.2.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 6555648, + "bytes": 2048 + } + }, + "blocks.2.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 6557696, + "bytes": 2048 + } + }, + "blocks.2.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 6559744, + "bytes": 2048 + } + }, + "blocks.2.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 6561792, + "bytes": 2048 + } + }, + "blocks.2.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 7, + "offset": 6563840, + "bytes": 1048576 + }, + "base_scales": { + "shard": 7, + "offset": 7612416, + "bytes": 32768 + }, + "residual_data": { + "shard": 7, + "offset": 7645184, + "bytes": 524288 + }, + "residual_scales": { + "shard": 7, + "offset": 8169472, + "bytes": 32768 + } + }, + "blocks.2.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 8202240, + "bytes": 6144 + } + }, + "blocks.2.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 7, + "offset": 8208384, + "bytes": 3145728 + }, + "base_scales": { + "shard": 7, + "offset": 11354112, + "bytes": 98304 + }, + "residual_data": { + "shard": 7, + "offset": 11452416, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 7, + "offset": 13025280, + "bytes": 98304 + } + }, + "blocks.20.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 13123584, + "bytes": 12288 + } + }, + "blocks.20.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 7, + "offset": 13135872, + "bytes": 6291456 + }, + "base_scales": { + "shard": 7, + "offset": 19427328, + "bytes": 196608 + }, + "residual_data": { + "shard": 7, + "offset": 19623936, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 7, + "offset": 22769664, + "bytes": 196608 + } + }, + "blocks.20.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 22966272, + "bytes": 4096 + } + }, + "blocks.20.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 7, + "offset": 22970368, + "bytes": 2097152 + }, + "base_scales": { + "shard": 7, + "offset": 25067520, + "bytes": 65536 + }, + "residual_data": { + "shard": 7, + "offset": 25133056, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 7, + "offset": 26181632, + "bytes": 65536 + } + }, + "blocks.20.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 26247168, + "bytes": 2048 + } + }, + "blocks.20.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 7, + "offset": 26249216, + "bytes": 1048576 + }, + "base_scales": { + "shard": 7, + "offset": 27297792, + "bytes": 32768 + }, + "residual_data": { + "shard": 7, + "offset": 27330560, + "bytes": 524288 + }, + "residual_scales": { + "shard": 7, + "offset": 27854848, + "bytes": 32768 + } + }, + "blocks.20.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 27887616, + "bytes": 2048 + } + }, + "blocks.20.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 7, + "offset": 27889664, + "bytes": 1048576 + }, + "base_scales": { + "shard": 7, + "offset": 28938240, + "bytes": 32768 + }, + "residual_data": { + "shard": 7, + "offset": 28971008, + "bytes": 524288 + }, + "residual_scales": { + "shard": 7, + "offset": 29495296, + "bytes": 32768 + } + }, + "blocks.20.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 29528064, + "bytes": 8192 + } + }, + "blocks.20.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 7, + "offset": 29536256, + "bytes": 4194304 + }, + "base_scales": { + "shard": 7, + "offset": 33730560, + "bytes": 131072 + }, + "residual_data": { + "shard": 7, + "offset": 33861632, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 7, + "offset": 35958784, + "bytes": 131072 + } + }, + "blocks.20.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 36089856, + "bytes": 2048 + } + }, + "blocks.20.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 7, + "offset": 36091904, + "bytes": 4194304 + }, + "base_scales": { + "shard": 7, + "offset": 40286208, + "bytes": 131072 + }, + "residual_data": { + "shard": 7, + "offset": 40417280, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 7, + "offset": 42514432, + "bytes": 131072 + } + }, + "blocks.20.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 42645504, + "bytes": 2048 + } + }, + "blocks.20.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 42647552, + "bytes": 2048 + } + }, + "blocks.20.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 42649600, + "bytes": 2048 + } + }, + "blocks.20.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 42651648, + "bytes": 2048 + } + }, + "blocks.20.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 42653696, + "bytes": 2048 + } + }, + "blocks.20.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 7, + "offset": 42655744, + "bytes": 1048576 + }, + "base_scales": { + "shard": 7, + "offset": 43704320, + "bytes": 32768 + }, + "residual_data": { + "shard": 7, + "offset": 43737088, + "bytes": 524288 + }, + "residual_scales": { + "shard": 7, + "offset": 44261376, + "bytes": 32768 + } + }, + "blocks.20.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 44294144, + "bytes": 6144 + } + }, + "blocks.20.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 7, + "offset": 44300288, + "bytes": 3145728 + }, + "base_scales": { + "shard": 7, + "offset": 47446016, + "bytes": 98304 + }, + "residual_data": { + "shard": 7, + "offset": 47544320, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 7, + "offset": 49117184, + "bytes": 98304 + } + }, + "blocks.21.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 49215488, + "bytes": 12288 + } + }, + "blocks.21.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 7, + "offset": 49227776, + "bytes": 6291456 + }, + "base_scales": { + "shard": 7, + "offset": 55519232, + "bytes": 196608 + }, + "residual_data": { + "shard": 7, + "offset": 55715840, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 7, + "offset": 58861568, + "bytes": 196608 + } + }, + "blocks.21.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 59058176, + "bytes": 4096 + } + }, + "blocks.21.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 7, + "offset": 59062272, + "bytes": 2097152 + }, + "base_scales": { + "shard": 7, + "offset": 61159424, + "bytes": 65536 + }, + "residual_data": { + "shard": 7, + "offset": 61224960, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 7, + "offset": 62273536, + "bytes": 65536 + } + }, + "blocks.21.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 62339072, + "bytes": 2048 + } + }, + "blocks.21.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 7, + "offset": 62341120, + "bytes": 1048576 + }, + "base_scales": { + "shard": 7, + "offset": 63389696, + "bytes": 32768 + }, + "residual_data": { + "shard": 7, + "offset": 63422464, + "bytes": 524288 + }, + "residual_scales": { + "shard": 7, + "offset": 63946752, + "bytes": 32768 + } + }, + "blocks.21.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 63979520, + "bytes": 2048 + } + }, + "blocks.21.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 7, + "offset": 63981568, + "bytes": 1048576 + }, + "base_scales": { + "shard": 7, + "offset": 65030144, + "bytes": 32768 + }, + "residual_data": { + "shard": 7, + "offset": 65062912, + "bytes": 524288 + }, + "residual_scales": { + "shard": 7, + "offset": 65587200, + "bytes": 32768 + } + }, + "blocks.21.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 7, + "offset": 65619968, + "bytes": 8192 + } + }, + "blocks.21.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 8, + "offset": 0, + "bytes": 4194304 + }, + "base_scales": { + "shard": 8, + "offset": 4194304, + "bytes": 131072 + }, + "residual_data": { + "shard": 8, + "offset": 4325376, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 8, + "offset": 6422528, + "bytes": 131072 + } + }, + "blocks.21.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 6553600, + "bytes": 2048 + } + }, + "blocks.21.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 8, + "offset": 6555648, + "bytes": 4194304 + }, + "base_scales": { + "shard": 8, + "offset": 10749952, + "bytes": 131072 + }, + "residual_data": { + "shard": 8, + "offset": 10881024, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 8, + "offset": 12978176, + "bytes": 131072 + } + }, + "blocks.21.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 13109248, + "bytes": 2048 + } + }, + "blocks.21.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 13111296, + "bytes": 2048 + } + }, + "blocks.21.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 13113344, + "bytes": 2048 + } + }, + "blocks.21.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 13115392, + "bytes": 2048 + } + }, + "blocks.21.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 13117440, + "bytes": 2048 + } + }, + "blocks.21.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 8, + "offset": 13119488, + "bytes": 1048576 + }, + "base_scales": { + "shard": 8, + "offset": 14168064, + "bytes": 32768 + }, + "residual_data": { + "shard": 8, + "offset": 14200832, + "bytes": 524288 + }, + "residual_scales": { + "shard": 8, + "offset": 14725120, + "bytes": 32768 + } + }, + "blocks.21.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 14757888, + "bytes": 6144 + } + }, + "blocks.21.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 8, + "offset": 14764032, + "bytes": 3145728 + }, + "base_scales": { + "shard": 8, + "offset": 17909760, + "bytes": 98304 + }, + "residual_data": { + "shard": 8, + "offset": 18008064, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 8, + "offset": 19580928, + "bytes": 98304 + } + }, + "blocks.22.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 19679232, + "bytes": 12288 + } + }, + "blocks.22.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 8, + "offset": 19691520, + "bytes": 6291456 + }, + "base_scales": { + "shard": 8, + "offset": 25982976, + "bytes": 196608 + }, + "residual_data": { + "shard": 8, + "offset": 26179584, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 8, + "offset": 29325312, + "bytes": 196608 + } + }, + "blocks.22.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 29521920, + "bytes": 4096 + } + }, + "blocks.22.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 8, + "offset": 29526016, + "bytes": 2097152 + }, + "base_scales": { + "shard": 8, + "offset": 31623168, + "bytes": 65536 + }, + "residual_data": { + "shard": 8, + "offset": 31688704, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 8, + "offset": 32737280, + "bytes": 65536 + } + }, + "blocks.22.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 32802816, + "bytes": 2048 + } + }, + "blocks.22.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 8, + "offset": 32804864, + "bytes": 1048576 + }, + "base_scales": { + "shard": 8, + "offset": 33853440, + "bytes": 32768 + }, + "residual_data": { + "shard": 8, + "offset": 33886208, + "bytes": 524288 + }, + "residual_scales": { + "shard": 8, + "offset": 34410496, + "bytes": 32768 + } + }, + "blocks.22.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 34443264, + "bytes": 2048 + } + }, + "blocks.22.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 8, + "offset": 34445312, + "bytes": 1048576 + }, + "base_scales": { + "shard": 8, + "offset": 35493888, + "bytes": 32768 + }, + "residual_data": { + "shard": 8, + "offset": 35526656, + "bytes": 524288 + }, + "residual_scales": { + "shard": 8, + "offset": 36050944, + "bytes": 32768 + } + }, + "blocks.22.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 36083712, + "bytes": 8192 + } + }, + "blocks.22.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 8, + "offset": 36091904, + "bytes": 4194304 + }, + "base_scales": { + "shard": 8, + "offset": 40286208, + "bytes": 131072 + }, + "residual_data": { + "shard": 8, + "offset": 40417280, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 8, + "offset": 42514432, + "bytes": 131072 + } + }, + "blocks.22.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 42645504, + "bytes": 2048 + } + }, + "blocks.22.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 8, + "offset": 42647552, + "bytes": 4194304 + }, + "base_scales": { + "shard": 8, + "offset": 46841856, + "bytes": 131072 + }, + "residual_data": { + "shard": 8, + "offset": 46972928, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 8, + "offset": 49070080, + "bytes": 131072 + } + }, + "blocks.22.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 49201152, + "bytes": 2048 + } + }, + "blocks.22.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 49203200, + "bytes": 2048 + } + }, + "blocks.22.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 49205248, + "bytes": 2048 + } + }, + "blocks.22.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 49207296, + "bytes": 2048 + } + }, + "blocks.22.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 49209344, + "bytes": 2048 + } + }, + "blocks.22.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 8, + "offset": 49211392, + "bytes": 1048576 + }, + "base_scales": { + "shard": 8, + "offset": 50259968, + "bytes": 32768 + }, + "residual_data": { + "shard": 8, + "offset": 50292736, + "bytes": 524288 + }, + "residual_scales": { + "shard": 8, + "offset": 50817024, + "bytes": 32768 + } + }, + "blocks.22.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 50849792, + "bytes": 6144 + } + }, + "blocks.22.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 8, + "offset": 50855936, + "bytes": 3145728 + }, + "base_scales": { + "shard": 8, + "offset": 54001664, + "bytes": 98304 + }, + "residual_data": { + "shard": 8, + "offset": 54099968, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 8, + "offset": 55672832, + "bytes": 98304 + } + }, + "blocks.23.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 55771136, + "bytes": 12288 + } + }, + "blocks.23.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 8, + "offset": 55783424, + "bytes": 6291456 + }, + "base_scales": { + "shard": 8, + "offset": 62074880, + "bytes": 196608 + }, + "residual_data": { + "shard": 8, + "offset": 62271488, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 8, + "offset": 65417216, + "bytes": 196608 + } + }, + "blocks.23.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 8, + "offset": 65613824, + "bytes": 4096 + } + }, + "blocks.23.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 9, + "offset": 0, + "bytes": 2097152 + }, + "base_scales": { + "shard": 9, + "offset": 2097152, + "bytes": 65536 + }, + "residual_data": { + "shard": 9, + "offset": 2162688, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 9, + "offset": 3211264, + "bytes": 65536 + } + }, + "blocks.23.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 3276800, + "bytes": 2048 + } + }, + "blocks.23.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 9, + "offset": 3278848, + "bytes": 1048576 + }, + "base_scales": { + "shard": 9, + "offset": 4327424, + "bytes": 32768 + }, + "residual_data": { + "shard": 9, + "offset": 4360192, + "bytes": 524288 + }, + "residual_scales": { + "shard": 9, + "offset": 4884480, + "bytes": 32768 + } + }, + "blocks.23.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 4917248, + "bytes": 2048 + } + }, + "blocks.23.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 9, + "offset": 4919296, + "bytes": 1048576 + }, + "base_scales": { + "shard": 9, + "offset": 5967872, + "bytes": 32768 + }, + "residual_data": { + "shard": 9, + "offset": 6000640, + "bytes": 524288 + }, + "residual_scales": { + "shard": 9, + "offset": 6524928, + "bytes": 32768 + } + }, + "blocks.23.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 6557696, + "bytes": 8192 + } + }, + "blocks.23.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 9, + "offset": 6565888, + "bytes": 4194304 + }, + "base_scales": { + "shard": 9, + "offset": 10760192, + "bytes": 131072 + }, + "residual_data": { + "shard": 9, + "offset": 10891264, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 9, + "offset": 12988416, + "bytes": 131072 + } + }, + "blocks.23.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 13119488, + "bytes": 2048 + } + }, + "blocks.23.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 9, + "offset": 13121536, + "bytes": 4194304 + }, + "base_scales": { + "shard": 9, + "offset": 17315840, + "bytes": 131072 + }, + "residual_data": { + "shard": 9, + "offset": 17446912, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 9, + "offset": 19544064, + "bytes": 131072 + } + }, + "blocks.23.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 19675136, + "bytes": 2048 + } + }, + "blocks.23.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 19677184, + "bytes": 2048 + } + }, + "blocks.23.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 19679232, + "bytes": 2048 + } + }, + "blocks.23.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 19681280, + "bytes": 2048 + } + }, + "blocks.23.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 19683328, + "bytes": 2048 + } + }, + "blocks.23.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 9, + "offset": 19685376, + "bytes": 1048576 + }, + "base_scales": { + "shard": 9, + "offset": 20733952, + "bytes": 32768 + }, + "residual_data": { + "shard": 9, + "offset": 20766720, + "bytes": 524288 + }, + "residual_scales": { + "shard": 9, + "offset": 21291008, + "bytes": 32768 + } + }, + "blocks.23.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 21323776, + "bytes": 6144 + } + }, + "blocks.23.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 9, + "offset": 21329920, + "bytes": 3145728 + }, + "base_scales": { + "shard": 9, + "offset": 24475648, + "bytes": 98304 + }, + "residual_data": { + "shard": 9, + "offset": 24573952, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 9, + "offset": 26146816, + "bytes": 98304 + } + }, + "blocks.3.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 26245120, + "bytes": 12288 + } + }, + "blocks.3.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 9, + "offset": 26257408, + "bytes": 6291456 + }, + "base_scales": { + "shard": 9, + "offset": 32548864, + "bytes": 196608 + }, + "residual_data": { + "shard": 9, + "offset": 32745472, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 9, + "offset": 35891200, + "bytes": 196608 + } + }, + "blocks.3.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 36087808, + "bytes": 4096 + } + }, + "blocks.3.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 9, + "offset": 36091904, + "bytes": 2097152 + }, + "base_scales": { + "shard": 9, + "offset": 38189056, + "bytes": 65536 + }, + "residual_data": { + "shard": 9, + "offset": 38254592, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 9, + "offset": 39303168, + "bytes": 65536 + } + }, + "blocks.3.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 39368704, + "bytes": 2048 + } + }, + "blocks.3.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 9, + "offset": 39370752, + "bytes": 1048576 + }, + "base_scales": { + "shard": 9, + "offset": 40419328, + "bytes": 32768 + }, + "residual_data": { + "shard": 9, + "offset": 40452096, + "bytes": 524288 + }, + "residual_scales": { + "shard": 9, + "offset": 40976384, + "bytes": 32768 + } + }, + "blocks.3.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 41009152, + "bytes": 2048 + } + }, + "blocks.3.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 9, + "offset": 41011200, + "bytes": 1048576 + }, + "base_scales": { + "shard": 9, + "offset": 42059776, + "bytes": 32768 + }, + "residual_data": { + "shard": 9, + "offset": 42092544, + "bytes": 524288 + }, + "residual_scales": { + "shard": 9, + "offset": 42616832, + "bytes": 32768 + } + }, + "blocks.3.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 42649600, + "bytes": 8192 + } + }, + "blocks.3.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 9, + "offset": 42657792, + "bytes": 4194304 + }, + "base_scales": { + "shard": 9, + "offset": 46852096, + "bytes": 131072 + }, + "residual_data": { + "shard": 9, + "offset": 46983168, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 9, + "offset": 49080320, + "bytes": 131072 + } + }, + "blocks.3.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 49211392, + "bytes": 2048 + } + }, + "blocks.3.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 9, + "offset": 49213440, + "bytes": 4194304 + }, + "base_scales": { + "shard": 9, + "offset": 53407744, + "bytes": 131072 + }, + "residual_data": { + "shard": 9, + "offset": 53538816, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 9, + "offset": 55635968, + "bytes": 131072 + } + }, + "blocks.3.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 55767040, + "bytes": 2048 + } + }, + "blocks.3.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 55769088, + "bytes": 2048 + } + }, + "blocks.3.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 55771136, + "bytes": 2048 + } + }, + "blocks.3.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 55773184, + "bytes": 2048 + } + }, + "blocks.3.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 55775232, + "bytes": 2048 + } + }, + "blocks.3.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 9, + "offset": 55777280, + "bytes": 1048576 + }, + "base_scales": { + "shard": 9, + "offset": 56825856, + "bytes": 32768 + }, + "residual_data": { + "shard": 9, + "offset": 56858624, + "bytes": 524288 + }, + "residual_scales": { + "shard": 9, + "offset": 57382912, + "bytes": 32768 + } + }, + "blocks.3.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 57415680, + "bytes": 6144 + } + }, + "blocks.3.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 9, + "offset": 57421824, + "bytes": 3145728 + }, + "base_scales": { + "shard": 9, + "offset": 60567552, + "bytes": 98304 + }, + "residual_data": { + "shard": 9, + "offset": 60665856, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 9, + "offset": 62238720, + "bytes": 98304 + } + }, + "blocks.4.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 9, + "offset": 62337024, + "bytes": 12288 + } + }, + "blocks.4.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 10, + "offset": 0, + "bytes": 6291456 + }, + "base_scales": { + "shard": 10, + "offset": 6291456, + "bytes": 196608 + }, + "residual_data": { + "shard": 10, + "offset": 6488064, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 10, + "offset": 9633792, + "bytes": 196608 + } + }, + "blocks.4.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 9830400, + "bytes": 4096 + } + }, + "blocks.4.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 10, + "offset": 9834496, + "bytes": 2097152 + }, + "base_scales": { + "shard": 10, + "offset": 11931648, + "bytes": 65536 + }, + "residual_data": { + "shard": 10, + "offset": 11997184, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 10, + "offset": 13045760, + "bytes": 65536 + } + }, + "blocks.4.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 13111296, + "bytes": 2048 + } + }, + "blocks.4.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 10, + "offset": 13113344, + "bytes": 1048576 + }, + "base_scales": { + "shard": 10, + "offset": 14161920, + "bytes": 32768 + }, + "residual_data": { + "shard": 10, + "offset": 14194688, + "bytes": 524288 + }, + "residual_scales": { + "shard": 10, + "offset": 14718976, + "bytes": 32768 + } + }, + "blocks.4.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 14751744, + "bytes": 2048 + } + }, + "blocks.4.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 10, + "offset": 14753792, + "bytes": 1048576 + }, + "base_scales": { + "shard": 10, + "offset": 15802368, + "bytes": 32768 + }, + "residual_data": { + "shard": 10, + "offset": 15835136, + "bytes": 524288 + }, + "residual_scales": { + "shard": 10, + "offset": 16359424, + "bytes": 32768 + } + }, + "blocks.4.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 16392192, + "bytes": 8192 + } + }, + "blocks.4.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 10, + "offset": 16400384, + "bytes": 4194304 + }, + "base_scales": { + "shard": 10, + "offset": 20594688, + "bytes": 131072 + }, + "residual_data": { + "shard": 10, + "offset": 20725760, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 10, + "offset": 22822912, + "bytes": 131072 + } + }, + "blocks.4.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 22953984, + "bytes": 2048 + } + }, + "blocks.4.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 10, + "offset": 22956032, + "bytes": 4194304 + }, + "base_scales": { + "shard": 10, + "offset": 27150336, + "bytes": 131072 + }, + "residual_data": { + "shard": 10, + "offset": 27281408, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 10, + "offset": 29378560, + "bytes": 131072 + } + }, + "blocks.4.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 29509632, + "bytes": 2048 + } + }, + "blocks.4.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 29511680, + "bytes": 2048 + } + }, + "blocks.4.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 29513728, + "bytes": 2048 + } + }, + "blocks.4.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 29515776, + "bytes": 2048 + } + }, + "blocks.4.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 29517824, + "bytes": 2048 + } + }, + "blocks.4.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 10, + "offset": 29519872, + "bytes": 1048576 + }, + "base_scales": { + "shard": 10, + "offset": 30568448, + "bytes": 32768 + }, + "residual_data": { + "shard": 10, + "offset": 30601216, + "bytes": 524288 + }, + "residual_scales": { + "shard": 10, + "offset": 31125504, + "bytes": 32768 + } + }, + "blocks.4.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 31158272, + "bytes": 6144 + } + }, + "blocks.4.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 10, + "offset": 31164416, + "bytes": 3145728 + }, + "base_scales": { + "shard": 10, + "offset": 34310144, + "bytes": 98304 + }, + "residual_data": { + "shard": 10, + "offset": 34408448, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 10, + "offset": 35981312, + "bytes": 98304 + } + }, + "blocks.5.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 36079616, + "bytes": 12288 + } + }, + "blocks.5.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 10, + "offset": 36091904, + "bytes": 6291456 + }, + "base_scales": { + "shard": 10, + "offset": 42383360, + "bytes": 196608 + }, + "residual_data": { + "shard": 10, + "offset": 42579968, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 10, + "offset": 45725696, + "bytes": 196608 + } + }, + "blocks.5.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 45922304, + "bytes": 4096 + } + }, + "blocks.5.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 10, + "offset": 45926400, + "bytes": 2097152 + }, + "base_scales": { + "shard": 10, + "offset": 48023552, + "bytes": 65536 + }, + "residual_data": { + "shard": 10, + "offset": 48089088, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 10, + "offset": 49137664, + "bytes": 65536 + } + }, + "blocks.5.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 49203200, + "bytes": 2048 + } + }, + "blocks.5.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 10, + "offset": 49205248, + "bytes": 1048576 + }, + "base_scales": { + "shard": 10, + "offset": 50253824, + "bytes": 32768 + }, + "residual_data": { + "shard": 10, + "offset": 50286592, + "bytes": 524288 + }, + "residual_scales": { + "shard": 10, + "offset": 50810880, + "bytes": 32768 + } + }, + "blocks.5.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 50843648, + "bytes": 2048 + } + }, + "blocks.5.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 10, + "offset": 50845696, + "bytes": 1048576 + }, + "base_scales": { + "shard": 10, + "offset": 51894272, + "bytes": 32768 + }, + "residual_data": { + "shard": 10, + "offset": 51927040, + "bytes": 524288 + }, + "residual_scales": { + "shard": 10, + "offset": 52451328, + "bytes": 32768 + } + }, + "blocks.5.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 52484096, + "bytes": 8192 + } + }, + "blocks.5.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 10, + "offset": 52492288, + "bytes": 4194304 + }, + "base_scales": { + "shard": 10, + "offset": 56686592, + "bytes": 131072 + }, + "residual_data": { + "shard": 10, + "offset": 56817664, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 10, + "offset": 58914816, + "bytes": 131072 + } + }, + "blocks.5.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 59045888, + "bytes": 2048 + } + }, + "blocks.5.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 10, + "offset": 59047936, + "bytes": 4194304 + }, + "base_scales": { + "shard": 10, + "offset": 63242240, + "bytes": 131072 + }, + "residual_data": { + "shard": 10, + "offset": 63373312, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 10, + "offset": 65470464, + "bytes": 131072 + } + }, + "blocks.5.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 65601536, + "bytes": 2048 + } + }, + "blocks.5.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 65603584, + "bytes": 2048 + } + }, + "blocks.5.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 65605632, + "bytes": 2048 + } + }, + "blocks.5.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 65607680, + "bytes": 2048 + } + }, + "blocks.5.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 10, + "offset": 65609728, + "bytes": 2048 + } + }, + "blocks.5.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 10, + "offset": 65611776, + "bytes": 1048576 + }, + "base_scales": { + "shard": 10, + "offset": 66660352, + "bytes": 32768 + }, + "residual_data": { + "shard": 11, + "offset": 0, + "bytes": 524288 + }, + "residual_scales": { + "shard": 11, + "offset": 524288, + "bytes": 32768 + } + }, + "blocks.5.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 557056, + "bytes": 6144 + } + }, + "blocks.5.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 11, + "offset": 563200, + "bytes": 3145728 + }, + "base_scales": { + "shard": 11, + "offset": 3708928, + "bytes": 98304 + }, + "residual_data": { + "shard": 11, + "offset": 3807232, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 11, + "offset": 5380096, + "bytes": 98304 + } + }, + "blocks.6.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 5478400, + "bytes": 12288 + } + }, + "blocks.6.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 11, + "offset": 5490688, + "bytes": 6291456 + }, + "base_scales": { + "shard": 11, + "offset": 11782144, + "bytes": 196608 + }, + "residual_data": { + "shard": 11, + "offset": 11978752, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 11, + "offset": 15124480, + "bytes": 196608 + } + }, + "blocks.6.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 15321088, + "bytes": 4096 + } + }, + "blocks.6.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 11, + "offset": 15325184, + "bytes": 2097152 + }, + "base_scales": { + "shard": 11, + "offset": 17422336, + "bytes": 65536 + }, + "residual_data": { + "shard": 11, + "offset": 17487872, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 11, + "offset": 18536448, + "bytes": 65536 + } + }, + "blocks.6.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 18601984, + "bytes": 2048 + } + }, + "blocks.6.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 11, + "offset": 18604032, + "bytes": 1048576 + }, + "base_scales": { + "shard": 11, + "offset": 19652608, + "bytes": 32768 + }, + "residual_data": { + "shard": 11, + "offset": 19685376, + "bytes": 524288 + }, + "residual_scales": { + "shard": 11, + "offset": 20209664, + "bytes": 32768 + } + }, + "blocks.6.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 20242432, + "bytes": 2048 + } + }, + "blocks.6.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 11, + "offset": 20244480, + "bytes": 1048576 + }, + "base_scales": { + "shard": 11, + "offset": 21293056, + "bytes": 32768 + }, + "residual_data": { + "shard": 11, + "offset": 21325824, + "bytes": 524288 + }, + "residual_scales": { + "shard": 11, + "offset": 21850112, + "bytes": 32768 + } + }, + "blocks.6.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 21882880, + "bytes": 8192 + } + }, + "blocks.6.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 11, + "offset": 21891072, + "bytes": 4194304 + }, + "base_scales": { + "shard": 11, + "offset": 26085376, + "bytes": 131072 + }, + "residual_data": { + "shard": 11, + "offset": 26216448, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 11, + "offset": 28313600, + "bytes": 131072 + } + }, + "blocks.6.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 28444672, + "bytes": 2048 + } + }, + "blocks.6.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 11, + "offset": 28446720, + "bytes": 4194304 + }, + "base_scales": { + "shard": 11, + "offset": 32641024, + "bytes": 131072 + }, + "residual_data": { + "shard": 11, + "offset": 32772096, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 11, + "offset": 34869248, + "bytes": 131072 + } + }, + "blocks.6.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 35000320, + "bytes": 2048 + } + }, + "blocks.6.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 35002368, + "bytes": 2048 + } + }, + "blocks.6.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 35004416, + "bytes": 2048 + } + }, + "blocks.6.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 35006464, + "bytes": 2048 + } + }, + "blocks.6.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 35008512, + "bytes": 2048 + } + }, + "blocks.6.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 11, + "offset": 35010560, + "bytes": 1048576 + }, + "base_scales": { + "shard": 11, + "offset": 36059136, + "bytes": 32768 + }, + "residual_data": { + "shard": 11, + "offset": 36091904, + "bytes": 524288 + }, + "residual_scales": { + "shard": 11, + "offset": 36616192, + "bytes": 32768 + } + }, + "blocks.6.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 36648960, + "bytes": 6144 + } + }, + "blocks.6.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 11, + "offset": 36655104, + "bytes": 3145728 + }, + "base_scales": { + "shard": 11, + "offset": 39800832, + "bytes": 98304 + }, + "residual_data": { + "shard": 11, + "offset": 39899136, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 11, + "offset": 41472000, + "bytes": 98304 + } + }, + "blocks.7.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 41570304, + "bytes": 12288 + } + }, + "blocks.7.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 11, + "offset": 41582592, + "bytes": 6291456 + }, + "base_scales": { + "shard": 11, + "offset": 47874048, + "bytes": 196608 + }, + "residual_data": { + "shard": 11, + "offset": 48070656, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 11, + "offset": 51216384, + "bytes": 196608 + } + }, + "blocks.7.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 51412992, + "bytes": 4096 + } + }, + "blocks.7.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 11, + "offset": 51417088, + "bytes": 2097152 + }, + "base_scales": { + "shard": 11, + "offset": 53514240, + "bytes": 65536 + }, + "residual_data": { + "shard": 11, + "offset": 53579776, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 11, + "offset": 54628352, + "bytes": 65536 + } + }, + "blocks.7.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 54693888, + "bytes": 2048 + } + }, + "blocks.7.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 11, + "offset": 54695936, + "bytes": 1048576 + }, + "base_scales": { + "shard": 11, + "offset": 55744512, + "bytes": 32768 + }, + "residual_data": { + "shard": 11, + "offset": 55777280, + "bytes": 524288 + }, + "residual_scales": { + "shard": 11, + "offset": 56301568, + "bytes": 32768 + } + }, + "blocks.7.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 56334336, + "bytes": 2048 + } + }, + "blocks.7.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 11, + "offset": 56336384, + "bytes": 1048576 + }, + "base_scales": { + "shard": 11, + "offset": 57384960, + "bytes": 32768 + }, + "residual_data": { + "shard": 11, + "offset": 57417728, + "bytes": 524288 + }, + "residual_scales": { + "shard": 11, + "offset": 57942016, + "bytes": 32768 + } + }, + "blocks.7.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 57974784, + "bytes": 8192 + } + }, + "blocks.7.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 11, + "offset": 57982976, + "bytes": 4194304 + }, + "base_scales": { + "shard": 11, + "offset": 62177280, + "bytes": 131072 + }, + "residual_data": { + "shard": 11, + "offset": 62308352, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 11, + "offset": 64405504, + "bytes": 131072 + } + }, + "blocks.7.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 11, + "offset": 64536576, + "bytes": 2048 + } + }, + "blocks.7.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 12, + "offset": 0, + "bytes": 4194304 + }, + "base_scales": { + "shard": 12, + "offset": 4194304, + "bytes": 131072 + }, + "residual_data": { + "shard": 12, + "offset": 4325376, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 12, + "offset": 6422528, + "bytes": 131072 + } + }, + "blocks.7.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 6553600, + "bytes": 2048 + } + }, + "blocks.7.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 6555648, + "bytes": 2048 + } + }, + "blocks.7.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 6557696, + "bytes": 2048 + } + }, + "blocks.7.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 6559744, + "bytes": 2048 + } + }, + "blocks.7.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 6561792, + "bytes": 2048 + } + }, + "blocks.7.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 12, + "offset": 6563840, + "bytes": 1048576 + }, + "base_scales": { + "shard": 12, + "offset": 7612416, + "bytes": 32768 + }, + "residual_data": { + "shard": 12, + "offset": 7645184, + "bytes": 524288 + }, + "residual_scales": { + "shard": 12, + "offset": 8169472, + "bytes": 32768 + } + }, + "blocks.7.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 8202240, + "bytes": 6144 + } + }, + "blocks.7.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 12, + "offset": 8208384, + "bytes": 3145728 + }, + "base_scales": { + "shard": 12, + "offset": 11354112, + "bytes": 98304 + }, + "residual_data": { + "shard": 12, + "offset": 11452416, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 12, + "offset": 13025280, + "bytes": 98304 + } + }, + "blocks.8.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 13123584, + "bytes": 12288 + } + }, + "blocks.8.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 12, + "offset": 13135872, + "bytes": 6291456 + }, + "base_scales": { + "shard": 12, + "offset": 19427328, + "bytes": 196608 + }, + "residual_data": { + "shard": 12, + "offset": 19623936, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 12, + "offset": 22769664, + "bytes": 196608 + } + }, + "blocks.8.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 22966272, + "bytes": 4096 + } + }, + "blocks.8.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 12, + "offset": 22970368, + "bytes": 2097152 + }, + "base_scales": { + "shard": 12, + "offset": 25067520, + "bytes": 65536 + }, + "residual_data": { + "shard": 12, + "offset": 25133056, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 12, + "offset": 26181632, + "bytes": 65536 + } + }, + "blocks.8.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 26247168, + "bytes": 2048 + } + }, + "blocks.8.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 12, + "offset": 26249216, + "bytes": 1048576 + }, + "base_scales": { + "shard": 12, + "offset": 27297792, + "bytes": 32768 + }, + "residual_data": { + "shard": 12, + "offset": 27330560, + "bytes": 524288 + }, + "residual_scales": { + "shard": 12, + "offset": 27854848, + "bytes": 32768 + } + }, + "blocks.8.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 27887616, + "bytes": 2048 + } + }, + "blocks.8.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 12, + "offset": 27889664, + "bytes": 1048576 + }, + "base_scales": { + "shard": 12, + "offset": 28938240, + "bytes": 32768 + }, + "residual_data": { + "shard": 12, + "offset": 28971008, + "bytes": 524288 + }, + "residual_scales": { + "shard": 12, + "offset": 29495296, + "bytes": 32768 + } + }, + "blocks.8.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 29528064, + "bytes": 8192 + } + }, + "blocks.8.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 12, + "offset": 29536256, + "bytes": 4194304 + }, + "base_scales": { + "shard": 12, + "offset": 33730560, + "bytes": 131072 + }, + "residual_data": { + "shard": 12, + "offset": 33861632, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 12, + "offset": 35958784, + "bytes": 131072 + } + }, + "blocks.8.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 36089856, + "bytes": 2048 + } + }, + "blocks.8.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 12, + "offset": 36091904, + "bytes": 4194304 + }, + "base_scales": { + "shard": 12, + "offset": 40286208, + "bytes": 131072 + }, + "residual_data": { + "shard": 12, + "offset": 40417280, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 12, + "offset": 42514432, + "bytes": 131072 + } + }, + "blocks.8.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 42645504, + "bytes": 2048 + } + }, + "blocks.8.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 42647552, + "bytes": 2048 + } + }, + "blocks.8.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 42649600, + "bytes": 2048 + } + }, + "blocks.8.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 42651648, + "bytes": 2048 + } + }, + "blocks.8.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 42653696, + "bytes": 2048 + } + }, + "blocks.8.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 12, + "offset": 42655744, + "bytes": 1048576 + }, + "base_scales": { + "shard": 12, + "offset": 43704320, + "bytes": 32768 + }, + "residual_data": { + "shard": 12, + "offset": 43737088, + "bytes": 524288 + }, + "residual_scales": { + "shard": 12, + "offset": 44261376, + "bytes": 32768 + } + }, + "blocks.8.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 44294144, + "bytes": 6144 + } + }, + "blocks.8.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 12, + "offset": 44300288, + "bytes": 3145728 + }, + "base_scales": { + "shard": 12, + "offset": 47446016, + "bytes": 98304 + }, + "residual_data": { + "shard": 12, + "offset": 47544320, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 12, + "offset": 49117184, + "bytes": 98304 + } + }, + "blocks.9.adaLN_modulation.1.bias": { + "shape": [ + 6144 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 49215488, + "bytes": 12288 + } + }, + "blocks.9.adaLN_modulation.1.weight": { + "shape": [ + 6144, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 6144, + "columns": 1024, + "padded_rows": 6144, + "padded_columns": 1024, + "base_data": { + "shard": 12, + "offset": 49227776, + "bytes": 6291456 + }, + "base_scales": { + "shard": 12, + "offset": 55519232, + "bytes": 196608 + }, + "residual_data": { + "shard": 12, + "offset": 55715840, + "bytes": 3145728 + }, + "residual_scales": { + "shard": 12, + "offset": 58861568, + "bytes": 196608 + } + }, + "blocks.9.cross_attn.to_kv.bias": { + "shape": [ + 2048 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 59058176, + "bytes": 4096 + } + }, + "blocks.9.cross_attn.to_kv.weight": { + "shape": [ + 2048, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 2048, + "columns": 1024, + "padded_rows": 2048, + "padded_columns": 1024, + "base_data": { + "shard": 12, + "offset": 59062272, + "bytes": 2097152 + }, + "base_scales": { + "shard": 12, + "offset": 61159424, + "bytes": 65536 + }, + "residual_data": { + "shard": 12, + "offset": 61224960, + "bytes": 1048576 + }, + "residual_scales": { + "shard": 12, + "offset": 62273536, + "bytes": 65536 + } + }, + "blocks.9.cross_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 62339072, + "bytes": 2048 + } + }, + "blocks.9.cross_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 12, + "offset": 62341120, + "bytes": 1048576 + }, + "base_scales": { + "shard": 12, + "offset": 63389696, + "bytes": 32768 + }, + "residual_data": { + "shard": 12, + "offset": 63422464, + "bytes": 524288 + }, + "residual_scales": { + "shard": 12, + "offset": 63946752, + "bytes": 32768 + } + }, + "blocks.9.cross_attn.to_q.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 63979520, + "bytes": 2048 + } + }, + "blocks.9.cross_attn.to_q.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 12, + "offset": 63981568, + "bytes": 1048576 + }, + "base_scales": { + "shard": 12, + "offset": 65030144, + "bytes": 32768 + }, + "residual_data": { + "shard": 12, + "offset": 65062912, + "bytes": 524288 + }, + "residual_scales": { + "shard": 12, + "offset": 65587200, + "bytes": 32768 + } + }, + "blocks.9.mlp.mlp.0.bias": { + "shape": [ + 4096 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 12, + "offset": 65619968, + "bytes": 8192 + } + }, + "blocks.9.mlp.mlp.0.weight": { + "shape": [ + 4096, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "base_data": { + "shard": 13, + "offset": 0, + "bytes": 4194304 + }, + "base_scales": { + "shard": 13, + "offset": 4194304, + "bytes": 131072 + }, + "residual_data": { + "shard": 13, + "offset": 4325376, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 13, + "offset": 6422528, + "bytes": 131072 + } + }, + "blocks.9.mlp.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 13, + "offset": 6553600, + "bytes": 2048 + } + }, + "blocks.9.mlp.mlp.2.weight": { + "shape": [ + 1024, + 4096 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 4096, + "padded_rows": 1024, + "padded_columns": 4096, + "base_data": { + "shard": 13, + "offset": 6555648, + "bytes": 4194304 + }, + "base_scales": { + "shard": 13, + "offset": 10749952, + "bytes": 131072 + }, + "residual_data": { + "shard": 13, + "offset": 10881024, + "bytes": 2097152 + }, + "residual_scales": { + "shard": 13, + "offset": 12978176, + "bytes": 131072 + } + }, + "blocks.9.norm2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 13, + "offset": 13109248, + "bytes": 2048 + } + }, + "blocks.9.norm2.weight": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 13, + "offset": 13111296, + "bytes": 2048 + } + }, + "blocks.9.self_attn.k_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 13, + "offset": 13113344, + "bytes": 2048 + } + }, + "blocks.9.self_attn.q_rms_norm.gamma": { + "shape": [ + 16, + 64 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 13, + "offset": 13115392, + "bytes": 2048 + } + }, + "blocks.9.self_attn.to_out.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 13, + "offset": 13117440, + "bytes": 2048 + } + }, + "blocks.9.self_attn.to_out.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 13, + "offset": 13119488, + "bytes": 1048576 + }, + "base_scales": { + "shard": 13, + "offset": 14168064, + "bytes": 32768 + }, + "residual_data": { + "shard": 13, + "offset": 14200832, + "bytes": 524288 + }, + "residual_scales": { + "shard": 13, + "offset": 14725120, + "bytes": 32768 + } + }, + "blocks.9.self_attn.to_qkv.bias": { + "shape": [ + 3072 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 13, + "offset": 14757888, + "bytes": 6144 + } + }, + "blocks.9.self_attn.to_qkv.weight": { + "shape": [ + 3072, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 3072, + "columns": 1024, + "padded_rows": 3072, + "padded_columns": 1024, + "base_data": { + "shard": 13, + "offset": 14764032, + "bytes": 3145728 + }, + "base_scales": { + "shard": 13, + "offset": 17909760, + "bytes": 98304 + }, + "residual_data": { + "shard": 13, + "offset": 18008064, + "bytes": 1572864 + }, + "residual_scales": { + "shard": 13, + "offset": 19580928, + "bytes": 98304 + } + }, + "input_layer.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 13, + "offset": 19679232, + "bytes": 2048 + } + }, + "input_layer.weight": { + "shape": [ + 1024, + 8 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 8, + "padded_rows": 1024, + "padded_columns": 64, + "base_data": { + "shard": 13, + "offset": 19681280, + "bytes": 65536 + }, + "base_scales": { + "shard": 13, + "offset": 19746816, + "bytes": 2048 + }, + "residual_data": { + "shard": 13, + "offset": 19748864, + "bytes": 32768 + }, + "residual_scales": { + "shard": 13, + "offset": 19781632, + "bytes": 2048 + } + }, + "out_layer.bias": { + "shape": [ + 8 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 13, + "offset": 19783680, + "bytes": 16 + } + }, + "out_layer.weight": { + "shape": [ + 8, + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 13, + "offset": 19783936, + "bytes": 16384 + } + }, + "pos_emb": { + "shape": [ + 4096, + 1024 + ], + "encoding": "symmetric-quantized", + "bits": 8, + "group_size": 64, + "output_tile": 16, + "rows": 4096, + "columns": 1024, + "padded_rows": 4096, + "padded_columns": 1024, + "data": { + "shard": 13, + "offset": 19800320, + "bytes": 4194304 + }, + "scales": { + "shard": 13, + "offset": 23994624, + "bytes": 131072 + } + }, + "t_embedder.mlp.0.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 13, + "offset": 24125696, + "bytes": 2048 + } + }, + "t_embedder.mlp.0.weight": { + "shape": [ + 1024, + 256 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 256, + "padded_rows": 1024, + "padded_columns": 256, + "base_data": { + "shard": 13, + "offset": 24127744, + "bytes": 262144 + }, + "base_scales": { + "shard": 13, + "offset": 24389888, + "bytes": 8192 + }, + "residual_data": { + "shard": 13, + "offset": 24398080, + "bytes": 131072 + }, + "residual_scales": { + "shard": 13, + "offset": 24529152, + "bytes": 8192 + } + }, + "t_embedder.mlp.2.bias": { + "shape": [ + 1024 + ], + "encoding": "raw", + "dtype": "f16", + "data": { + "shard": 13, + "offset": 24537344, + "bytes": 2048 + } + }, + "t_embedder.mlp.2.weight": { + "shape": [ + 1024, + 1024 + ], + "encoding": "residual-quantized", + "group_size": 64, + "output_tile": 16, + "rows": 1024, + "columns": 1024, + "padded_rows": 1024, + "padded_columns": 1024, + "base_data": { + "shard": 13, + "offset": 24539392, + "bytes": 1048576 + }, + "base_scales": { + "shard": 13, + "offset": 25587968, + "bytes": 32768 + }, + "residual_data": { + "shard": 13, + "offset": 25620736, + "bytes": 524288 + }, + "residual_scales": { + "shard": 13, + "offset": 26145024, + "bytes": 32768 + } + } + } +}