| { |
| "schema": "trellis-webgpu-pack", |
| "version": 1, |
| "stage": "sparseStructureFlow", |
| "upstreamRevision": "f8e8cf00c40d53dea26b718e49169ce83cf24c67", |
| "shards": [ |
| { |
| "file": "stage-000.bin", |
| "bytes": 67094016, |
| "sha256": "9cdadc00a0921d95135a25c76d026f19f80779de5a0700dccba1705630c43975" |
| }, |
| { |
| "file": "stage-001.bin", |
| "bytes": 67084800, |
| "sha256": "996637aebb87422ceef9f17e0f116412ce57749a17db15325dcff9d47e47c97f" |
| }, |
| { |
| "file": "stage-002.bin", |
| "bytes": 30948096, |
| "sha256": "85b9ab636d6dca02c19b016429d30ea23cf1d36ae2398a67f202f7a8d6b211cf" |
| } |
| ], |
| "tensors": { |
| "blocks.0.adaLN_modulation.1.bias": { |
| "shape": [ |
| 4608 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 0, |
| "bytes": 9216 |
| } |
| }, |
| "blocks.0.adaLN_modulation.1.weight": { |
| "shape": [ |
| 4608, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4608, |
| "columns": 768, |
| "padded_rows": 4608, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 9216, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 3548160, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.0.cross_attn.to_kv.bias": { |
| "shape": [ |
| 1536 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 3658752, |
| "bytes": 3072 |
| } |
| }, |
| "blocks.0.cross_attn.to_kv.weight": { |
| "shape": [ |
| 1536, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1536, |
| "columns": 768, |
| "padded_rows": 1536, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 3661824, |
| "bytes": 1179648 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 4841472, |
| "bytes": 36864 |
| } |
| }, |
| "blocks.0.cross_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 4878336, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.0.cross_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 4879872, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 5469696, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.0.cross_attn.to_q.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 5488128, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.0.cross_attn.to_q.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 5489664, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 6079488, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.0.mlp.mlp.0.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 6097920, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.0.mlp.mlp.0.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 768, |
| "padded_rows": 3072, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 6104064, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 8463360, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.0.mlp.mlp.2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 8537088, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.0.mlp.mlp.2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 3072, |
| "padded_rows": 768, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 0, |
| "offset": 8538624, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 10897920, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.0.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 10971648, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.0.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 10973184, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.0.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 10974720, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.0.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 10976256, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.0.self_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 10977792, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.0.self_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 10979328, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 11569152, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.0.self_attn.to_qkv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 11587584, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.0.self_attn.to_qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 768, |
| "padded_rows": 2304, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 11592192, |
| "bytes": 1769472 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 13361664, |
| "bytes": 55296 |
| } |
| }, |
| "blocks.1.adaLN_modulation.1.bias": { |
| "shape": [ |
| 4608 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 13416960, |
| "bytes": 9216 |
| } |
| }, |
| "blocks.1.adaLN_modulation.1.weight": { |
| "shape": [ |
| 4608, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4608, |
| "columns": 768, |
| "padded_rows": 4608, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 13426176, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 16965120, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.1.cross_attn.to_kv.bias": { |
| "shape": [ |
| 1536 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 17075712, |
| "bytes": 3072 |
| } |
| }, |
| "blocks.1.cross_attn.to_kv.weight": { |
| "shape": [ |
| 1536, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1536, |
| "columns": 768, |
| "padded_rows": 1536, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 17078784, |
| "bytes": 1179648 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 18258432, |
| "bytes": 36864 |
| } |
| }, |
| "blocks.1.cross_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 18295296, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.1.cross_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 18296832, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 18886656, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.1.cross_attn.to_q.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 18905088, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.1.cross_attn.to_q.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 18906624, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 19496448, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.1.mlp.mlp.0.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 19514880, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.1.mlp.mlp.0.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 768, |
| "padded_rows": 3072, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 19521024, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 21880320, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.1.mlp.mlp.2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 21954048, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.1.mlp.mlp.2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 3072, |
| "padded_rows": 768, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 0, |
| "offset": 21955584, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 24314880, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.1.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 24388608, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.1.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 24390144, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.1.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 24391680, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.1.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 24393216, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.1.self_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 24394752, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.1.self_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 24396288, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 24986112, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.1.self_attn.to_qkv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 25004544, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.1.self_attn.to_qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 768, |
| "padded_rows": 2304, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 25009152, |
| "bytes": 1769472 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 26778624, |
| "bytes": 55296 |
| } |
| }, |
| "blocks.10.adaLN_modulation.1.bias": { |
| "shape": [ |
| 4608 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 26833920, |
| "bytes": 9216 |
| } |
| }, |
| "blocks.10.adaLN_modulation.1.weight": { |
| "shape": [ |
| 4608, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4608, |
| "columns": 768, |
| "padded_rows": 4608, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 26843136, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 30382080, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.10.cross_attn.to_kv.bias": { |
| "shape": [ |
| 1536 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 30492672, |
| "bytes": 3072 |
| } |
| }, |
| "blocks.10.cross_attn.to_kv.weight": { |
| "shape": [ |
| 1536, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1536, |
| "columns": 768, |
| "padded_rows": 1536, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 30495744, |
| "bytes": 1179648 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 31675392, |
| "bytes": 36864 |
| } |
| }, |
| "blocks.10.cross_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 31712256, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.10.cross_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 31713792, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 32303616, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.10.cross_attn.to_q.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 32322048, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.10.cross_attn.to_q.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 32323584, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 32913408, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.10.mlp.mlp.0.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 32931840, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.10.mlp.mlp.0.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 768, |
| "padded_rows": 3072, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 32937984, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 35297280, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.10.mlp.mlp.2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 35371008, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.10.mlp.mlp.2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 3072, |
| "padded_rows": 768, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 0, |
| "offset": 35372544, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 37731840, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.10.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 37805568, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.10.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 37807104, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.10.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 37808640, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.10.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 37810176, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.10.self_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 37811712, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.10.self_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 37813248, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 38403072, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.10.self_attn.to_qkv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 38421504, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.10.self_attn.to_qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 768, |
| "padded_rows": 2304, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 38426112, |
| "bytes": 1769472 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 40195584, |
| "bytes": 55296 |
| } |
| }, |
| "blocks.11.adaLN_modulation.1.bias": { |
| "shape": [ |
| 4608 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 40250880, |
| "bytes": 9216 |
| } |
| }, |
| "blocks.11.adaLN_modulation.1.weight": { |
| "shape": [ |
| 4608, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4608, |
| "columns": 768, |
| "padded_rows": 4608, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 40260096, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 43799040, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.11.cross_attn.to_kv.bias": { |
| "shape": [ |
| 1536 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 43909632, |
| "bytes": 3072 |
| } |
| }, |
| "blocks.11.cross_attn.to_kv.weight": { |
| "shape": [ |
| 1536, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1536, |
| "columns": 768, |
| "padded_rows": 1536, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 43912704, |
| "bytes": 1179648 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 45092352, |
| "bytes": 36864 |
| } |
| }, |
| "blocks.11.cross_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 45129216, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.11.cross_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 45130752, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 45720576, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.11.cross_attn.to_q.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 45739008, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.11.cross_attn.to_q.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 45740544, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 46330368, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.11.mlp.mlp.0.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 46348800, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.11.mlp.mlp.0.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 768, |
| "padded_rows": 3072, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 46354944, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 48714240, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.11.mlp.mlp.2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 48787968, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.11.mlp.mlp.2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 3072, |
| "padded_rows": 768, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 0, |
| "offset": 48789504, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 51148800, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.11.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 51222528, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.11.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 51224064, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.11.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 51225600, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.11.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 51227136, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.11.self_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 51228672, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.11.self_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 51230208, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 51820032, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.11.self_attn.to_qkv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 51838464, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.11.self_attn.to_qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 768, |
| "padded_rows": 2304, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 51843072, |
| "bytes": 1769472 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 53612544, |
| "bytes": 55296 |
| } |
| }, |
| "blocks.2.adaLN_modulation.1.bias": { |
| "shape": [ |
| 4608 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 53667840, |
| "bytes": 9216 |
| } |
| }, |
| "blocks.2.adaLN_modulation.1.weight": { |
| "shape": [ |
| 4608, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4608, |
| "columns": 768, |
| "padded_rows": 4608, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 53677056, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 57216000, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.2.cross_attn.to_kv.bias": { |
| "shape": [ |
| 1536 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 57326592, |
| "bytes": 3072 |
| } |
| }, |
| "blocks.2.cross_attn.to_kv.weight": { |
| "shape": [ |
| 1536, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1536, |
| "columns": 768, |
| "padded_rows": 1536, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 57329664, |
| "bytes": 1179648 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 58509312, |
| "bytes": 36864 |
| } |
| }, |
| "blocks.2.cross_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 58546176, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.2.cross_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 58547712, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 59137536, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.2.cross_attn.to_q.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 59155968, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.2.cross_attn.to_q.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 59157504, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 59747328, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.2.mlp.mlp.0.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 59765760, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.2.mlp.mlp.0.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 768, |
| "padded_rows": 3072, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 59771904, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 62131200, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.2.mlp.mlp.2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 62204928, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.2.mlp.mlp.2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 3072, |
| "padded_rows": 768, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 0, |
| "offset": 62206464, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 64565760, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.2.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 64639488, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.2.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 64641024, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.2.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 64642560, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.2.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 64644096, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.2.self_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 64645632, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.2.self_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 64647168, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 65236992, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.2.self_attn.to_qkv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 65255424, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.2.self_attn.to_qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 768, |
| "padded_rows": 2304, |
| "padded_columns": 768, |
| "data": { |
| "shard": 0, |
| "offset": 65260032, |
| "bytes": 1769472 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 67029504, |
| "bytes": 55296 |
| } |
| }, |
| "blocks.3.adaLN_modulation.1.bias": { |
| "shape": [ |
| 4608 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 67084800, |
| "bytes": 9216 |
| } |
| }, |
| "blocks.3.adaLN_modulation.1.weight": { |
| "shape": [ |
| 4608, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4608, |
| "columns": 768, |
| "padded_rows": 4608, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 0, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 3538944, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.3.cross_attn.to_kv.bias": { |
| "shape": [ |
| 1536 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 3649536, |
| "bytes": 3072 |
| } |
| }, |
| "blocks.3.cross_attn.to_kv.weight": { |
| "shape": [ |
| 1536, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1536, |
| "columns": 768, |
| "padded_rows": 1536, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 3652608, |
| "bytes": 1179648 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 4832256, |
| "bytes": 36864 |
| } |
| }, |
| "blocks.3.cross_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 4869120, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.3.cross_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 4870656, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 5460480, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.3.cross_attn.to_q.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 5478912, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.3.cross_attn.to_q.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 5480448, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 6070272, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.3.mlp.mlp.0.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 6088704, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.3.mlp.mlp.0.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 768, |
| "padded_rows": 3072, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 6094848, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 8454144, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.3.mlp.mlp.2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 8527872, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.3.mlp.mlp.2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 3072, |
| "padded_rows": 768, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 1, |
| "offset": 8529408, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 10888704, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.3.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 10962432, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.3.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 10963968, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.3.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 10965504, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.3.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 10967040, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.3.self_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 10968576, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.3.self_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 10970112, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 11559936, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.3.self_attn.to_qkv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 11578368, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.3.self_attn.to_qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 768, |
| "padded_rows": 2304, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 11582976, |
| "bytes": 1769472 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 13352448, |
| "bytes": 55296 |
| } |
| }, |
| "blocks.4.adaLN_modulation.1.bias": { |
| "shape": [ |
| 4608 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 13407744, |
| "bytes": 9216 |
| } |
| }, |
| "blocks.4.adaLN_modulation.1.weight": { |
| "shape": [ |
| 4608, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4608, |
| "columns": 768, |
| "padded_rows": 4608, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 13416960, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 16955904, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.4.cross_attn.to_kv.bias": { |
| "shape": [ |
| 1536 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 17066496, |
| "bytes": 3072 |
| } |
| }, |
| "blocks.4.cross_attn.to_kv.weight": { |
| "shape": [ |
| 1536, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1536, |
| "columns": 768, |
| "padded_rows": 1536, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 17069568, |
| "bytes": 1179648 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 18249216, |
| "bytes": 36864 |
| } |
| }, |
| "blocks.4.cross_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 18286080, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.4.cross_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 18287616, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 18877440, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.4.cross_attn.to_q.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 18895872, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.4.cross_attn.to_q.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 18897408, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 19487232, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.4.mlp.mlp.0.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 19505664, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.4.mlp.mlp.0.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 768, |
| "padded_rows": 3072, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 19511808, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 21871104, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.4.mlp.mlp.2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 21944832, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.4.mlp.mlp.2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 3072, |
| "padded_rows": 768, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 1, |
| "offset": 21946368, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 24305664, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.4.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 24379392, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.4.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 24380928, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.4.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 24382464, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.4.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 24384000, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.4.self_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 24385536, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.4.self_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 24387072, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 24976896, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.4.self_attn.to_qkv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 24995328, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.4.self_attn.to_qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 768, |
| "padded_rows": 2304, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 24999936, |
| "bytes": 1769472 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 26769408, |
| "bytes": 55296 |
| } |
| }, |
| "blocks.5.adaLN_modulation.1.bias": { |
| "shape": [ |
| 4608 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 26824704, |
| "bytes": 9216 |
| } |
| }, |
| "blocks.5.adaLN_modulation.1.weight": { |
| "shape": [ |
| 4608, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4608, |
| "columns": 768, |
| "padded_rows": 4608, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 26833920, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 30372864, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.5.cross_attn.to_kv.bias": { |
| "shape": [ |
| 1536 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 30483456, |
| "bytes": 3072 |
| } |
| }, |
| "blocks.5.cross_attn.to_kv.weight": { |
| "shape": [ |
| 1536, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1536, |
| "columns": 768, |
| "padded_rows": 1536, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 30486528, |
| "bytes": 1179648 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 31666176, |
| "bytes": 36864 |
| } |
| }, |
| "blocks.5.cross_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 31703040, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.5.cross_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 31704576, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 32294400, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.5.cross_attn.to_q.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 32312832, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.5.cross_attn.to_q.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 32314368, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 32904192, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.5.mlp.mlp.0.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 32922624, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.5.mlp.mlp.0.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 768, |
| "padded_rows": 3072, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 32928768, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 35288064, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.5.mlp.mlp.2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 35361792, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.5.mlp.mlp.2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 3072, |
| "padded_rows": 768, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 1, |
| "offset": 35363328, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 37722624, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.5.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 37796352, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.5.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 37797888, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.5.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 37799424, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.5.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 37800960, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.5.self_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 37802496, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.5.self_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 37804032, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 38393856, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.5.self_attn.to_qkv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 38412288, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.5.self_attn.to_qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 768, |
| "padded_rows": 2304, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 38416896, |
| "bytes": 1769472 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 40186368, |
| "bytes": 55296 |
| } |
| }, |
| "blocks.6.adaLN_modulation.1.bias": { |
| "shape": [ |
| 4608 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 40241664, |
| "bytes": 9216 |
| } |
| }, |
| "blocks.6.adaLN_modulation.1.weight": { |
| "shape": [ |
| 4608, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4608, |
| "columns": 768, |
| "padded_rows": 4608, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 40250880, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 43789824, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.6.cross_attn.to_kv.bias": { |
| "shape": [ |
| 1536 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 43900416, |
| "bytes": 3072 |
| } |
| }, |
| "blocks.6.cross_attn.to_kv.weight": { |
| "shape": [ |
| 1536, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1536, |
| "columns": 768, |
| "padded_rows": 1536, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 43903488, |
| "bytes": 1179648 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 45083136, |
| "bytes": 36864 |
| } |
| }, |
| "blocks.6.cross_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 45120000, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.6.cross_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 45121536, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 45711360, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.6.cross_attn.to_q.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 45729792, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.6.cross_attn.to_q.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 45731328, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 46321152, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.6.mlp.mlp.0.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 46339584, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.6.mlp.mlp.0.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 768, |
| "padded_rows": 3072, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 46345728, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 48705024, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.6.mlp.mlp.2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 48778752, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.6.mlp.mlp.2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 3072, |
| "padded_rows": 768, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 1, |
| "offset": 48780288, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 51139584, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.6.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 51213312, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.6.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 51214848, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.6.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 51216384, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.6.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 51217920, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.6.self_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 51219456, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.6.self_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 51220992, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 51810816, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.6.self_attn.to_qkv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 51829248, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.6.self_attn.to_qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 768, |
| "padded_rows": 2304, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 51833856, |
| "bytes": 1769472 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 53603328, |
| "bytes": 55296 |
| } |
| }, |
| "blocks.7.adaLN_modulation.1.bias": { |
| "shape": [ |
| 4608 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 53658624, |
| "bytes": 9216 |
| } |
| }, |
| "blocks.7.adaLN_modulation.1.weight": { |
| "shape": [ |
| 4608, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4608, |
| "columns": 768, |
| "padded_rows": 4608, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 53667840, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 57206784, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.7.cross_attn.to_kv.bias": { |
| "shape": [ |
| 1536 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 57317376, |
| "bytes": 3072 |
| } |
| }, |
| "blocks.7.cross_attn.to_kv.weight": { |
| "shape": [ |
| 1536, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1536, |
| "columns": 768, |
| "padded_rows": 1536, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 57320448, |
| "bytes": 1179648 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 58500096, |
| "bytes": 36864 |
| } |
| }, |
| "blocks.7.cross_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 58536960, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.7.cross_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 58538496, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 59128320, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.7.cross_attn.to_q.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 59146752, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.7.cross_attn.to_q.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 59148288, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 59738112, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.7.mlp.mlp.0.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 59756544, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.7.mlp.mlp.0.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 768, |
| "padded_rows": 3072, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 59762688, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 62121984, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.7.mlp.mlp.2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 62195712, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.7.mlp.mlp.2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 3072, |
| "padded_rows": 768, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 1, |
| "offset": 62197248, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 64556544, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.7.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 64630272, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.7.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 64631808, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.7.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 64633344, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.7.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 64634880, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.7.self_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 64636416, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.7.self_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 64637952, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 65227776, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.7.self_attn.to_qkv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 65246208, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.7.self_attn.to_qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 768, |
| "padded_rows": 2304, |
| "padded_columns": 768, |
| "data": { |
| "shard": 1, |
| "offset": 65250816, |
| "bytes": 1769472 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 67020288, |
| "bytes": 55296 |
| } |
| }, |
| "blocks.8.adaLN_modulation.1.bias": { |
| "shape": [ |
| 4608 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 67075584, |
| "bytes": 9216 |
| } |
| }, |
| "blocks.8.adaLN_modulation.1.weight": { |
| "shape": [ |
| 4608, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4608, |
| "columns": 768, |
| "padded_rows": 4608, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 0, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 3538944, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.8.cross_attn.to_kv.bias": { |
| "shape": [ |
| 1536 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 3649536, |
| "bytes": 3072 |
| } |
| }, |
| "blocks.8.cross_attn.to_kv.weight": { |
| "shape": [ |
| 1536, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1536, |
| "columns": 768, |
| "padded_rows": 1536, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 3652608, |
| "bytes": 1179648 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 4832256, |
| "bytes": 36864 |
| } |
| }, |
| "blocks.8.cross_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 4869120, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.8.cross_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 4870656, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 5460480, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.8.cross_attn.to_q.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 5478912, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.8.cross_attn.to_q.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 5480448, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 6070272, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.8.mlp.mlp.0.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 6088704, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.8.mlp.mlp.0.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 768, |
| "padded_rows": 3072, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 6094848, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 8454144, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.8.mlp.mlp.2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 8527872, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.8.mlp.mlp.2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 3072, |
| "padded_rows": 768, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 2, |
| "offset": 8529408, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 10888704, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.8.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 10962432, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.8.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 10963968, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.8.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 10965504, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.8.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 10967040, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.8.self_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 10968576, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.8.self_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 10970112, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 11559936, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.8.self_attn.to_qkv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 11578368, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.8.self_attn.to_qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 768, |
| "padded_rows": 2304, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 11582976, |
| "bytes": 1769472 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 13352448, |
| "bytes": 55296 |
| } |
| }, |
| "blocks.9.adaLN_modulation.1.bias": { |
| "shape": [ |
| 4608 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 13407744, |
| "bytes": 9216 |
| } |
| }, |
| "blocks.9.adaLN_modulation.1.weight": { |
| "shape": [ |
| 4608, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4608, |
| "columns": 768, |
| "padded_rows": 4608, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 13416960, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 16955904, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.9.cross_attn.to_kv.bias": { |
| "shape": [ |
| 1536 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 17066496, |
| "bytes": 3072 |
| } |
| }, |
| "blocks.9.cross_attn.to_kv.weight": { |
| "shape": [ |
| 1536, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1536, |
| "columns": 768, |
| "padded_rows": 1536, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 17069568, |
| "bytes": 1179648 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 18249216, |
| "bytes": 36864 |
| } |
| }, |
| "blocks.9.cross_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 18286080, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.9.cross_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 18287616, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 18877440, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.9.cross_attn.to_q.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 18895872, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.9.cross_attn.to_q.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 18897408, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 19487232, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.9.mlp.mlp.0.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 19505664, |
| "bytes": 6144 |
| } |
| }, |
| "blocks.9.mlp.mlp.0.weight": { |
| "shape": [ |
| 3072, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 768, |
| "padded_rows": 3072, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 19511808, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 21871104, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.9.mlp.mlp.2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 21944832, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.9.mlp.mlp.2.weight": { |
| "shape": [ |
| 768, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 3072, |
| "padded_rows": 768, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 2, |
| "offset": 21946368, |
| "bytes": 2359296 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 24305664, |
| "bytes": 73728 |
| } |
| }, |
| "blocks.9.norm2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 24379392, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.9.norm2.weight": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 24380928, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.9.self_attn.k_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 24382464, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.9.self_attn.q_rms_norm.gamma": { |
| "shape": [ |
| 12, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 24384000, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.9.self_attn.to_out.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 24385536, |
| "bytes": 1536 |
| } |
| }, |
| "blocks.9.self_attn.to_out.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 24387072, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 24976896, |
| "bytes": 18432 |
| } |
| }, |
| "blocks.9.self_attn.to_qkv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 24995328, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.9.self_attn.to_qkv.weight": { |
| "shape": [ |
| 2304, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 768, |
| "padded_rows": 2304, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 24999936, |
| "bytes": 1769472 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 26769408, |
| "bytes": 55296 |
| } |
| }, |
| "input_layer.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 26824704, |
| "bytes": 1536 |
| } |
| }, |
| "input_layer.weight": { |
| "shape": [ |
| 768, |
| 8 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 8, |
| "padded_rows": 768, |
| "padded_columns": 64, |
| "data": { |
| "shard": 2, |
| "offset": 26826240, |
| "bytes": 49152 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 26875392, |
| "bytes": 1536 |
| } |
| }, |
| "out_layer.bias": { |
| "shape": [ |
| 8 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 26876928, |
| "bytes": 16 |
| } |
| }, |
| "out_layer.weight": { |
| "shape": [ |
| 8, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 8, |
| "columns": 768, |
| "padded_rows": 16, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 26877184, |
| "bytes": 12288 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 26889472, |
| "bytes": 384 |
| } |
| }, |
| "pos_emb": { |
| "shape": [ |
| 4096, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 4096, |
| "columns": 768, |
| "padded_rows": 4096, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 26889984, |
| "bytes": 3145728 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 30035712, |
| "bytes": 98304 |
| } |
| }, |
| "t_embedder.mlp.0.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 30134016, |
| "bytes": 1536 |
| } |
| }, |
| "t_embedder.mlp.0.weight": { |
| "shape": [ |
| 768, |
| 256 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 256, |
| "padded_rows": 768, |
| "padded_columns": 256, |
| "data": { |
| "shard": 2, |
| "offset": 30135552, |
| "bytes": 196608 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 30332160, |
| "bytes": 6144 |
| } |
| }, |
| "t_embedder.mlp.2.bias": { |
| "shape": [ |
| 768 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 30338304, |
| "bytes": 1536 |
| } |
| }, |
| "t_embedder.mlp.2.weight": { |
| "shape": [ |
| 768, |
| 768 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 768, |
| "columns": 768, |
| "padded_rows": 768, |
| "padded_columns": 768, |
| "data": { |
| "shard": 2, |
| "offset": 30339840, |
| "bytes": 589824 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 30929664, |
| "bytes": 18432 |
| } |
| } |
| } |
| } |
|
|