| { |
| "schema": "trellis-webgpu-pack", |
| "version": 1, |
| "stage": "litoDit", |
| "upstreamRevision": "apple/ml-lito", |
| "shards": [ |
| { |
| "file": "stage-000.bin", |
| "bytes": 65852928, |
| "sha256": "7de9a67bae06a6d2301dacf0e85814b34d984931ba280cb3c11b34ea6b1c5d7b" |
| }, |
| { |
| "file": "stage-001.bin", |
| "bytes": 65850624, |
| "sha256": "fd42f8efedce6b16e2f8022121575d4d5aecf9ffc3fc7fe745f2642f22347dd6" |
| }, |
| { |
| "file": "stage-002.bin", |
| "bytes": 65850624, |
| "sha256": "ba1a61741d9f14d940f6c98f1433ad31d80ce266e053223061c0d69acce8393b" |
| }, |
| { |
| "file": "stage-003.bin", |
| "bytes": 65850624, |
| "sha256": "3d7d5a3196d49b21269e4e1e754a03197a423d9f33048d16ede6a29b092c5930" |
| }, |
| { |
| "file": "stage-004.bin", |
| "bytes": 65850624, |
| "sha256": "b20864fc25a9626d6c942c5aab058ac254217d5a134541a17a01f73eb407b60c" |
| }, |
| { |
| "file": "stage-005.bin", |
| "bytes": 65850624, |
| "sha256": "69ec3be242bb5fe5b4ddc69705068785659ee4e5da7635802b1a9a30f1c8c224" |
| }, |
| { |
| "file": "stage-006.bin", |
| "bytes": 65850624, |
| "sha256": "7c0b1e060f36672f74d09106a5fb027b1e8d3eadbbaab2aaa82dda58de2ff5b4" |
| }, |
| { |
| "file": "stage-007.bin", |
| "bytes": 65850624, |
| "sha256": "bab9ac5b12fee5b9507ccc0d1bfe73c42f2bec19768e7c5092a5e927cabf7727" |
| }, |
| { |
| "file": "stage-008.bin", |
| "bytes": 65850624, |
| "sha256": "02155f6450092552a7e9e1f58d2fc5615a9da8a7ae1df6325a08a70f1332129c" |
| }, |
| { |
| "file": "stage-009.bin", |
| "bytes": 58917632, |
| "sha256": "d47a586161cc2cf9c6def5e4a7e9bd75555da3b1fc23f1c359fc15eac9b386d5" |
| }, |
| { |
| "file": "stage-010.bin", |
| "bytes": 18812416, |
| "sha256": "2c60600195e1c3a15a861282301de6633afc1dabd599479f9214a83b3f55ff06" |
| } |
| ], |
| "tensors": { |
| "blocks.0.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 0, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.0.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 2304, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 1329408, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.0.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 1370880, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.0.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 1377792, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 5359104, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.0.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 5483520, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.0.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 5485824, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.0.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 5488128, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.0.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 5490432, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.0.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 5492736, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.0.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 5495040, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.0.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 5497344, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.0.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 5501952, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 8156160, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.0.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 8239104, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.0.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 8241408, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 9568512, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.0.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 9609984, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.0.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 9612288, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 10939392, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.0.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 10980864, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.0.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 10983168, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.0.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 10985472, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 14524416, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.0.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 14635008, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.0.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 0, |
| "offset": 14637312, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 18176256, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.0.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 18286848, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 21825792, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.0.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 21936384, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.1.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 21950208, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.1.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 21952512, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 23279616, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.1.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 23321088, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.1.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 23328000, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 27309312, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.1.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 27433728, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.1.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 27436032, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.1.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 27438336, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.1.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 27440640, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.1.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 27442944, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.1.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 27445248, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.1.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 27447552, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.1.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 27452160, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 30106368, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.1.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 30189312, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.1.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 30191616, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 31518720, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.1.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 31560192, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.1.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 31562496, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 32889600, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.1.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 32931072, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.1.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 32933376, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.1.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 32935680, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 36474624, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.1.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 36585216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.1.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 0, |
| "offset": 36587520, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 40126464, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.1.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 40237056, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 43776000, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.1.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 43886592, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.10.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 43900416, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.10.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 43902720, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 45229824, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.10.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 45271296, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.10.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 45278208, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 49259520, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.10.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 49383936, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.10.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 49386240, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.10.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 49388544, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.10.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 49390848, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.10.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 49393152, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.10.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 49395456, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.10.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 49397760, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.10.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 49402368, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 52056576, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.10.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 52139520, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.10.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 52141824, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 53468928, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.10.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 53510400, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.10.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 53512704, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 54839808, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.10.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 54881280, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.10.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 54883584, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.10.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 54885888, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 58424832, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.10.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 58535424, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.10.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 0, |
| "offset": 58537728, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 62076672, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.10.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 0, |
| "offset": 62187264, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 65726208, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.10.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 65836800, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.11.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 65850624, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.11.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 0, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 1327104, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.11.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 1368576, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.11.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 1375488, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 5356800, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.11.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 5481216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.11.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 5483520, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.11.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 5485824, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.11.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 5488128, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.11.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 5490432, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.11.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 5492736, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.11.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 5495040, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.11.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 5499648, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 8153856, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.11.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 8236800, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.11.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 8239104, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 9566208, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.11.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 9607680, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.11.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 9609984, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 10937088, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.11.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 10978560, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.11.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 10980864, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.11.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 10983168, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 14522112, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.11.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 14632704, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.11.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 1, |
| "offset": 14635008, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 18173952, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.11.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 18284544, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 21823488, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.11.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 21934080, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.12.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 21947904, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.12.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 21950208, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 23277312, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.12.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 23318784, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.12.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 23325696, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 27307008, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.12.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 27431424, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.12.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 27433728, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.12.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 27436032, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.12.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 27438336, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.12.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 27440640, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.12.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 27442944, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.12.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 27445248, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.12.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 27449856, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 30104064, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.12.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 30187008, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.12.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 30189312, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 31516416, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.12.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 31557888, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.12.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 31560192, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 32887296, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.12.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 32928768, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.12.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 32931072, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.12.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 32933376, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 36472320, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.12.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 36582912, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.12.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 1, |
| "offset": 36585216, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 40124160, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.12.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 40234752, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 43773696, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.12.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 43884288, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.13.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 43898112, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.13.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 43900416, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 45227520, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.13.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 45268992, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.13.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 45275904, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 49257216, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.13.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 49381632, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.13.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 49383936, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.13.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 49386240, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.13.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 49388544, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.13.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 49390848, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.13.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 49393152, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.13.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 49395456, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.13.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 49400064, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 52054272, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.13.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 52137216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.13.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 52139520, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 53466624, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.13.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 53508096, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.13.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 53510400, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 54837504, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.13.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 54878976, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.13.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 54881280, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.13.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 54883584, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 58422528, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.13.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 58533120, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.13.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 1, |
| "offset": 58535424, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 62074368, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.13.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 1, |
| "offset": 62184960, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 65723904, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.13.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 65834496, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.14.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 65848320, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.14.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 0, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 1327104, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.14.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 1368576, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.14.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 1375488, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 5356800, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.14.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 5481216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.14.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 5483520, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.14.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 5485824, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.14.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 5488128, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.14.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 5490432, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.14.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 5492736, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.14.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 5495040, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.14.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 5499648, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 8153856, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.14.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 8236800, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.14.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 8239104, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 9566208, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.14.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 9607680, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.14.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 9609984, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 10937088, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.14.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 10978560, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.14.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 10980864, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.14.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 10983168, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 14522112, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.14.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 14632704, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.14.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 2, |
| "offset": 14635008, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 18173952, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.14.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 18284544, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 21823488, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.14.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 21934080, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.15.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 21947904, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.15.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 21950208, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 23277312, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.15.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 23318784, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.15.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 23325696, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 27307008, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.15.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 27431424, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.15.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 27433728, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.15.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 27436032, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.15.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 27438336, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.15.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 27440640, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.15.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 27442944, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.15.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 27445248, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.15.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 27449856, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 30104064, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.15.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 30187008, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.15.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 30189312, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 31516416, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.15.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 31557888, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.15.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 31560192, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 32887296, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.15.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 32928768, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.15.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 32931072, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.15.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 32933376, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 36472320, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.15.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 36582912, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.15.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 2, |
| "offset": 36585216, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 40124160, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.15.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 40234752, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 43773696, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.15.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 43884288, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.16.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 43898112, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.16.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 43900416, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 45227520, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.16.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 45268992, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.16.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 45275904, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 49257216, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.16.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 49381632, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.16.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 49383936, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.16.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 49386240, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.16.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 49388544, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.16.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 49390848, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.16.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 49393152, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.16.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 49395456, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.16.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 49400064, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 52054272, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.16.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 52137216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.16.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 52139520, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 53466624, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.16.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 53508096, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.16.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 53510400, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 54837504, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.16.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 54878976, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.16.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 54881280, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.16.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 54883584, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 58422528, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.16.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 58533120, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.16.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 2, |
| "offset": 58535424, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 62074368, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.16.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 2, |
| "offset": 62184960, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 65723904, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.16.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 65834496, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.17.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 65848320, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.17.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 0, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 1327104, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.17.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 1368576, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.17.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 1375488, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 5356800, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.17.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 5481216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.17.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 5483520, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.17.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 5485824, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.17.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 5488128, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.17.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 5490432, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.17.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 5492736, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.17.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 5495040, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.17.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 5499648, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 8153856, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.17.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 8236800, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.17.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 8239104, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 9566208, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.17.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 9607680, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.17.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 9609984, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 10937088, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.17.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 10978560, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.17.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 10980864, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.17.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 10983168, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 14522112, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.17.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 14632704, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.17.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 3, |
| "offset": 14635008, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 18173952, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.17.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 18284544, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 21823488, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.17.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 21934080, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.18.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 21947904, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.18.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 21950208, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 23277312, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.18.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 23318784, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.18.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 23325696, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 27307008, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.18.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 27431424, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.18.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 27433728, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.18.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 27436032, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.18.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 27438336, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.18.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 27440640, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.18.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 27442944, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.18.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 27445248, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.18.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 27449856, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 30104064, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.18.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 30187008, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.18.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 30189312, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 31516416, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.18.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 31557888, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.18.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 31560192, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 32887296, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.18.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 32928768, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.18.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 32931072, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.18.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 32933376, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 36472320, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.18.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 36582912, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.18.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 3, |
| "offset": 36585216, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 40124160, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.18.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 40234752, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 43773696, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.18.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 43884288, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.19.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 43898112, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.19.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 43900416, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 45227520, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.19.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 45268992, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.19.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 45275904, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 49257216, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.19.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 49381632, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.19.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 49383936, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.19.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 49386240, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.19.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 49388544, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.19.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 49390848, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.19.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 49393152, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.19.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 49395456, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.19.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 49400064, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 52054272, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.19.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 52137216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.19.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 52139520, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 53466624, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.19.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 53508096, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.19.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 53510400, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 54837504, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.19.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 54878976, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.19.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 54881280, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.19.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 54883584, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 58422528, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.19.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 58533120, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.19.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 3, |
| "offset": 58535424, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 62074368, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.19.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 3, |
| "offset": 62184960, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 65723904, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.19.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 65834496, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.2.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 65848320, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.2.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 0, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 1327104, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.2.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 1368576, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.2.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 1375488, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 5356800, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.2.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 5481216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.2.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 5483520, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.2.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 5485824, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.2.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 5488128, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.2.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 5490432, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.2.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 5492736, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.2.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 5495040, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.2.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 5499648, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 8153856, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.2.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 8236800, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.2.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 8239104, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 9566208, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.2.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 9607680, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.2.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 9609984, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 10937088, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.2.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 10978560, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.2.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 10980864, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.2.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 10983168, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 14522112, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.2.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 14632704, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.2.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 4, |
| "offset": 14635008, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 18173952, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.2.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 18284544, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 21823488, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.2.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 21934080, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.20.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 21947904, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.20.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 21950208, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 23277312, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.20.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 23318784, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.20.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 23325696, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 27307008, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.20.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 27431424, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.20.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 27433728, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.20.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 27436032, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.20.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 27438336, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.20.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 27440640, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.20.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 27442944, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.20.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 27445248, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.20.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 27449856, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 30104064, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.20.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 30187008, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.20.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 30189312, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 31516416, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.20.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 31557888, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.20.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 31560192, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 32887296, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.20.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 32928768, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.20.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 32931072, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.20.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 32933376, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 36472320, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.20.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 36582912, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.20.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 4, |
| "offset": 36585216, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 40124160, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.20.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 40234752, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 43773696, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.20.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 43884288, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.21.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 43898112, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.21.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 43900416, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 45227520, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.21.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 45268992, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.21.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 45275904, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 49257216, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.21.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 49381632, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.21.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 49383936, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.21.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 49386240, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.21.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 49388544, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.21.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 49390848, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.21.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 49393152, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.21.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 49395456, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.21.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 49400064, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 52054272, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.21.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 52137216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.21.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 52139520, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 53466624, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.21.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 53508096, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.21.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 53510400, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 54837504, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.21.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 54878976, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.21.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 54881280, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.21.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 54883584, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 58422528, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.21.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 58533120, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.21.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 4, |
| "offset": 58535424, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 62074368, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.21.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 4, |
| "offset": 62184960, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 65723904, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.21.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 65834496, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.22.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 65848320, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.22.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 0, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 1327104, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.22.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 1368576, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.22.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 1375488, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 5356800, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.22.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 5481216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.22.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 5483520, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.22.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 5485824, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.22.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 5488128, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.22.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 5490432, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.22.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 5492736, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.22.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 5495040, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.22.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 5499648, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 8153856, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.22.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 8236800, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.22.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 8239104, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 9566208, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.22.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 9607680, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.22.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 9609984, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 10937088, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.22.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 10978560, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.22.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 10980864, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.22.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 10983168, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 14522112, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.22.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 14632704, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.22.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 5, |
| "offset": 14635008, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 18173952, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.22.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 18284544, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 21823488, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.22.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 21934080, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.23.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 21947904, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.23.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 21950208, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 23277312, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.23.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 23318784, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.23.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 23325696, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 27307008, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.23.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 27431424, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.23.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 27433728, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.23.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 27436032, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.23.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 27438336, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.23.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 27440640, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.23.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 27442944, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.23.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 27445248, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.23.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 27449856, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 30104064, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.23.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 30187008, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.23.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 30189312, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 31516416, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.23.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 31557888, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.23.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 31560192, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 32887296, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.23.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 32928768, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.23.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 32931072, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.23.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 32933376, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 36472320, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.23.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 36582912, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.23.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 5, |
| "offset": 36585216, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 40124160, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.23.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 40234752, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 43773696, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.23.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 43884288, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.24.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 43898112, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.24.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 43900416, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 45227520, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.24.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 45268992, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.24.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 45275904, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 49257216, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.24.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 49381632, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.24.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 49383936, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.24.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 49386240, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.24.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 49388544, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.24.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 49390848, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.24.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 49393152, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.24.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 49395456, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.24.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 49400064, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 52054272, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.24.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 52137216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.24.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 52139520, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 53466624, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.24.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 53508096, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.24.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 53510400, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 54837504, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.24.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 54878976, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.24.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 54881280, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.24.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 54883584, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 58422528, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.24.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 58533120, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.24.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 5, |
| "offset": 58535424, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 62074368, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.24.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 5, |
| "offset": 62184960, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 65723904, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.24.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 65834496, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.25.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 65848320, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.25.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 0, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 1327104, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.25.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 1368576, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.25.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 1375488, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 5356800, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.25.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 5481216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.25.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 5483520, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.25.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 5485824, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.25.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 5488128, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.25.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 5490432, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.25.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 5492736, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.25.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 5495040, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.25.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 5499648, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 8153856, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.25.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 8236800, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.25.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 8239104, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 9566208, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.25.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 9607680, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.25.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 9609984, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 10937088, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.25.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 10978560, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.25.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 10980864, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.25.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 10983168, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 14522112, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.25.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 14632704, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.25.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 6, |
| "offset": 14635008, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 18173952, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.25.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 18284544, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 21823488, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.25.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 21934080, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.26.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 21947904, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.26.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 21950208, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 23277312, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.26.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 23318784, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.26.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 23325696, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 27307008, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.26.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 27431424, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.26.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 27433728, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.26.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 27436032, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.26.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 27438336, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.26.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 27440640, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.26.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 27442944, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.26.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 27445248, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.26.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 27449856, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 30104064, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.26.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 30187008, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.26.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 30189312, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 31516416, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.26.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 31557888, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.26.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 31560192, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 32887296, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.26.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 32928768, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.26.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 32931072, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.26.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 32933376, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 36472320, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.26.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 36582912, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.26.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 6, |
| "offset": 36585216, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 40124160, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.26.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 40234752, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 43773696, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.26.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 43884288, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.27.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 43898112, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.27.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 43900416, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 45227520, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.27.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 45268992, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.27.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 45275904, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 49257216, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.27.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 49381632, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.27.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 49383936, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.27.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 49386240, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.27.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 49388544, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.27.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 49390848, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.27.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 49393152, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.27.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 49395456, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.27.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 49400064, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 52054272, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.27.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 52137216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.27.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 52139520, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 53466624, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.27.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 53508096, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.27.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 53510400, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 54837504, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.27.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 54878976, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.27.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 54881280, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.27.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 54883584, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 58422528, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.27.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 58533120, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.27.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 6, |
| "offset": 58535424, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 62074368, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.27.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 6, |
| "offset": 62184960, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 65723904, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.27.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 65834496, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.3.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 65848320, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.3.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 0, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 1327104, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.3.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 1368576, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.3.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 1375488, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 5356800, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.3.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 5481216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.3.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 5483520, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.3.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 5485824, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.3.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 5488128, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.3.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 5490432, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.3.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 5492736, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.3.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 5495040, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.3.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 5499648, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 8153856, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.3.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 8236800, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.3.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 8239104, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 9566208, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.3.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 9607680, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.3.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 9609984, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 10937088, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.3.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 10978560, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.3.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 10980864, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.3.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 10983168, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 14522112, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.3.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 14632704, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.3.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 7, |
| "offset": 14635008, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 18173952, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.3.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 18284544, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 21823488, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.3.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 21934080, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.4.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 21947904, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.4.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 21950208, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 23277312, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.4.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 23318784, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.4.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 23325696, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 27307008, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.4.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 27431424, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.4.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 27433728, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.4.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 27436032, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.4.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 27438336, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.4.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 27440640, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.4.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 27442944, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.4.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 27445248, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.4.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 27449856, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 30104064, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.4.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 30187008, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.4.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 30189312, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 31516416, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.4.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 31557888, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.4.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 31560192, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 32887296, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.4.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 32928768, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.4.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 32931072, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.4.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 32933376, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 36472320, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.4.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 36582912, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.4.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 7, |
| "offset": 36585216, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 40124160, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.4.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 40234752, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 43773696, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.4.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 43884288, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.5.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 43898112, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.5.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 43900416, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 45227520, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.5.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 45268992, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.5.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 45275904, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 49257216, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.5.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 49381632, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.5.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 49383936, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.5.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 49386240, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.5.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 49388544, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.5.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 49390848, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.5.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 49393152, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.5.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 49395456, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.5.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 49400064, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 52054272, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.5.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 52137216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.5.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 52139520, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 53466624, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.5.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 53508096, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.5.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 53510400, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 54837504, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.5.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 54878976, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.5.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 54881280, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.5.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 54883584, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 58422528, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.5.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 58533120, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.5.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 7, |
| "offset": 58535424, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 62074368, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.5.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 7, |
| "offset": 62184960, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 65723904, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.5.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 65834496, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.6.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 65848320, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.6.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 0, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 1327104, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.6.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 1368576, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.6.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 1375488, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 5356800, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.6.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 5481216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.6.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 5483520, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.6.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 5485824, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.6.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 5488128, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.6.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 5490432, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.6.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 5492736, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.6.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 5495040, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.6.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 5499648, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 8153856, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.6.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 8236800, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.6.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 8239104, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 9566208, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.6.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 9607680, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.6.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 9609984, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 10937088, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.6.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 10978560, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.6.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 10980864, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.6.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 10983168, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 14522112, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.6.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 14632704, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.6.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 8, |
| "offset": 14635008, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 18173952, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.6.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 18284544, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 21823488, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.6.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 21934080, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.7.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 21947904, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.7.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 21950208, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 23277312, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.7.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 23318784, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.7.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 23325696, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 27307008, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.7.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 27431424, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.7.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 27433728, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.7.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 27436032, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.7.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 27438336, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.7.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 27440640, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.7.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 27442944, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.7.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 27445248, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.7.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 27449856, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 30104064, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.7.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 30187008, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.7.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 30189312, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 31516416, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.7.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 31557888, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.7.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 31560192, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 32887296, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.7.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 32928768, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.7.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 32931072, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.7.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 32933376, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 36472320, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.7.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 36582912, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.7.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 8, |
| "offset": 36585216, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 40124160, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.7.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 40234752, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 43773696, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.7.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 43884288, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.8.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 43898112, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.8.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 43900416, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 45227520, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.8.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 45268992, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.8.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 45275904, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 49257216, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.8.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 49381632, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.8.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 49383936, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.8.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 49386240, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.8.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 49388544, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.8.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 49390848, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.8.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 49393152, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.8.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 49395456, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.8.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 49400064, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 52054272, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.8.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 52137216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.8.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 52139520, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 53466624, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.8.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 53508096, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.8.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 53510400, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 54837504, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.8.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 54878976, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.8.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 54881280, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.8.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 54883584, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 58422528, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.8.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 58533120, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.8.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 8, |
| "offset": 58535424, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 62074368, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.8.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 8, |
| "offset": 62184960, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 65723904, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.8.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 65834496, |
| "bytes": 13824 |
| } |
| }, |
| "blocks.9.attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 65848320, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.9.attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 9, |
| "offset": 0, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 1327104, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.9.attn.linear_qkv.bias": { |
| "shape": [ |
| 3456 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 1368576, |
| "bytes": 6912 |
| } |
| }, |
| "blocks.9.attn.linear_qkv.weight": { |
| "shape": [ |
| 3456, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3456, |
| "columns": 1152, |
| "padded_rows": 3456, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 9, |
| "offset": 1375488, |
| "bytes": 3981312 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 5356800, |
| "bytes": 124416 |
| } |
| }, |
| "blocks.9.attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 5481216, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.9.attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 5483520, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.9.cross_attn.layernorm_kv.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 5485824, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.9.cross_attn.layernorm_kv.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 5488128, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.9.cross_attn.layernorm_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 5490432, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.9.cross_attn.layernorm_q.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 5492736, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.9.cross_attn.linear_kv.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 5495040, |
| "bytes": 4608 |
| } |
| }, |
| "blocks.9.cross_attn.linear_kv.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 2304, |
| "columns": 1152, |
| "padded_rows": 2304, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 9, |
| "offset": 5499648, |
| "bytes": 2654208 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 8153856, |
| "bytes": 82944 |
| } |
| }, |
| "blocks.9.cross_attn.linear_out.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 8236800, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.9.cross_attn.linear_out.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 9, |
| "offset": 8239104, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 9566208, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.9.cross_attn.linear_q.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 9607680, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.9.cross_attn.linear_q.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 1152, |
| "padded_rows": 1152, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 9, |
| "offset": 9609984, |
| "bytes": 1327104 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 10937088, |
| "bytes": 41472 |
| } |
| }, |
| "blocks.9.cross_attn.rmsnorm_k.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 10978560, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.9.cross_attn.rmsnorm_q.scale": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 10980864, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.9.mlp.w1.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 9, |
| "offset": 10983168, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 14522112, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.9.mlp.w2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 14632704, |
| "bytes": 2304 |
| } |
| }, |
| "blocks.9.mlp.w2.weight": { |
| "shape": [ |
| 1152, |
| 3072 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1152, |
| "columns": 3072, |
| "padded_rows": 1152, |
| "padded_columns": 3072, |
| "data": { |
| "shard": 9, |
| "offset": 14635008, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 18173952, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.9.mlp.w3.weight": { |
| "shape": [ |
| 3072, |
| 1152 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 3072, |
| "columns": 1152, |
| "padded_rows": 3072, |
| "padded_columns": 1152, |
| "data": { |
| "shard": 9, |
| "offset": 18284544, |
| "bytes": 3538944 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 21823488, |
| "bytes": 110592 |
| } |
| }, |
| "blocks.9.scale_shift_table": { |
| "shape": [ |
| 6, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 21934080, |
| "bytes": 13824 |
| } |
| }, |
| "cond_embedder.y_embedding": { |
| "shape": [ |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 21947904, |
| "bytes": 4096 |
| } |
| }, |
| "cond_embedder.y_proj.fc1.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 21952000, |
| "bytes": 2304 |
| } |
| }, |
| "cond_embedder.y_proj.fc1.weight": { |
| "shape": [ |
| 1152, |
| 2048 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 21954304, |
| "bytes": 4718592 |
| } |
| }, |
| "cond_embedder.y_proj.fc2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 26672896, |
| "bytes": 2304 |
| } |
| }, |
| "cond_embedder.y_proj.fc2.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 26675200, |
| "bytes": 2654208 |
| } |
| }, |
| "final_layer.adaLN_linear1.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 29329408, |
| "bytes": 2304 |
| } |
| }, |
| "final_layer.adaLN_linear1.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 29331712, |
| "bytes": 2654208 |
| } |
| }, |
| "final_layer.adaLN_linear2.bias": { |
| "shape": [ |
| 2304 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 31985920, |
| "bytes": 4608 |
| } |
| }, |
| "final_layer.adaLN_linear2.weight": { |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 31990528, |
| "bytes": 5308416 |
| } |
| }, |
| "final_layer.linear.bias": { |
| "shape": [ |
| 32 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 37298944, |
| "bytes": 64 |
| } |
| }, |
| "final_layer.linear.weight": { |
| "shape": [ |
| 32, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 37299200, |
| "bytes": 73728 |
| } |
| }, |
| "pos_mtx": { |
| "shape": [ |
| 8192, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 37372928, |
| "bytes": 18874368 |
| } |
| }, |
| "pos_proj.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 56247296, |
| "bytes": 2304 |
| } |
| }, |
| "pos_proj.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 56249600, |
| "bytes": 2654208 |
| } |
| }, |
| "t0_proj_linear.bias": { |
| "shape": [ |
| 6912 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 58903808, |
| "bytes": 13824 |
| } |
| }, |
| "t0_proj_linear.weight": { |
| "shape": [ |
| 6912, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 0, |
| "bytes": 15925248 |
| } |
| }, |
| "t_embedder.freq_bands": { |
| "shape": [ |
| 32 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 15925248, |
| "bytes": 64 |
| } |
| }, |
| "t_proj_linear1.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 15925504, |
| "bytes": 2304 |
| } |
| }, |
| "t_proj_linear1.weight": { |
| "shape": [ |
| 1152, |
| 64 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 15927808, |
| "bytes": 147456 |
| } |
| }, |
| "t_proj_linear2.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 16075264, |
| "bytes": 2304 |
| } |
| }, |
| "t_proj_linear2.weight": { |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 16077568, |
| "bytes": 2654208 |
| } |
| }, |
| "z_proj.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 18731776, |
| "bytes": 2304 |
| } |
| }, |
| "z_proj.weight": { |
| "shape": [ |
| 1152, |
| 32 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 18734080, |
| "bytes": 73728 |
| } |
| }, |
| "z_proj_ln.bias": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 18807808, |
| "bytes": 2304 |
| } |
| }, |
| "z_proj_ln.weight": { |
| "shape": [ |
| 1152 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 18810112, |
| "bytes": 2304 |
| } |
| } |
| } |
| } |
|
|
