| { |
| "schema": "trellis-webgpu-pack", |
| "version": 1, |
| "stage": "idolImageEncoder", |
| "upstreamRevision": "c807c9eeea853df70aec4069e6f56b28ddc82acc", |
| "shards": [ |
| { |
| "file": "stage-000-4efa5887da632c71dbb72a4f49c2e5862b644f4df059d3304f3b261f9fed9b7c.bin", |
| "bytes": 62965760, |
| "sha256": "4efa5887da632c71dbb72a4f49c2e5862b644f4df059d3304f3b261f9fed9b7c" |
| }, |
| { |
| "file": "stage-001-e2a77645ef350e46311b75baa16b73cc6d3bcab5ac62fac842ff0c79efc35fb7.bin", |
| "bytes": 60958720, |
| "sha256": "e2a77645ef350e46311b75baa16b73cc6d3bcab5ac62fac842ff0c79efc35fb7" |
| }, |
| { |
| "file": "stage-002-3466a0562d586e5ec7cf1f91565a30053bb1e98693e68c106413de2c3f5a4255.bin", |
| "bytes": 60958720, |
| "sha256": "3466a0562d586e5ec7cf1f91565a30053bb1e98693e68c106413de2c3f5a4255" |
| }, |
| { |
| "file": "stage-003-1166b2f2a95392f30273cb5fb8119c1277549a78bf5abfadb208ff8c3c9163f7.bin", |
| "bytes": 66045440, |
| "sha256": "1166b2f2a95392f30273cb5fb8119c1277549a78bf5abfadb208ff8c3c9163f7" |
| }, |
| { |
| "file": "stage-004-114839fb2170c54de590d0c85e4137b61101fb9a26f4c58249f0c62ad0461279.bin", |
| "bytes": 62661120, |
| "sha256": "114839fb2170c54de590d0c85e4137b61101fb9a26f4c58249f0c62ad0461279" |
| }, |
| { |
| "file": "stage-005-63cc926ba0b7a49999c02d7038d6a6eb01471632ac82e2f8baeb89e49cc2ae24.bin", |
| "bytes": 60958720, |
| "sha256": "63cc926ba0b7a49999c02d7038d6a6eb01471632ac82e2f8baeb89e49cc2ae24" |
| }, |
| { |
| "file": "stage-006-08fa8d08e95924e941936a7b32f067315170539b94b0e509fdee1131f17ed224.bin", |
| "bytes": 66045440, |
| "sha256": "08fa8d08e95924e941936a7b32f067315170539b94b0e509fdee1131f17ed224" |
| }, |
| { |
| "file": "stage-007-7f9f2c072d7f23155fa6c64265e9428176012a309207da63ceefe8017ec7efc0.bin", |
| "bytes": 62661120, |
| "sha256": "7f9f2c072d7f23155fa6c64265e9428176012a309207da63ceefe8017ec7efc0" |
| }, |
| { |
| "file": "stage-008-5cf4735c48ed29c31568196cde7220853de5ee941726c018e605bc87e84bca14.bin", |
| "bytes": 60958720, |
| "sha256": "5cf4735c48ed29c31568196cde7220853de5ee941726c018e605bc87e84bca14" |
| }, |
| { |
| "file": "stage-009-67f348278ccc791280d69408c20d7069276172e47579293866104165afff1fac.bin", |
| "bytes": 66045440, |
| "sha256": "67f348278ccc791280d69408c20d7069276172e47579293866104165afff1fac" |
| }, |
| { |
| "file": "stage-010-44691ab584b208202ea2eb3b88438c98ce4514a76a2dfe9bcde6b025457ea025.bin", |
| "bytes": 62661120, |
| "sha256": "44691ab584b208202ea2eb3b88438c98ce4514a76a2dfe9bcde6b025457ea025" |
| }, |
| { |
| "file": "stage-011-e93a12f51e968be3b3abd3a9233d62c84bb529a637ebc944da89b32b0aa0ee0e.bin", |
| "bytes": 60958720, |
| "sha256": "e93a12f51e968be3b3abd3a9233d62c84bb529a637ebc944da89b32b0aa0ee0e" |
| }, |
| { |
| "file": "stage-012-4c5131918ef0a6f6477e7af3fe4f0a5821abe344e0d3bb59919bc087ace33970.bin", |
| "bytes": 66045440, |
| "sha256": "4c5131918ef0a6f6477e7af3fe4f0a5821abe344e0d3bb59919bc087ace33970" |
| }, |
| { |
| "file": "stage-013-5e1fe8107016cb2e503bd6569c5a4d7551e291315c757486c3815ee4f4e6fc7c.bin", |
| "bytes": 49123840, |
| "sha256": "5e1fe8107016cb2e503bd6569c5a4d7551e291315c757486c3815ee4f4e6fc7c" |
| } |
| ], |
| "tensors": { |
| "embeddings.cls_token": { |
| "shape": [ |
| 1, |
| 1, |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 0, |
| "bytes": 5120 |
| } |
| }, |
| "embeddings.mask_token": { |
| "shape": [ |
| 1, |
| 1, |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 5120, |
| "bytes": 2560 |
| } |
| }, |
| "embeddings.patch_embeddings.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 7680, |
| "bytes": 2560 |
| } |
| }, |
| "embeddings.patch_embeddings.weight": { |
| "shape": [ |
| 1280, |
| 3, |
| 16, |
| 16 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 10240, |
| "bytes": 1966080 |
| } |
| }, |
| "embeddings.register_tokens": { |
| "shape": [ |
| 1, |
| 4, |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 1976320, |
| "bytes": 20480 |
| } |
| }, |
| "layer.0.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 1996800, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 3635200, |
| "bytes": 51200 |
| } |
| }, |
| "layer.0.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 3686400, |
| "bytes": 2560 |
| } |
| }, |
| "layer.0.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 3688960, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 5327360, |
| "bytes": 51200 |
| } |
| }, |
| "layer.0.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 5378560, |
| "bytes": 2560 |
| } |
| }, |
| "layer.0.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 5381120, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 7019520, |
| "bytes": 51200 |
| } |
| }, |
| "layer.0.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 7070720, |
| "bytes": 2560 |
| } |
| }, |
| "layer.0.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 7073280, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 8711680, |
| "bytes": 51200 |
| } |
| }, |
| "layer.0.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 8762880, |
| "bytes": 5120 |
| } |
| }, |
| "layer.0.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 8768000, |
| "bytes": 5120 |
| } |
| }, |
| "layer.0.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 8773120, |
| "bytes": 2560 |
| } |
| }, |
| "layer.0.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 0, |
| "offset": 8775680, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 15329280, |
| "bytes": 204800 |
| } |
| }, |
| "layer.0.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 15534080, |
| "bytes": 10240 |
| } |
| }, |
| "layer.0.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 15544320, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 22097920, |
| "bytes": 204800 |
| } |
| }, |
| "layer.0.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 22302720, |
| "bytes": 10240 |
| } |
| }, |
| "layer.0.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 22312960, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 28866560, |
| "bytes": 204800 |
| } |
| }, |
| "layer.0.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 29071360, |
| "bytes": 5120 |
| } |
| }, |
| "layer.0.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 29076480, |
| "bytes": 5120 |
| } |
| }, |
| "layer.0.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 29081600, |
| "bytes": 5120 |
| } |
| }, |
| "layer.0.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 29086720, |
| "bytes": 5120 |
| } |
| }, |
| "layer.1.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 29091840, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 30730240, |
| "bytes": 51200 |
| } |
| }, |
| "layer.1.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 30781440, |
| "bytes": 2560 |
| } |
| }, |
| "layer.1.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 30784000, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 32422400, |
| "bytes": 51200 |
| } |
| }, |
| "layer.1.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 32473600, |
| "bytes": 2560 |
| } |
| }, |
| "layer.1.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 32476160, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 34114560, |
| "bytes": 51200 |
| } |
| }, |
| "layer.1.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 34165760, |
| "bytes": 2560 |
| } |
| }, |
| "layer.1.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 34168320, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 35806720, |
| "bytes": 51200 |
| } |
| }, |
| "layer.1.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 35857920, |
| "bytes": 5120 |
| } |
| }, |
| "layer.1.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 35863040, |
| "bytes": 5120 |
| } |
| }, |
| "layer.1.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 35868160, |
| "bytes": 2560 |
| } |
| }, |
| "layer.1.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 0, |
| "offset": 35870720, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 42424320, |
| "bytes": 204800 |
| } |
| }, |
| "layer.1.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 42629120, |
| "bytes": 10240 |
| } |
| }, |
| "layer.1.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 42639360, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 49192960, |
| "bytes": 204800 |
| } |
| }, |
| "layer.1.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 49397760, |
| "bytes": 10240 |
| } |
| }, |
| "layer.1.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 49408000, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 55961600, |
| "bytes": 204800 |
| } |
| }, |
| "layer.1.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 56166400, |
| "bytes": 5120 |
| } |
| }, |
| "layer.1.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 56171520, |
| "bytes": 5120 |
| } |
| }, |
| "layer.1.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 56176640, |
| "bytes": 5120 |
| } |
| }, |
| "layer.1.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 56181760, |
| "bytes": 5120 |
| } |
| }, |
| "layer.10.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 56186880, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 57825280, |
| "bytes": 51200 |
| } |
| }, |
| "layer.10.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 57876480, |
| "bytes": 2560 |
| } |
| }, |
| "layer.10.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 57879040, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 59517440, |
| "bytes": 51200 |
| } |
| }, |
| "layer.10.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 59568640, |
| "bytes": 2560 |
| } |
| }, |
| "layer.10.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 59571200, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 61209600, |
| "bytes": 51200 |
| } |
| }, |
| "layer.10.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 61260800, |
| "bytes": 2560 |
| } |
| }, |
| "layer.10.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 0, |
| "offset": 61263360, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 0, |
| "offset": 62901760, |
| "bytes": 51200 |
| } |
| }, |
| "layer.10.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 62952960, |
| "bytes": 5120 |
| } |
| }, |
| "layer.10.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 0, |
| "offset": 62958080, |
| "bytes": 5120 |
| } |
| }, |
| "layer.10.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 0, |
| "offset": 62963200, |
| "bytes": 2560 |
| } |
| }, |
| "layer.10.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 1, |
| "offset": 0, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 6553600, |
| "bytes": 204800 |
| } |
| }, |
| "layer.10.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 6758400, |
| "bytes": 10240 |
| } |
| }, |
| "layer.10.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 1, |
| "offset": 6768640, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 13322240, |
| "bytes": 204800 |
| } |
| }, |
| "layer.10.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 13527040, |
| "bytes": 10240 |
| } |
| }, |
| "layer.10.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 1, |
| "offset": 13537280, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 20090880, |
| "bytes": 204800 |
| } |
| }, |
| "layer.10.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 20295680, |
| "bytes": 5120 |
| } |
| }, |
| "layer.10.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 20300800, |
| "bytes": 5120 |
| } |
| }, |
| "layer.10.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 20305920, |
| "bytes": 5120 |
| } |
| }, |
| "layer.10.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 20311040, |
| "bytes": 5120 |
| } |
| }, |
| "layer.11.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 1, |
| "offset": 20316160, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 21954560, |
| "bytes": 51200 |
| } |
| }, |
| "layer.11.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 22005760, |
| "bytes": 2560 |
| } |
| }, |
| "layer.11.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 1, |
| "offset": 22008320, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 23646720, |
| "bytes": 51200 |
| } |
| }, |
| "layer.11.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 23697920, |
| "bytes": 2560 |
| } |
| }, |
| "layer.11.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 1, |
| "offset": 23700480, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 25338880, |
| "bytes": 51200 |
| } |
| }, |
| "layer.11.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 25390080, |
| "bytes": 2560 |
| } |
| }, |
| "layer.11.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 1, |
| "offset": 25392640, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 27031040, |
| "bytes": 51200 |
| } |
| }, |
| "layer.11.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 27082240, |
| "bytes": 5120 |
| } |
| }, |
| "layer.11.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 27087360, |
| "bytes": 5120 |
| } |
| }, |
| "layer.11.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 27092480, |
| "bytes": 2560 |
| } |
| }, |
| "layer.11.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 1, |
| "offset": 27095040, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 33648640, |
| "bytes": 204800 |
| } |
| }, |
| "layer.11.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 33853440, |
| "bytes": 10240 |
| } |
| }, |
| "layer.11.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 1, |
| "offset": 33863680, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 40417280, |
| "bytes": 204800 |
| } |
| }, |
| "layer.11.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 40622080, |
| "bytes": 10240 |
| } |
| }, |
| "layer.11.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 1, |
| "offset": 40632320, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 47185920, |
| "bytes": 204800 |
| } |
| }, |
| "layer.11.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 47390720, |
| "bytes": 5120 |
| } |
| }, |
| "layer.11.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 47395840, |
| "bytes": 5120 |
| } |
| }, |
| "layer.11.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 47400960, |
| "bytes": 5120 |
| } |
| }, |
| "layer.11.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 47406080, |
| "bytes": 5120 |
| } |
| }, |
| "layer.12.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 1, |
| "offset": 47411200, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 49049600, |
| "bytes": 51200 |
| } |
| }, |
| "layer.12.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 49100800, |
| "bytes": 2560 |
| } |
| }, |
| "layer.12.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 1, |
| "offset": 49103360, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 50741760, |
| "bytes": 51200 |
| } |
| }, |
| "layer.12.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 50792960, |
| "bytes": 2560 |
| } |
| }, |
| "layer.12.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 1, |
| "offset": 50795520, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 52433920, |
| "bytes": 51200 |
| } |
| }, |
| "layer.12.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 52485120, |
| "bytes": 2560 |
| } |
| }, |
| "layer.12.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 1, |
| "offset": 52487680, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 54126080, |
| "bytes": 51200 |
| } |
| }, |
| "layer.12.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 54177280, |
| "bytes": 5120 |
| } |
| }, |
| "layer.12.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 1, |
| "offset": 54182400, |
| "bytes": 5120 |
| } |
| }, |
| "layer.12.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 54187520, |
| "bytes": 2560 |
| } |
| }, |
| "layer.12.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 1, |
| "offset": 54190080, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 1, |
| "offset": 60743680, |
| "bytes": 204800 |
| } |
| }, |
| "layer.12.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 1, |
| "offset": 60948480, |
| "bytes": 10240 |
| } |
| }, |
| "layer.12.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 2, |
| "offset": 0, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 6553600, |
| "bytes": 204800 |
| } |
| }, |
| "layer.12.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 6758400, |
| "bytes": 10240 |
| } |
| }, |
| "layer.12.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 2, |
| "offset": 6768640, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 13322240, |
| "bytes": 204800 |
| } |
| }, |
| "layer.12.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 13527040, |
| "bytes": 5120 |
| } |
| }, |
| "layer.12.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 13532160, |
| "bytes": 5120 |
| } |
| }, |
| "layer.12.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 13537280, |
| "bytes": 5120 |
| } |
| }, |
| "layer.12.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 13542400, |
| "bytes": 5120 |
| } |
| }, |
| "layer.13.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 2, |
| "offset": 13547520, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 15185920, |
| "bytes": 51200 |
| } |
| }, |
| "layer.13.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 15237120, |
| "bytes": 2560 |
| } |
| }, |
| "layer.13.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 2, |
| "offset": 15239680, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 16878080, |
| "bytes": 51200 |
| } |
| }, |
| "layer.13.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 16929280, |
| "bytes": 2560 |
| } |
| }, |
| "layer.13.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 2, |
| "offset": 16931840, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 18570240, |
| "bytes": 51200 |
| } |
| }, |
| "layer.13.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 18621440, |
| "bytes": 2560 |
| } |
| }, |
| "layer.13.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 2, |
| "offset": 18624000, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 20262400, |
| "bytes": 51200 |
| } |
| }, |
| "layer.13.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 20313600, |
| "bytes": 5120 |
| } |
| }, |
| "layer.13.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 20318720, |
| "bytes": 5120 |
| } |
| }, |
| "layer.13.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 20323840, |
| "bytes": 2560 |
| } |
| }, |
| "layer.13.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 2, |
| "offset": 20326400, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 26880000, |
| "bytes": 204800 |
| } |
| }, |
| "layer.13.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 27084800, |
| "bytes": 10240 |
| } |
| }, |
| "layer.13.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 2, |
| "offset": 27095040, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 33648640, |
| "bytes": 204800 |
| } |
| }, |
| "layer.13.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 33853440, |
| "bytes": 10240 |
| } |
| }, |
| "layer.13.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 2, |
| "offset": 33863680, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 40417280, |
| "bytes": 204800 |
| } |
| }, |
| "layer.13.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 40622080, |
| "bytes": 5120 |
| } |
| }, |
| "layer.13.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 40627200, |
| "bytes": 5120 |
| } |
| }, |
| "layer.13.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 40632320, |
| "bytes": 5120 |
| } |
| }, |
| "layer.13.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 40637440, |
| "bytes": 5120 |
| } |
| }, |
| "layer.14.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 2, |
| "offset": 40642560, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 42280960, |
| "bytes": 51200 |
| } |
| }, |
| "layer.14.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 42332160, |
| "bytes": 2560 |
| } |
| }, |
| "layer.14.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 2, |
| "offset": 42334720, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 43973120, |
| "bytes": 51200 |
| } |
| }, |
| "layer.14.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 44024320, |
| "bytes": 2560 |
| } |
| }, |
| "layer.14.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 2, |
| "offset": 44026880, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 45665280, |
| "bytes": 51200 |
| } |
| }, |
| "layer.14.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 45716480, |
| "bytes": 2560 |
| } |
| }, |
| "layer.14.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 2, |
| "offset": 45719040, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 47357440, |
| "bytes": 51200 |
| } |
| }, |
| "layer.14.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 47408640, |
| "bytes": 5120 |
| } |
| }, |
| "layer.14.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 2, |
| "offset": 47413760, |
| "bytes": 5120 |
| } |
| }, |
| "layer.14.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 47418880, |
| "bytes": 2560 |
| } |
| }, |
| "layer.14.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 2, |
| "offset": 47421440, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 53975040, |
| "bytes": 204800 |
| } |
| }, |
| "layer.14.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 54179840, |
| "bytes": 10240 |
| } |
| }, |
| "layer.14.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 2, |
| "offset": 54190080, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 2, |
| "offset": 60743680, |
| "bytes": 204800 |
| } |
| }, |
| "layer.14.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 2, |
| "offset": 60948480, |
| "bytes": 10240 |
| } |
| }, |
| "layer.14.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 0, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 6553600, |
| "bytes": 204800 |
| } |
| }, |
| "layer.14.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 6758400, |
| "bytes": 5120 |
| } |
| }, |
| "layer.14.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 6763520, |
| "bytes": 5120 |
| } |
| }, |
| "layer.14.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 6768640, |
| "bytes": 5120 |
| } |
| }, |
| "layer.14.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 6773760, |
| "bytes": 5120 |
| } |
| }, |
| "layer.15.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 6778880, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 8417280, |
| "bytes": 51200 |
| } |
| }, |
| "layer.15.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 8468480, |
| "bytes": 2560 |
| } |
| }, |
| "layer.15.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 8471040, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 10109440, |
| "bytes": 51200 |
| } |
| }, |
| "layer.15.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 10160640, |
| "bytes": 2560 |
| } |
| }, |
| "layer.15.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 10163200, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 11801600, |
| "bytes": 51200 |
| } |
| }, |
| "layer.15.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 11852800, |
| "bytes": 2560 |
| } |
| }, |
| "layer.15.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 11855360, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 13493760, |
| "bytes": 51200 |
| } |
| }, |
| "layer.15.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 13544960, |
| "bytes": 5120 |
| } |
| }, |
| "layer.15.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 13550080, |
| "bytes": 5120 |
| } |
| }, |
| "layer.15.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 13555200, |
| "bytes": 2560 |
| } |
| }, |
| "layer.15.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 3, |
| "offset": 13557760, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 20111360, |
| "bytes": 204800 |
| } |
| }, |
| "layer.15.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 20316160, |
| "bytes": 10240 |
| } |
| }, |
| "layer.15.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 20326400, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 26880000, |
| "bytes": 204800 |
| } |
| }, |
| "layer.15.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 27084800, |
| "bytes": 10240 |
| } |
| }, |
| "layer.15.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 27095040, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 33648640, |
| "bytes": 204800 |
| } |
| }, |
| "layer.15.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 33853440, |
| "bytes": 5120 |
| } |
| }, |
| "layer.15.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 33858560, |
| "bytes": 5120 |
| } |
| }, |
| "layer.15.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 33863680, |
| "bytes": 5120 |
| } |
| }, |
| "layer.15.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 33868800, |
| "bytes": 5120 |
| } |
| }, |
| "layer.16.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 33873920, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 35512320, |
| "bytes": 51200 |
| } |
| }, |
| "layer.16.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 35563520, |
| "bytes": 2560 |
| } |
| }, |
| "layer.16.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 35566080, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 37204480, |
| "bytes": 51200 |
| } |
| }, |
| "layer.16.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 37255680, |
| "bytes": 2560 |
| } |
| }, |
| "layer.16.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 37258240, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 38896640, |
| "bytes": 51200 |
| } |
| }, |
| "layer.16.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 38947840, |
| "bytes": 2560 |
| } |
| }, |
| "layer.16.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 38950400, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 40588800, |
| "bytes": 51200 |
| } |
| }, |
| "layer.16.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 40640000, |
| "bytes": 5120 |
| } |
| }, |
| "layer.16.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 40645120, |
| "bytes": 5120 |
| } |
| }, |
| "layer.16.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 40650240, |
| "bytes": 2560 |
| } |
| }, |
| "layer.16.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 3, |
| "offset": 40652800, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 47206400, |
| "bytes": 204800 |
| } |
| }, |
| "layer.16.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 47411200, |
| "bytes": 10240 |
| } |
| }, |
| "layer.16.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 47421440, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 53975040, |
| "bytes": 204800 |
| } |
| }, |
| "layer.16.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 54179840, |
| "bytes": 10240 |
| } |
| }, |
| "layer.16.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 54190080, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 60743680, |
| "bytes": 204800 |
| } |
| }, |
| "layer.16.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 60948480, |
| "bytes": 5120 |
| } |
| }, |
| "layer.16.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 60953600, |
| "bytes": 5120 |
| } |
| }, |
| "layer.16.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 60958720, |
| "bytes": 5120 |
| } |
| }, |
| "layer.16.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 3, |
| "offset": 60963840, |
| "bytes": 5120 |
| } |
| }, |
| "layer.17.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 60968960, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 62607360, |
| "bytes": 51200 |
| } |
| }, |
| "layer.17.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 62658560, |
| "bytes": 2560 |
| } |
| }, |
| "layer.17.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 62661120, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 64299520, |
| "bytes": 51200 |
| } |
| }, |
| "layer.17.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 64350720, |
| "bytes": 2560 |
| } |
| }, |
| "layer.17.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 3, |
| "offset": 64353280, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 3, |
| "offset": 65991680, |
| "bytes": 51200 |
| } |
| }, |
| "layer.17.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 3, |
| "offset": 66042880, |
| "bytes": 2560 |
| } |
| }, |
| "layer.17.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 4, |
| "offset": 0, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 1638400, |
| "bytes": 51200 |
| } |
| }, |
| "layer.17.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 4, |
| "offset": 1689600, |
| "bytes": 5120 |
| } |
| }, |
| "layer.17.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 4, |
| "offset": 1694720, |
| "bytes": 5120 |
| } |
| }, |
| "layer.17.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 1699840, |
| "bytes": 2560 |
| } |
| }, |
| "layer.17.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 4, |
| "offset": 1702400, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 8256000, |
| "bytes": 204800 |
| } |
| }, |
| "layer.17.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 8460800, |
| "bytes": 10240 |
| } |
| }, |
| "layer.17.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 4, |
| "offset": 8471040, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 15024640, |
| "bytes": 204800 |
| } |
| }, |
| "layer.17.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 15229440, |
| "bytes": 10240 |
| } |
| }, |
| "layer.17.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 4, |
| "offset": 15239680, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 21793280, |
| "bytes": 204800 |
| } |
| }, |
| "layer.17.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 4, |
| "offset": 21998080, |
| "bytes": 5120 |
| } |
| }, |
| "layer.17.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 4, |
| "offset": 22003200, |
| "bytes": 5120 |
| } |
| }, |
| "layer.17.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 4, |
| "offset": 22008320, |
| "bytes": 5120 |
| } |
| }, |
| "layer.17.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 4, |
| "offset": 22013440, |
| "bytes": 5120 |
| } |
| }, |
| "layer.18.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 4, |
| "offset": 22018560, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 23656960, |
| "bytes": 51200 |
| } |
| }, |
| "layer.18.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 23708160, |
| "bytes": 2560 |
| } |
| }, |
| "layer.18.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 4, |
| "offset": 23710720, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 25349120, |
| "bytes": 51200 |
| } |
| }, |
| "layer.18.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 25400320, |
| "bytes": 2560 |
| } |
| }, |
| "layer.18.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 4, |
| "offset": 25402880, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 27041280, |
| "bytes": 51200 |
| } |
| }, |
| "layer.18.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 27092480, |
| "bytes": 2560 |
| } |
| }, |
| "layer.18.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 4, |
| "offset": 27095040, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 28733440, |
| "bytes": 51200 |
| } |
| }, |
| "layer.18.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 4, |
| "offset": 28784640, |
| "bytes": 5120 |
| } |
| }, |
| "layer.18.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 4, |
| "offset": 28789760, |
| "bytes": 5120 |
| } |
| }, |
| "layer.18.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 28794880, |
| "bytes": 2560 |
| } |
| }, |
| "layer.18.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 4, |
| "offset": 28797440, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 35351040, |
| "bytes": 204800 |
| } |
| }, |
| "layer.18.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 35555840, |
| "bytes": 10240 |
| } |
| }, |
| "layer.18.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 4, |
| "offset": 35566080, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 42119680, |
| "bytes": 204800 |
| } |
| }, |
| "layer.18.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 42324480, |
| "bytes": 10240 |
| } |
| }, |
| "layer.18.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 4, |
| "offset": 42334720, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 48888320, |
| "bytes": 204800 |
| } |
| }, |
| "layer.18.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 4, |
| "offset": 49093120, |
| "bytes": 5120 |
| } |
| }, |
| "layer.18.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 4, |
| "offset": 49098240, |
| "bytes": 5120 |
| } |
| }, |
| "layer.18.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 4, |
| "offset": 49103360, |
| "bytes": 5120 |
| } |
| }, |
| "layer.18.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 4, |
| "offset": 49108480, |
| "bytes": 5120 |
| } |
| }, |
| "layer.19.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 4, |
| "offset": 49113600, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 50752000, |
| "bytes": 51200 |
| } |
| }, |
| "layer.19.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 50803200, |
| "bytes": 2560 |
| } |
| }, |
| "layer.19.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 4, |
| "offset": 50805760, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 52444160, |
| "bytes": 51200 |
| } |
| }, |
| "layer.19.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 52495360, |
| "bytes": 2560 |
| } |
| }, |
| "layer.19.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 4, |
| "offset": 52497920, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 54136320, |
| "bytes": 51200 |
| } |
| }, |
| "layer.19.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 54187520, |
| "bytes": 2560 |
| } |
| }, |
| "layer.19.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 4, |
| "offset": 54190080, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 55828480, |
| "bytes": 51200 |
| } |
| }, |
| "layer.19.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 4, |
| "offset": 55879680, |
| "bytes": 5120 |
| } |
| }, |
| "layer.19.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 4, |
| "offset": 55884800, |
| "bytes": 5120 |
| } |
| }, |
| "layer.19.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 55889920, |
| "bytes": 2560 |
| } |
| }, |
| "layer.19.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 4, |
| "offset": 55892480, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 4, |
| "offset": 62446080, |
| "bytes": 204800 |
| } |
| }, |
| "layer.19.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 4, |
| "offset": 62650880, |
| "bytes": 10240 |
| } |
| }, |
| "layer.19.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 5, |
| "offset": 0, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 6553600, |
| "bytes": 204800 |
| } |
| }, |
| "layer.19.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 6758400, |
| "bytes": 10240 |
| } |
| }, |
| "layer.19.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 5, |
| "offset": 6768640, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 13322240, |
| "bytes": 204800 |
| } |
| }, |
| "layer.19.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 5, |
| "offset": 13527040, |
| "bytes": 5120 |
| } |
| }, |
| "layer.19.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 5, |
| "offset": 13532160, |
| "bytes": 5120 |
| } |
| }, |
| "layer.19.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 5, |
| "offset": 13537280, |
| "bytes": 5120 |
| } |
| }, |
| "layer.19.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 5, |
| "offset": 13542400, |
| "bytes": 5120 |
| } |
| }, |
| "layer.2.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 5, |
| "offset": 13547520, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 15185920, |
| "bytes": 51200 |
| } |
| }, |
| "layer.2.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 15237120, |
| "bytes": 2560 |
| } |
| }, |
| "layer.2.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 5, |
| "offset": 15239680, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 16878080, |
| "bytes": 51200 |
| } |
| }, |
| "layer.2.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 16929280, |
| "bytes": 2560 |
| } |
| }, |
| "layer.2.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 5, |
| "offset": 16931840, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 18570240, |
| "bytes": 51200 |
| } |
| }, |
| "layer.2.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 18621440, |
| "bytes": 2560 |
| } |
| }, |
| "layer.2.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 5, |
| "offset": 18624000, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 20262400, |
| "bytes": 51200 |
| } |
| }, |
| "layer.2.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 5, |
| "offset": 20313600, |
| "bytes": 5120 |
| } |
| }, |
| "layer.2.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 5, |
| "offset": 20318720, |
| "bytes": 5120 |
| } |
| }, |
| "layer.2.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 20323840, |
| "bytes": 2560 |
| } |
| }, |
| "layer.2.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 5, |
| "offset": 20326400, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 26880000, |
| "bytes": 204800 |
| } |
| }, |
| "layer.2.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 27084800, |
| "bytes": 10240 |
| } |
| }, |
| "layer.2.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 5, |
| "offset": 27095040, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 33648640, |
| "bytes": 204800 |
| } |
| }, |
| "layer.2.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 33853440, |
| "bytes": 10240 |
| } |
| }, |
| "layer.2.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 5, |
| "offset": 33863680, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 40417280, |
| "bytes": 204800 |
| } |
| }, |
| "layer.2.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 5, |
| "offset": 40622080, |
| "bytes": 5120 |
| } |
| }, |
| "layer.2.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 5, |
| "offset": 40627200, |
| "bytes": 5120 |
| } |
| }, |
| "layer.2.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 5, |
| "offset": 40632320, |
| "bytes": 5120 |
| } |
| }, |
| "layer.2.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 5, |
| "offset": 40637440, |
| "bytes": 5120 |
| } |
| }, |
| "layer.20.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 5, |
| "offset": 40642560, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 42280960, |
| "bytes": 51200 |
| } |
| }, |
| "layer.20.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 42332160, |
| "bytes": 2560 |
| } |
| }, |
| "layer.20.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 5, |
| "offset": 42334720, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 43973120, |
| "bytes": 51200 |
| } |
| }, |
| "layer.20.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 44024320, |
| "bytes": 2560 |
| } |
| }, |
| "layer.20.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 5, |
| "offset": 44026880, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 45665280, |
| "bytes": 51200 |
| } |
| }, |
| "layer.20.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 45716480, |
| "bytes": 2560 |
| } |
| }, |
| "layer.20.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 5, |
| "offset": 45719040, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 47357440, |
| "bytes": 51200 |
| } |
| }, |
| "layer.20.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 5, |
| "offset": 47408640, |
| "bytes": 5120 |
| } |
| }, |
| "layer.20.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 5, |
| "offset": 47413760, |
| "bytes": 5120 |
| } |
| }, |
| "layer.20.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 47418880, |
| "bytes": 2560 |
| } |
| }, |
| "layer.20.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 5, |
| "offset": 47421440, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 53975040, |
| "bytes": 204800 |
| } |
| }, |
| "layer.20.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 54179840, |
| "bytes": 10240 |
| } |
| }, |
| "layer.20.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 5, |
| "offset": 54190080, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 5, |
| "offset": 60743680, |
| "bytes": 204800 |
| } |
| }, |
| "layer.20.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 5, |
| "offset": 60948480, |
| "bytes": 10240 |
| } |
| }, |
| "layer.20.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 0, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 6553600, |
| "bytes": 204800 |
| } |
| }, |
| "layer.20.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 6758400, |
| "bytes": 5120 |
| } |
| }, |
| "layer.20.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 6763520, |
| "bytes": 5120 |
| } |
| }, |
| "layer.20.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 6768640, |
| "bytes": 5120 |
| } |
| }, |
| "layer.20.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 6773760, |
| "bytes": 5120 |
| } |
| }, |
| "layer.21.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 6778880, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 8417280, |
| "bytes": 51200 |
| } |
| }, |
| "layer.21.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 8468480, |
| "bytes": 2560 |
| } |
| }, |
| "layer.21.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 8471040, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 10109440, |
| "bytes": 51200 |
| } |
| }, |
| "layer.21.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 10160640, |
| "bytes": 2560 |
| } |
| }, |
| "layer.21.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 10163200, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 11801600, |
| "bytes": 51200 |
| } |
| }, |
| "layer.21.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 11852800, |
| "bytes": 2560 |
| } |
| }, |
| "layer.21.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 11855360, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 13493760, |
| "bytes": 51200 |
| } |
| }, |
| "layer.21.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 13544960, |
| "bytes": 5120 |
| } |
| }, |
| "layer.21.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 13550080, |
| "bytes": 5120 |
| } |
| }, |
| "layer.21.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 13555200, |
| "bytes": 2560 |
| } |
| }, |
| "layer.21.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 6, |
| "offset": 13557760, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 20111360, |
| "bytes": 204800 |
| } |
| }, |
| "layer.21.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 20316160, |
| "bytes": 10240 |
| } |
| }, |
| "layer.21.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 20326400, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 26880000, |
| "bytes": 204800 |
| } |
| }, |
| "layer.21.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 27084800, |
| "bytes": 10240 |
| } |
| }, |
| "layer.21.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 27095040, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 33648640, |
| "bytes": 204800 |
| } |
| }, |
| "layer.21.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 33853440, |
| "bytes": 5120 |
| } |
| }, |
| "layer.21.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 33858560, |
| "bytes": 5120 |
| } |
| }, |
| "layer.21.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 33863680, |
| "bytes": 5120 |
| } |
| }, |
| "layer.21.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 33868800, |
| "bytes": 5120 |
| } |
| }, |
| "layer.22.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 33873920, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 35512320, |
| "bytes": 51200 |
| } |
| }, |
| "layer.22.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 35563520, |
| "bytes": 2560 |
| } |
| }, |
| "layer.22.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 35566080, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 37204480, |
| "bytes": 51200 |
| } |
| }, |
| "layer.22.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 37255680, |
| "bytes": 2560 |
| } |
| }, |
| "layer.22.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 37258240, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 38896640, |
| "bytes": 51200 |
| } |
| }, |
| "layer.22.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 38947840, |
| "bytes": 2560 |
| } |
| }, |
| "layer.22.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 38950400, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 40588800, |
| "bytes": 51200 |
| } |
| }, |
| "layer.22.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 40640000, |
| "bytes": 5120 |
| } |
| }, |
| "layer.22.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 40645120, |
| "bytes": 5120 |
| } |
| }, |
| "layer.22.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 40650240, |
| "bytes": 2560 |
| } |
| }, |
| "layer.22.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 6, |
| "offset": 40652800, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 47206400, |
| "bytes": 204800 |
| } |
| }, |
| "layer.22.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 47411200, |
| "bytes": 10240 |
| } |
| }, |
| "layer.22.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 47421440, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 53975040, |
| "bytes": 204800 |
| } |
| }, |
| "layer.22.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 54179840, |
| "bytes": 10240 |
| } |
| }, |
| "layer.22.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 54190080, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 60743680, |
| "bytes": 204800 |
| } |
| }, |
| "layer.22.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 60948480, |
| "bytes": 5120 |
| } |
| }, |
| "layer.22.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 60953600, |
| "bytes": 5120 |
| } |
| }, |
| "layer.22.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 60958720, |
| "bytes": 5120 |
| } |
| }, |
| "layer.22.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 6, |
| "offset": 60963840, |
| "bytes": 5120 |
| } |
| }, |
| "layer.23.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 60968960, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 62607360, |
| "bytes": 51200 |
| } |
| }, |
| "layer.23.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 62658560, |
| "bytes": 2560 |
| } |
| }, |
| "layer.23.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 62661120, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 64299520, |
| "bytes": 51200 |
| } |
| }, |
| "layer.23.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 64350720, |
| "bytes": 2560 |
| } |
| }, |
| "layer.23.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 6, |
| "offset": 64353280, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 6, |
| "offset": 65991680, |
| "bytes": 51200 |
| } |
| }, |
| "layer.23.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 6, |
| "offset": 66042880, |
| "bytes": 2560 |
| } |
| }, |
| "layer.23.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 7, |
| "offset": 0, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 1638400, |
| "bytes": 51200 |
| } |
| }, |
| "layer.23.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 7, |
| "offset": 1689600, |
| "bytes": 5120 |
| } |
| }, |
| "layer.23.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 7, |
| "offset": 1694720, |
| "bytes": 5120 |
| } |
| }, |
| "layer.23.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 1699840, |
| "bytes": 2560 |
| } |
| }, |
| "layer.23.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 7, |
| "offset": 1702400, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 8256000, |
| "bytes": 204800 |
| } |
| }, |
| "layer.23.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 8460800, |
| "bytes": 10240 |
| } |
| }, |
| "layer.23.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 7, |
| "offset": 8471040, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 15024640, |
| "bytes": 204800 |
| } |
| }, |
| "layer.23.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 15229440, |
| "bytes": 10240 |
| } |
| }, |
| "layer.23.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 7, |
| "offset": 15239680, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 21793280, |
| "bytes": 204800 |
| } |
| }, |
| "layer.23.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 7, |
| "offset": 21998080, |
| "bytes": 5120 |
| } |
| }, |
| "layer.23.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 7, |
| "offset": 22003200, |
| "bytes": 5120 |
| } |
| }, |
| "layer.23.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 7, |
| "offset": 22008320, |
| "bytes": 5120 |
| } |
| }, |
| "layer.23.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 7, |
| "offset": 22013440, |
| "bytes": 5120 |
| } |
| }, |
| "layer.24.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 7, |
| "offset": 22018560, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 23656960, |
| "bytes": 51200 |
| } |
| }, |
| "layer.24.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 23708160, |
| "bytes": 2560 |
| } |
| }, |
| "layer.24.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 7, |
| "offset": 23710720, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 25349120, |
| "bytes": 51200 |
| } |
| }, |
| "layer.24.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 25400320, |
| "bytes": 2560 |
| } |
| }, |
| "layer.24.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 7, |
| "offset": 25402880, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 27041280, |
| "bytes": 51200 |
| } |
| }, |
| "layer.24.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 27092480, |
| "bytes": 2560 |
| } |
| }, |
| "layer.24.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 7, |
| "offset": 27095040, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 28733440, |
| "bytes": 51200 |
| } |
| }, |
| "layer.24.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 7, |
| "offset": 28784640, |
| "bytes": 5120 |
| } |
| }, |
| "layer.24.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 7, |
| "offset": 28789760, |
| "bytes": 5120 |
| } |
| }, |
| "layer.24.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 28794880, |
| "bytes": 2560 |
| } |
| }, |
| "layer.24.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 7, |
| "offset": 28797440, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 35351040, |
| "bytes": 204800 |
| } |
| }, |
| "layer.24.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 35555840, |
| "bytes": 10240 |
| } |
| }, |
| "layer.24.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 7, |
| "offset": 35566080, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 42119680, |
| "bytes": 204800 |
| } |
| }, |
| "layer.24.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 42324480, |
| "bytes": 10240 |
| } |
| }, |
| "layer.24.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 7, |
| "offset": 42334720, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 48888320, |
| "bytes": 204800 |
| } |
| }, |
| "layer.24.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 7, |
| "offset": 49093120, |
| "bytes": 5120 |
| } |
| }, |
| "layer.24.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 7, |
| "offset": 49098240, |
| "bytes": 5120 |
| } |
| }, |
| "layer.24.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 7, |
| "offset": 49103360, |
| "bytes": 5120 |
| } |
| }, |
| "layer.24.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 7, |
| "offset": 49108480, |
| "bytes": 5120 |
| } |
| }, |
| "layer.25.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 7, |
| "offset": 49113600, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 50752000, |
| "bytes": 51200 |
| } |
| }, |
| "layer.25.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 50803200, |
| "bytes": 2560 |
| } |
| }, |
| "layer.25.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 7, |
| "offset": 50805760, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 52444160, |
| "bytes": 51200 |
| } |
| }, |
| "layer.25.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 52495360, |
| "bytes": 2560 |
| } |
| }, |
| "layer.25.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 7, |
| "offset": 52497920, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 54136320, |
| "bytes": 51200 |
| } |
| }, |
| "layer.25.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 54187520, |
| "bytes": 2560 |
| } |
| }, |
| "layer.25.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 7, |
| "offset": 54190080, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 55828480, |
| "bytes": 51200 |
| } |
| }, |
| "layer.25.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 7, |
| "offset": 55879680, |
| "bytes": 5120 |
| } |
| }, |
| "layer.25.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 7, |
| "offset": 55884800, |
| "bytes": 5120 |
| } |
| }, |
| "layer.25.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 55889920, |
| "bytes": 2560 |
| } |
| }, |
| "layer.25.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 7, |
| "offset": 55892480, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 7, |
| "offset": 62446080, |
| "bytes": 204800 |
| } |
| }, |
| "layer.25.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 7, |
| "offset": 62650880, |
| "bytes": 10240 |
| } |
| }, |
| "layer.25.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 8, |
| "offset": 0, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 6553600, |
| "bytes": 204800 |
| } |
| }, |
| "layer.25.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 6758400, |
| "bytes": 10240 |
| } |
| }, |
| "layer.25.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 8, |
| "offset": 6768640, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 13322240, |
| "bytes": 204800 |
| } |
| }, |
| "layer.25.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 8, |
| "offset": 13527040, |
| "bytes": 5120 |
| } |
| }, |
| "layer.25.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 8, |
| "offset": 13532160, |
| "bytes": 5120 |
| } |
| }, |
| "layer.25.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 8, |
| "offset": 13537280, |
| "bytes": 5120 |
| } |
| }, |
| "layer.25.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 8, |
| "offset": 13542400, |
| "bytes": 5120 |
| } |
| }, |
| "layer.26.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 8, |
| "offset": 13547520, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 15185920, |
| "bytes": 51200 |
| } |
| }, |
| "layer.26.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 15237120, |
| "bytes": 2560 |
| } |
| }, |
| "layer.26.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 8, |
| "offset": 15239680, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 16878080, |
| "bytes": 51200 |
| } |
| }, |
| "layer.26.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 16929280, |
| "bytes": 2560 |
| } |
| }, |
| "layer.26.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 8, |
| "offset": 16931840, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 18570240, |
| "bytes": 51200 |
| } |
| }, |
| "layer.26.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 18621440, |
| "bytes": 2560 |
| } |
| }, |
| "layer.26.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 8, |
| "offset": 18624000, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 20262400, |
| "bytes": 51200 |
| } |
| }, |
| "layer.26.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 8, |
| "offset": 20313600, |
| "bytes": 5120 |
| } |
| }, |
| "layer.26.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 8, |
| "offset": 20318720, |
| "bytes": 5120 |
| } |
| }, |
| "layer.26.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 20323840, |
| "bytes": 2560 |
| } |
| }, |
| "layer.26.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 8, |
| "offset": 20326400, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 26880000, |
| "bytes": 204800 |
| } |
| }, |
| "layer.26.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 27084800, |
| "bytes": 10240 |
| } |
| }, |
| "layer.26.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 8, |
| "offset": 27095040, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 33648640, |
| "bytes": 204800 |
| } |
| }, |
| "layer.26.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 33853440, |
| "bytes": 10240 |
| } |
| }, |
| "layer.26.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 8, |
| "offset": 33863680, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 40417280, |
| "bytes": 204800 |
| } |
| }, |
| "layer.26.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 8, |
| "offset": 40622080, |
| "bytes": 5120 |
| } |
| }, |
| "layer.26.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 8, |
| "offset": 40627200, |
| "bytes": 5120 |
| } |
| }, |
| "layer.26.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 8, |
| "offset": 40632320, |
| "bytes": 5120 |
| } |
| }, |
| "layer.26.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 8, |
| "offset": 40637440, |
| "bytes": 5120 |
| } |
| }, |
| "layer.27.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 8, |
| "offset": 40642560, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 42280960, |
| "bytes": 51200 |
| } |
| }, |
| "layer.27.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 42332160, |
| "bytes": 2560 |
| } |
| }, |
| "layer.27.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 8, |
| "offset": 42334720, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 43973120, |
| "bytes": 51200 |
| } |
| }, |
| "layer.27.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 44024320, |
| "bytes": 2560 |
| } |
| }, |
| "layer.27.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 8, |
| "offset": 44026880, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 45665280, |
| "bytes": 51200 |
| } |
| }, |
| "layer.27.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 45716480, |
| "bytes": 2560 |
| } |
| }, |
| "layer.27.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 8, |
| "offset": 45719040, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 47357440, |
| "bytes": 51200 |
| } |
| }, |
| "layer.27.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 8, |
| "offset": 47408640, |
| "bytes": 5120 |
| } |
| }, |
| "layer.27.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 8, |
| "offset": 47413760, |
| "bytes": 5120 |
| } |
| }, |
| "layer.27.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 47418880, |
| "bytes": 2560 |
| } |
| }, |
| "layer.27.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 8, |
| "offset": 47421440, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 53975040, |
| "bytes": 204800 |
| } |
| }, |
| "layer.27.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 54179840, |
| "bytes": 10240 |
| } |
| }, |
| "layer.27.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 8, |
| "offset": 54190080, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 8, |
| "offset": 60743680, |
| "bytes": 204800 |
| } |
| }, |
| "layer.27.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 8, |
| "offset": 60948480, |
| "bytes": 10240 |
| } |
| }, |
| "layer.27.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 0, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 6553600, |
| "bytes": 204800 |
| } |
| }, |
| "layer.27.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 6758400, |
| "bytes": 5120 |
| } |
| }, |
| "layer.27.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 6763520, |
| "bytes": 5120 |
| } |
| }, |
| "layer.27.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 6768640, |
| "bytes": 5120 |
| } |
| }, |
| "layer.27.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 6773760, |
| "bytes": 5120 |
| } |
| }, |
| "layer.28.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 6778880, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 8417280, |
| "bytes": 51200 |
| } |
| }, |
| "layer.28.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 8468480, |
| "bytes": 2560 |
| } |
| }, |
| "layer.28.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 8471040, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 10109440, |
| "bytes": 51200 |
| } |
| }, |
| "layer.28.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 10160640, |
| "bytes": 2560 |
| } |
| }, |
| "layer.28.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 10163200, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 11801600, |
| "bytes": 51200 |
| } |
| }, |
| "layer.28.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 11852800, |
| "bytes": 2560 |
| } |
| }, |
| "layer.28.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 11855360, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 13493760, |
| "bytes": 51200 |
| } |
| }, |
| "layer.28.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 13544960, |
| "bytes": 5120 |
| } |
| }, |
| "layer.28.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 13550080, |
| "bytes": 5120 |
| } |
| }, |
| "layer.28.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 13555200, |
| "bytes": 2560 |
| } |
| }, |
| "layer.28.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 9, |
| "offset": 13557760, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 20111360, |
| "bytes": 204800 |
| } |
| }, |
| "layer.28.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 20316160, |
| "bytes": 10240 |
| } |
| }, |
| "layer.28.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 20326400, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 26880000, |
| "bytes": 204800 |
| } |
| }, |
| "layer.28.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 27084800, |
| "bytes": 10240 |
| } |
| }, |
| "layer.28.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 27095040, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 33648640, |
| "bytes": 204800 |
| } |
| }, |
| "layer.28.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 33853440, |
| "bytes": 5120 |
| } |
| }, |
| "layer.28.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 33858560, |
| "bytes": 5120 |
| } |
| }, |
| "layer.28.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 33863680, |
| "bytes": 5120 |
| } |
| }, |
| "layer.28.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 33868800, |
| "bytes": 5120 |
| } |
| }, |
| "layer.29.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 33873920, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 35512320, |
| "bytes": 51200 |
| } |
| }, |
| "layer.29.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 35563520, |
| "bytes": 2560 |
| } |
| }, |
| "layer.29.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 35566080, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 37204480, |
| "bytes": 51200 |
| } |
| }, |
| "layer.29.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 37255680, |
| "bytes": 2560 |
| } |
| }, |
| "layer.29.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 37258240, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 38896640, |
| "bytes": 51200 |
| } |
| }, |
| "layer.29.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 38947840, |
| "bytes": 2560 |
| } |
| }, |
| "layer.29.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 38950400, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 40588800, |
| "bytes": 51200 |
| } |
| }, |
| "layer.29.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 40640000, |
| "bytes": 5120 |
| } |
| }, |
| "layer.29.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 40645120, |
| "bytes": 5120 |
| } |
| }, |
| "layer.29.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 40650240, |
| "bytes": 2560 |
| } |
| }, |
| "layer.29.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 9, |
| "offset": 40652800, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 47206400, |
| "bytes": 204800 |
| } |
| }, |
| "layer.29.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 47411200, |
| "bytes": 10240 |
| } |
| }, |
| "layer.29.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 47421440, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 53975040, |
| "bytes": 204800 |
| } |
| }, |
| "layer.29.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 54179840, |
| "bytes": 10240 |
| } |
| }, |
| "layer.29.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 54190080, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 60743680, |
| "bytes": 204800 |
| } |
| }, |
| "layer.29.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 60948480, |
| "bytes": 5120 |
| } |
| }, |
| "layer.29.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 60953600, |
| "bytes": 5120 |
| } |
| }, |
| "layer.29.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 60958720, |
| "bytes": 5120 |
| } |
| }, |
| "layer.29.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 9, |
| "offset": 60963840, |
| "bytes": 5120 |
| } |
| }, |
| "layer.3.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 60968960, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 62607360, |
| "bytes": 51200 |
| } |
| }, |
| "layer.3.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 62658560, |
| "bytes": 2560 |
| } |
| }, |
| "layer.3.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 62661120, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 64299520, |
| "bytes": 51200 |
| } |
| }, |
| "layer.3.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 64350720, |
| "bytes": 2560 |
| } |
| }, |
| "layer.3.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 9, |
| "offset": 64353280, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 9, |
| "offset": 65991680, |
| "bytes": 51200 |
| } |
| }, |
| "layer.3.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 9, |
| "offset": 66042880, |
| "bytes": 2560 |
| } |
| }, |
| "layer.3.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 10, |
| "offset": 0, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 1638400, |
| "bytes": 51200 |
| } |
| }, |
| "layer.3.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 10, |
| "offset": 1689600, |
| "bytes": 5120 |
| } |
| }, |
| "layer.3.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 10, |
| "offset": 1694720, |
| "bytes": 5120 |
| } |
| }, |
| "layer.3.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 1699840, |
| "bytes": 2560 |
| } |
| }, |
| "layer.3.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 10, |
| "offset": 1702400, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 8256000, |
| "bytes": 204800 |
| } |
| }, |
| "layer.3.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 8460800, |
| "bytes": 10240 |
| } |
| }, |
| "layer.3.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 10, |
| "offset": 8471040, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 15024640, |
| "bytes": 204800 |
| } |
| }, |
| "layer.3.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 15229440, |
| "bytes": 10240 |
| } |
| }, |
| "layer.3.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 10, |
| "offset": 15239680, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 21793280, |
| "bytes": 204800 |
| } |
| }, |
| "layer.3.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 10, |
| "offset": 21998080, |
| "bytes": 5120 |
| } |
| }, |
| "layer.3.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 10, |
| "offset": 22003200, |
| "bytes": 5120 |
| } |
| }, |
| "layer.3.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 10, |
| "offset": 22008320, |
| "bytes": 5120 |
| } |
| }, |
| "layer.3.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 10, |
| "offset": 22013440, |
| "bytes": 5120 |
| } |
| }, |
| "layer.30.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 10, |
| "offset": 22018560, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 23656960, |
| "bytes": 51200 |
| } |
| }, |
| "layer.30.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 23708160, |
| "bytes": 2560 |
| } |
| }, |
| "layer.30.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 10, |
| "offset": 23710720, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 25349120, |
| "bytes": 51200 |
| } |
| }, |
| "layer.30.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 25400320, |
| "bytes": 2560 |
| } |
| }, |
| "layer.30.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 10, |
| "offset": 25402880, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 27041280, |
| "bytes": 51200 |
| } |
| }, |
| "layer.30.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 27092480, |
| "bytes": 2560 |
| } |
| }, |
| "layer.30.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 10, |
| "offset": 27095040, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 28733440, |
| "bytes": 51200 |
| } |
| }, |
| "layer.30.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 10, |
| "offset": 28784640, |
| "bytes": 5120 |
| } |
| }, |
| "layer.30.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 10, |
| "offset": 28789760, |
| "bytes": 5120 |
| } |
| }, |
| "layer.30.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 28794880, |
| "bytes": 2560 |
| } |
| }, |
| "layer.30.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 10, |
| "offset": 28797440, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 35351040, |
| "bytes": 204800 |
| } |
| }, |
| "layer.30.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 35555840, |
| "bytes": 10240 |
| } |
| }, |
| "layer.30.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 10, |
| "offset": 35566080, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 42119680, |
| "bytes": 204800 |
| } |
| }, |
| "layer.30.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 42324480, |
| "bytes": 10240 |
| } |
| }, |
| "layer.30.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 10, |
| "offset": 42334720, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 48888320, |
| "bytes": 204800 |
| } |
| }, |
| "layer.30.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 10, |
| "offset": 49093120, |
| "bytes": 5120 |
| } |
| }, |
| "layer.30.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 10, |
| "offset": 49098240, |
| "bytes": 5120 |
| } |
| }, |
| "layer.30.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 10, |
| "offset": 49103360, |
| "bytes": 5120 |
| } |
| }, |
| "layer.30.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 10, |
| "offset": 49108480, |
| "bytes": 5120 |
| } |
| }, |
| "layer.31.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 10, |
| "offset": 49113600, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 50752000, |
| "bytes": 51200 |
| } |
| }, |
| "layer.31.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 50803200, |
| "bytes": 2560 |
| } |
| }, |
| "layer.31.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 10, |
| "offset": 50805760, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 52444160, |
| "bytes": 51200 |
| } |
| }, |
| "layer.31.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 52495360, |
| "bytes": 2560 |
| } |
| }, |
| "layer.31.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 10, |
| "offset": 52497920, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 54136320, |
| "bytes": 51200 |
| } |
| }, |
| "layer.31.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 54187520, |
| "bytes": 2560 |
| } |
| }, |
| "layer.31.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 10, |
| "offset": 54190080, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 55828480, |
| "bytes": 51200 |
| } |
| }, |
| "layer.31.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 10, |
| "offset": 55879680, |
| "bytes": 5120 |
| } |
| }, |
| "layer.31.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 10, |
| "offset": 55884800, |
| "bytes": 5120 |
| } |
| }, |
| "layer.31.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 55889920, |
| "bytes": 2560 |
| } |
| }, |
| "layer.31.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 10, |
| "offset": 55892480, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 10, |
| "offset": 62446080, |
| "bytes": 204800 |
| } |
| }, |
| "layer.31.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 10, |
| "offset": 62650880, |
| "bytes": 10240 |
| } |
| }, |
| "layer.31.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 11, |
| "offset": 0, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 6553600, |
| "bytes": 204800 |
| } |
| }, |
| "layer.31.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 11, |
| "offset": 6758400, |
| "bytes": 10240 |
| } |
| }, |
| "layer.31.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 11, |
| "offset": 6768640, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 13322240, |
| "bytes": 204800 |
| } |
| }, |
| "layer.31.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 11, |
| "offset": 13527040, |
| "bytes": 5120 |
| } |
| }, |
| "layer.31.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 11, |
| "offset": 13532160, |
| "bytes": 5120 |
| } |
| }, |
| "layer.31.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 11, |
| "offset": 13537280, |
| "bytes": 5120 |
| } |
| }, |
| "layer.31.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 11, |
| "offset": 13542400, |
| "bytes": 5120 |
| } |
| }, |
| "layer.4.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 11, |
| "offset": 13547520, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 15185920, |
| "bytes": 51200 |
| } |
| }, |
| "layer.4.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 11, |
| "offset": 15237120, |
| "bytes": 2560 |
| } |
| }, |
| "layer.4.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 11, |
| "offset": 15239680, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 16878080, |
| "bytes": 51200 |
| } |
| }, |
| "layer.4.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 11, |
| "offset": 16929280, |
| "bytes": 2560 |
| } |
| }, |
| "layer.4.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 11, |
| "offset": 16931840, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 18570240, |
| "bytes": 51200 |
| } |
| }, |
| "layer.4.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 11, |
| "offset": 18621440, |
| "bytes": 2560 |
| } |
| }, |
| "layer.4.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 11, |
| "offset": 18624000, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 20262400, |
| "bytes": 51200 |
| } |
| }, |
| "layer.4.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 11, |
| "offset": 20313600, |
| "bytes": 5120 |
| } |
| }, |
| "layer.4.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 11, |
| "offset": 20318720, |
| "bytes": 5120 |
| } |
| }, |
| "layer.4.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 11, |
| "offset": 20323840, |
| "bytes": 2560 |
| } |
| }, |
| "layer.4.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 11, |
| "offset": 20326400, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 26880000, |
| "bytes": 204800 |
| } |
| }, |
| "layer.4.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 11, |
| "offset": 27084800, |
| "bytes": 10240 |
| } |
| }, |
| "layer.4.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 11, |
| "offset": 27095040, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 33648640, |
| "bytes": 204800 |
| } |
| }, |
| "layer.4.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 11, |
| "offset": 33853440, |
| "bytes": 10240 |
| } |
| }, |
| "layer.4.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 11, |
| "offset": 33863680, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 40417280, |
| "bytes": 204800 |
| } |
| }, |
| "layer.4.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 11, |
| "offset": 40622080, |
| "bytes": 5120 |
| } |
| }, |
| "layer.4.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 11, |
| "offset": 40627200, |
| "bytes": 5120 |
| } |
| }, |
| "layer.4.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 11, |
| "offset": 40632320, |
| "bytes": 5120 |
| } |
| }, |
| "layer.4.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 11, |
| "offset": 40637440, |
| "bytes": 5120 |
| } |
| }, |
| "layer.5.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 11, |
| "offset": 40642560, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 42280960, |
| "bytes": 51200 |
| } |
| }, |
| "layer.5.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 11, |
| "offset": 42332160, |
| "bytes": 2560 |
| } |
| }, |
| "layer.5.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 11, |
| "offset": 42334720, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 43973120, |
| "bytes": 51200 |
| } |
| }, |
| "layer.5.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 11, |
| "offset": 44024320, |
| "bytes": 2560 |
| } |
| }, |
| "layer.5.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 11, |
| "offset": 44026880, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 45665280, |
| "bytes": 51200 |
| } |
| }, |
| "layer.5.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 11, |
| "offset": 45716480, |
| "bytes": 2560 |
| } |
| }, |
| "layer.5.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 11, |
| "offset": 45719040, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 47357440, |
| "bytes": 51200 |
| } |
| }, |
| "layer.5.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 11, |
| "offset": 47408640, |
| "bytes": 5120 |
| } |
| }, |
| "layer.5.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 11, |
| "offset": 47413760, |
| "bytes": 5120 |
| } |
| }, |
| "layer.5.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 11, |
| "offset": 47418880, |
| "bytes": 2560 |
| } |
| }, |
| "layer.5.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 11, |
| "offset": 47421440, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 53975040, |
| "bytes": 204800 |
| } |
| }, |
| "layer.5.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 11, |
| "offset": 54179840, |
| "bytes": 10240 |
| } |
| }, |
| "layer.5.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 11, |
| "offset": 54190080, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 11, |
| "offset": 60743680, |
| "bytes": 204800 |
| } |
| }, |
| "layer.5.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 11, |
| "offset": 60948480, |
| "bytes": 10240 |
| } |
| }, |
| "layer.5.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 0, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 6553600, |
| "bytes": 204800 |
| } |
| }, |
| "layer.5.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 6758400, |
| "bytes": 5120 |
| } |
| }, |
| "layer.5.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 6763520, |
| "bytes": 5120 |
| } |
| }, |
| "layer.5.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 6768640, |
| "bytes": 5120 |
| } |
| }, |
| "layer.5.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 6773760, |
| "bytes": 5120 |
| } |
| }, |
| "layer.6.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 6778880, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 8417280, |
| "bytes": 51200 |
| } |
| }, |
| "layer.6.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 8468480, |
| "bytes": 2560 |
| } |
| }, |
| "layer.6.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 8471040, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 10109440, |
| "bytes": 51200 |
| } |
| }, |
| "layer.6.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 10160640, |
| "bytes": 2560 |
| } |
| }, |
| "layer.6.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 10163200, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 11801600, |
| "bytes": 51200 |
| } |
| }, |
| "layer.6.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 11852800, |
| "bytes": 2560 |
| } |
| }, |
| "layer.6.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 11855360, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 13493760, |
| "bytes": 51200 |
| } |
| }, |
| "layer.6.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 13544960, |
| "bytes": 5120 |
| } |
| }, |
| "layer.6.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 13550080, |
| "bytes": 5120 |
| } |
| }, |
| "layer.6.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 13555200, |
| "bytes": 2560 |
| } |
| }, |
| "layer.6.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 12, |
| "offset": 13557760, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 20111360, |
| "bytes": 204800 |
| } |
| }, |
| "layer.6.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 20316160, |
| "bytes": 10240 |
| } |
| }, |
| "layer.6.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 20326400, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 26880000, |
| "bytes": 204800 |
| } |
| }, |
| "layer.6.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 27084800, |
| "bytes": 10240 |
| } |
| }, |
| "layer.6.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 27095040, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 33648640, |
| "bytes": 204800 |
| } |
| }, |
| "layer.6.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 33853440, |
| "bytes": 5120 |
| } |
| }, |
| "layer.6.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 33858560, |
| "bytes": 5120 |
| } |
| }, |
| "layer.6.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 33863680, |
| "bytes": 5120 |
| } |
| }, |
| "layer.6.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 33868800, |
| "bytes": 5120 |
| } |
| }, |
| "layer.7.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 33873920, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 35512320, |
| "bytes": 51200 |
| } |
| }, |
| "layer.7.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 35563520, |
| "bytes": 2560 |
| } |
| }, |
| "layer.7.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 35566080, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 37204480, |
| "bytes": 51200 |
| } |
| }, |
| "layer.7.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 37255680, |
| "bytes": 2560 |
| } |
| }, |
| "layer.7.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 37258240, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 38896640, |
| "bytes": 51200 |
| } |
| }, |
| "layer.7.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 38947840, |
| "bytes": 2560 |
| } |
| }, |
| "layer.7.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 38950400, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 40588800, |
| "bytes": 51200 |
| } |
| }, |
| "layer.7.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 40640000, |
| "bytes": 5120 |
| } |
| }, |
| "layer.7.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 40645120, |
| "bytes": 5120 |
| } |
| }, |
| "layer.7.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 40650240, |
| "bytes": 2560 |
| } |
| }, |
| "layer.7.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 12, |
| "offset": 40652800, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 47206400, |
| "bytes": 204800 |
| } |
| }, |
| "layer.7.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 47411200, |
| "bytes": 10240 |
| } |
| }, |
| "layer.7.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 47421440, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 53975040, |
| "bytes": 204800 |
| } |
| }, |
| "layer.7.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 54179840, |
| "bytes": 10240 |
| } |
| }, |
| "layer.7.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 54190080, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 60743680, |
| "bytes": 204800 |
| } |
| }, |
| "layer.7.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 60948480, |
| "bytes": 5120 |
| } |
| }, |
| "layer.7.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 60953600, |
| "bytes": 5120 |
| } |
| }, |
| "layer.7.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 60958720, |
| "bytes": 5120 |
| } |
| }, |
| "layer.7.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 12, |
| "offset": 60963840, |
| "bytes": 5120 |
| } |
| }, |
| "layer.8.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 60968960, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 62607360, |
| "bytes": 51200 |
| } |
| }, |
| "layer.8.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 62658560, |
| "bytes": 2560 |
| } |
| }, |
| "layer.8.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 62661120, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 64299520, |
| "bytes": 51200 |
| } |
| }, |
| "layer.8.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 64350720, |
| "bytes": 2560 |
| } |
| }, |
| "layer.8.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 12, |
| "offset": 64353280, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 12, |
| "offset": 65991680, |
| "bytes": 51200 |
| } |
| }, |
| "layer.8.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 12, |
| "offset": 66042880, |
| "bytes": 2560 |
| } |
| }, |
| "layer.8.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 13, |
| "offset": 0, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 13, |
| "offset": 1638400, |
| "bytes": 51200 |
| } |
| }, |
| "layer.8.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 13, |
| "offset": 1689600, |
| "bytes": 5120 |
| } |
| }, |
| "layer.8.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 13, |
| "offset": 1694720, |
| "bytes": 5120 |
| } |
| }, |
| "layer.8.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 13, |
| "offset": 1699840, |
| "bytes": 2560 |
| } |
| }, |
| "layer.8.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 13, |
| "offset": 1702400, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 13, |
| "offset": 8256000, |
| "bytes": 204800 |
| } |
| }, |
| "layer.8.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 13, |
| "offset": 8460800, |
| "bytes": 10240 |
| } |
| }, |
| "layer.8.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 13, |
| "offset": 8471040, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 13, |
| "offset": 15024640, |
| "bytes": 204800 |
| } |
| }, |
| "layer.8.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 13, |
| "offset": 15229440, |
| "bytes": 10240 |
| } |
| }, |
| "layer.8.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 13, |
| "offset": 15239680, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 13, |
| "offset": 21793280, |
| "bytes": 204800 |
| } |
| }, |
| "layer.8.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 13, |
| "offset": 21998080, |
| "bytes": 5120 |
| } |
| }, |
| "layer.8.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 13, |
| "offset": 22003200, |
| "bytes": 5120 |
| } |
| }, |
| "layer.8.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 13, |
| "offset": 22008320, |
| "bytes": 5120 |
| } |
| }, |
| "layer.8.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 13, |
| "offset": 22013440, |
| "bytes": 5120 |
| } |
| }, |
| "layer.9.attention.k_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 13, |
| "offset": 22018560, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 13, |
| "offset": 23656960, |
| "bytes": 51200 |
| } |
| }, |
| "layer.9.attention.o_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 13, |
| "offset": 23708160, |
| "bytes": 2560 |
| } |
| }, |
| "layer.9.attention.o_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 13, |
| "offset": 23710720, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 13, |
| "offset": 25349120, |
| "bytes": 51200 |
| } |
| }, |
| "layer.9.attention.q_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 13, |
| "offset": 25400320, |
| "bytes": 2560 |
| } |
| }, |
| "layer.9.attention.q_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 13, |
| "offset": 25402880, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 13, |
| "offset": 27041280, |
| "bytes": 51200 |
| } |
| }, |
| "layer.9.attention.v_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 13, |
| "offset": 27092480, |
| "bytes": 2560 |
| } |
| }, |
| "layer.9.attention.v_proj.weight": { |
| "shape": [ |
| 1280, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 1280, |
| "padded_rows": 1280, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 13, |
| "offset": 27095040, |
| "bytes": 1638400 |
| }, |
| "scales": { |
| "shard": 13, |
| "offset": 28733440, |
| "bytes": 51200 |
| } |
| }, |
| "layer.9.layer_scale1.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 13, |
| "offset": 28784640, |
| "bytes": 5120 |
| } |
| }, |
| "layer.9.layer_scale2.lambda1": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 13, |
| "offset": 28789760, |
| "bytes": 5120 |
| } |
| }, |
| "layer.9.mlp.down_proj.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 13, |
| "offset": 28794880, |
| "bytes": 2560 |
| } |
| }, |
| "layer.9.mlp.down_proj.weight": { |
| "shape": [ |
| 1280, |
| 5120 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 1280, |
| "columns": 5120, |
| "padded_rows": 1280, |
| "padded_columns": 5120, |
| "data": { |
| "shard": 13, |
| "offset": 28797440, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 13, |
| "offset": 35351040, |
| "bytes": 204800 |
| } |
| }, |
| "layer.9.mlp.gate_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 13, |
| "offset": 35555840, |
| "bytes": 10240 |
| } |
| }, |
| "layer.9.mlp.gate_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 13, |
| "offset": 35566080, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 13, |
| "offset": 42119680, |
| "bytes": 204800 |
| } |
| }, |
| "layer.9.mlp.up_proj.bias": { |
| "shape": [ |
| 5120 |
| ], |
| "encoding": "raw", |
| "dtype": "f16", |
| "data": { |
| "shard": 13, |
| "offset": 42324480, |
| "bytes": 10240 |
| } |
| }, |
| "layer.9.mlp.up_proj.weight": { |
| "shape": [ |
| 5120, |
| 1280 |
| ], |
| "encoding": "symmetric-quantized", |
| "bits": 8, |
| "group_size": 64, |
| "output_tile": 16, |
| "rows": 5120, |
| "columns": 1280, |
| "padded_rows": 5120, |
| "padded_columns": 1280, |
| "data": { |
| "shard": 13, |
| "offset": 42334720, |
| "bytes": 6553600 |
| }, |
| "scales": { |
| "shard": 13, |
| "offset": 48888320, |
| "bytes": 204800 |
| } |
| }, |
| "layer.9.norm1.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 13, |
| "offset": 49093120, |
| "bytes": 5120 |
| } |
| }, |
| "layer.9.norm1.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 13, |
| "offset": 49098240, |
| "bytes": 5120 |
| } |
| }, |
| "layer.9.norm2.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 13, |
| "offset": 49103360, |
| "bytes": 5120 |
| } |
| }, |
| "layer.9.norm2.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 13, |
| "offset": 49108480, |
| "bytes": 5120 |
| } |
| }, |
| "norm.bias": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 13, |
| "offset": 49113600, |
| "bytes": 5120 |
| } |
| }, |
| "norm.weight": { |
| "shape": [ |
| 1280 |
| ], |
| "encoding": "raw", |
| "dtype": "f32", |
| "data": { |
| "shard": 13, |
| "offset": 49118720, |
| "bytes": 5120 |
| } |
| } |
| } |
| } |
|
|
