{ "metadata": { "ParamSize": 579, "ParamBytes": 2631839744.0, "BitsPerParam": 4.518310627966404 }, "records": [ { "dataPath": "params_shard_0.bin", "format": "raw-shard", "nbytes": 317849600, "records": [ { "name": "model.embed_tokens.q_weight", "shape": [ 248320, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 317849600, "byteOffset": 0 } ], "md5sum": "2d967a1aa36e68653ddde2155e3ebdc0" }, { "dataPath": "params_shard_1.bin", "format": "raw-shard", "nbytes": 39731200, "records": [ { "name": "model.embed_tokens.q_scale", "shape": [ 248320, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 39731200, "byteOffset": 0 } ], "md5sum": "08bffe5e5b0ee09e447b90c42c862491" }, { "dataPath": "params_shard_2.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.1.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "0bef63597eb18dcfcd3f9226c3dbf18d" }, { "dataPath": "params_shard_3.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.10.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "2618d59acf6747d7d5c8822f3e8a53bf" }, { "dataPath": "params_shard_4.bin", "format": "raw-shard", "nbytes": 32440320, "records": [ { "name": "model.layers.1.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 0 }, { "name": "model.layers.1.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.1.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.10.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 16220160 }, { "name": "model.layers.10.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.10.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "363a1d84722d12cf76c8b2be6fc51c85" }, { "dataPath": "params_shard_5.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.11.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "77fe5b7c3c95ec38e65b80c58c82b36c" }, { "dataPath": "params_shard_6.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.18.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "19067070c97671fa1c2a206d9acc3caa" }, { "dataPath": "params_shard_7.bin", "format": "raw-shard", "nbytes": 32440320, "records": [ { "name": "model.layers.11.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 0 }, { "name": "model.layers.11.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.11.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.18.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 16220160 }, { "name": "model.layers.18.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.18.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "ea5dd6363ea3a2fd133ac9f86f0f28f2" }, { "dataPath": "params_shard_8.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.19.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "c07809ae81e5e9673ebac477d3039cbf" }, { "dataPath": "params_shard_9.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.12.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "e15f254d04d4d04fcd69b11e467fb2fd" }, { "dataPath": "params_shard_10.bin", "format": "raw-shard", "nbytes": 32440320, "records": [ { "name": "model.layers.19.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 0 }, { "name": "model.layers.19.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.19.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.12.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 16220160 }, { "name": "model.layers.12.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.12.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "80627ee935049c97f27d10e40502e8ef" }, { "dataPath": "params_shard_11.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.13.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "6b62e7e2ca387f7500d8ecf8051748c7" }, { "dataPath": "params_shard_12.bin", "format": "raw-shard", "nbytes": 29491200, "records": [ { "name": "model.layers.13.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 0 }, { "name": "model.layers.13.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.13.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.7.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 16220160 }, { "name": "model.layers.7.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 } ], "md5sum": "e6e21711a952a92e41a7ae93393bd659" }, { "dataPath": "params_shard_13.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.8.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "b487027fea788e50f30ee3c11b251470" }, { "dataPath": "params_shard_14.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.21.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "41141eb7e06379ce469611920bc8d394" }, { "dataPath": "params_shard_15.bin", "format": "raw-shard", "nbytes": 32440320, "records": [ { "name": "model.layers.8.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 0 }, { "name": "model.layers.8.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.8.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.21.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 16220160 }, { "name": "model.layers.21.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.21.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "9e9cf351a07320b1a9688a02c655d32a" }, { "dataPath": "params_shard_16.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.22.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "499fab533c1c36ad0466736d7ac539c9" }, { "dataPath": "params_shard_17.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.23.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "4efd0207ed1df392cfe9d781365908fe" }, { "dataPath": "params_shard_18.bin", "format": "raw-shard", "nbytes": 32440320, "records": [ { "name": "model.layers.22.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 0 }, { "name": "model.layers.22.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.22.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.23.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 16220160 }, { "name": "model.layers.23.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.23.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "250260fcc9912a977ea7e7aa319fd17b" }, { "dataPath": "params_shard_19.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.31.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "406395d061ed45d47a48fc70f047cffe" }, { "dataPath": "params_shard_20.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.4.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "ca4230da4061a253dba883009c735fe0" }, { "dataPath": "params_shard_21.bin", "format": "raw-shard", "nbytes": 32440320, "records": [ { "name": "model.layers.31.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 0 }, { "name": "model.layers.31.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.31.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.4.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 16220160 }, { "name": "model.layers.4.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.4.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "ca1c8ac93aa7f1e42461c3ec568e4cce" }, { "dataPath": "params_shard_22.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.14.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "58c279f5d476252015d66acc0fb6795c" }, { "dataPath": "params_shard_23.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.15.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "6f5ae6a16cf9ccdc2578422b2ea108d6" }, { "dataPath": "params_shard_24.bin", "format": "raw-shard", "nbytes": 32440320, "records": [ { "name": "model.layers.14.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 0 }, { "name": "model.layers.14.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.14.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.15.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 16220160 }, { "name": "model.layers.15.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.15.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "1e18032442aea858d2cb84249cb20820" }, { "dataPath": "params_shard_25.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.9.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "8b9fcef98d82ba4185ce6a7a8616fce4" }, { "dataPath": "params_shard_26.bin", "format": "raw-shard", "nbytes": 29491200, "records": [ { "name": "model.layers.9.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 0 }, { "name": "model.layers.9.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.9.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.29.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 16220160 }, { "name": "model.layers.29.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 } ], "md5sum": "f24b32e33fac4946ea1069c6e46a3913" }, { "dataPath": "params_shard_27.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.3.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "c3b7baea531c4e01c0a8777d5ec29a00" }, { "dataPath": "params_shard_28.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.30.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "425aa9bfd2d90a9cf0b95c3b6007455b" }, { "dataPath": "params_shard_29.bin", "format": "raw-shard", "nbytes": 32440320, "records": [ { "name": "model.layers.3.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 0 }, { "name": "model.layers.3.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.3.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.30.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 16220160 }, { "name": "model.layers.30.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.30.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "d14ff85b5f25521e7119c0b63037bf23" }, { "dataPath": "params_shard_30.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.16.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "f8a4d8bd40b9e758b3c5ac24c8ba77fa" }, { "dataPath": "params_shard_31.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.17.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "16d93ea18776803bae860404465fa784" }, { "dataPath": "params_shard_32.bin", "format": "raw-shard", "nbytes": 32440320, "records": [ { "name": "model.layers.16.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 0 }, { "name": "model.layers.16.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.16.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.17.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 16220160 }, { "name": "model.layers.17.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.17.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "7c245f8f0f1a1e41a573aa40c9223519" }, { "dataPath": "params_shard_33.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.24.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "2b4af8f6666688a4106ecb3386b511e3" }, { "dataPath": "params_shard_34.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.5.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "d3e37355a009bfc949b85297aa557d7c" }, { "dataPath": "params_shard_35.bin", "format": "raw-shard", "nbytes": 32440320, "records": [ { "name": "model.layers.24.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 0 }, { "name": "model.layers.24.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.24.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.5.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 16220160 }, { "name": "model.layers.5.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.5.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "2f5d7cf31ce431136d09cf2df9aab7c5" }, { "dataPath": "params_shard_36.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.6.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "aef9ba7bdd7978dd87a1b1d55e497b86" }, { "dataPath": "params_shard_37.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.7.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "84bc4ac96a0afc806a61b6db87b46af9" }, { "dataPath": "params_shard_38.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.2.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "96d2341912c6c46e59783543009aada1" }, { "dataPath": "params_shard_39.bin", "format": "raw-shard", "nbytes": 32440320, "records": [ { "name": "model.layers.6.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 0 }, { "name": "model.layers.6.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.6.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.7.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 16220160 }, { "name": "model.layers.2.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 19169280 }, { "name": "model.layers.2.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 30965760 } ], "md5sum": "3a10a4ccdfa6d6bd6bcab8d0110c30ad" }, { "dataPath": "params_shard_40.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.20.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "5c5f9c24a09677b3ca04af8951445f8b" }, { "dataPath": "params_shard_41.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.25.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "b2a33973468f8a8efaae96e0384f5071" }, { "dataPath": "params_shard_42.bin", "format": "raw-shard", "nbytes": 32440320, "records": [ { "name": "model.layers.2.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 0 }, { "name": "model.layers.20.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 2949120 }, { "name": "model.layers.20.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 14745600 }, { "name": "model.layers.20.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 16220160 }, { "name": "model.layers.25.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 19169280 }, { "name": "model.layers.25.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 30965760 } ], "md5sum": "14a89f7d2e175bd96983d06b7c9d96c4" }, { "dataPath": "params_shard_43.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.26.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "1704e7ff0d3eed87df1f9e5532bf0242" }, { "dataPath": "params_shard_44.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.27.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "9c849a9b7cbc56f824e9514ab082da1b" }, { "dataPath": "params_shard_45.bin", "format": "raw-shard", "nbytes": 32440320, "records": [ { "name": "model.layers.25.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 0 }, { "name": "model.layers.26.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 2949120 }, { "name": "model.layers.26.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 14745600 }, { "name": "model.layers.26.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 16220160 }, { "name": "model.layers.27.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 19169280 }, { "name": "model.layers.27.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 30965760 } ], "md5sum": "905114febb08efa025527551fb8fcbb8" }, { "dataPath": "params_shard_46.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.28.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "8767bf094b4a855737722346e879d7e5" }, { "dataPath": "params_shard_47.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.29.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "2f42daac7e22bfb5b317ee54e3edec31" }, { "dataPath": "params_shard_48.bin", "format": "raw-shard", "nbytes": 22118400, "records": [ { "name": "model.layers.27.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 0 }, { "name": "model.layers.28.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 2949120 }, { "name": "model.layers.28.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 14745600 }, { "name": "model.layers.28.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 16220160 }, { "name": "model.layers.29.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 19169280 } ], "md5sum": "19368f0f937a27b36eb079f8ebd8d0b1" }, { "dataPath": "params_shard_49.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.0.mlp.gate_up_proj.q_weight", "shape": [ 18432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 23592960, "byteOffset": 0 } ], "md5sum": "b60742deaf0f75a94fbd582ecd6c4548" }, { "dataPath": "params_shard_50.bin", "format": "raw-shard", "nbytes": 28016640, "records": [ { "name": "model.layers.0.mlp.down_proj.q_weight", "shape": [ 2560, 1152 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 11796480, "byteOffset": 0 }, { "name": "model.layers.0.mlp.down_proj.q_scale", "shape": [ 2560, 288 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.0.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.10.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 16220160 }, { "name": "model.layers.10.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 26705920 } ], "md5sum": "b06089d400502ca9ef84392c1fb60020" }, { "dataPath": "params_shard_51.bin", "format": "raw-shard", "nbytes": 26542080, "records": [ { "name": "model.layers.11.self_attn.c_attn.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "model.layers.11.self_attn.c_attn.q_scale", "shape": [ 10240, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "model.layers.18.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 14745600 }, { "name": "model.layers.18.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 25231360 } ], "md5sum": "88e07abe82e4a021d0f1a6ae2d5e64da" }, { "dataPath": "params_shard_52.bin", "format": "raw-shard", "nbytes": 26542080, "records": [ { "name": "model.layers.19.self_attn.c_attn.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "model.layers.19.self_attn.c_attn.q_scale", "shape": [ 10240, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "model.layers.2.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 14745600 }, { "name": "model.layers.2.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 25231360 } ], "md5sum": "488b160eab7984b8569c37dd4a783252" }, { "dataPath": "params_shard_53.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.12.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 0 }, { "name": "model.layers.12.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 10485760 }, { "name": "model.layers.13.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 11796480 }, { "name": "model.layers.13.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 22282240 } ], "md5sum": "fd0733fd056aee17e58a06933cc18bcb" }, { "dataPath": "params_shard_54.bin", "format": "raw-shard", "nbytes": 26542080, "records": [ { "name": "model.layers.7.self_attn.c_attn.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "model.layers.7.self_attn.c_attn.q_scale", "shape": [ 10240, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "model.layers.8.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 14745600 }, { "name": "model.layers.8.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 25231360 } ], "md5sum": "c9d3ef7f9e27c21a3d788c6fa7d45ec9" }, { "dataPath": "params_shard_55.bin", "format": "raw-shard", "nbytes": 26542080, "records": [ { "name": "model.layers.22.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 0 }, { "name": "model.layers.22.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 10485760 }, { "name": "model.layers.23.self_attn.c_attn.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 11796480 }, { "name": "model.layers.23.self_attn.c_attn.q_scale", "shape": [ 10240, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 24903680 } ], "md5sum": "38b21b36ef6486caa1247d70ec1fa0d4" }, { "dataPath": "params_shard_56.bin", "format": "raw-shard", "nbytes": 26542080, "records": [ { "name": "model.layers.31.self_attn.c_attn.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "model.layers.31.self_attn.c_attn.q_scale", "shape": [ 10240, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "model.layers.4.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 14745600 }, { "name": "model.layers.4.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 25231360 } ], "md5sum": "4bf8342c26bd592a0afb45f5c6478504" }, { "dataPath": "params_shard_57.bin", "format": "raw-shard", "nbytes": 26542080, "records": [ { "name": "model.layers.14.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 0 }, { "name": "model.layers.14.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 10485760 }, { "name": "model.layers.15.self_attn.c_attn.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 11796480 }, { "name": "model.layers.15.self_attn.c_attn.q_scale", "shape": [ 10240, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 24903680 } ], "md5sum": "868065a930c4d4252e47402cf649a424" }, { "dataPath": "params_shard_58.bin", "format": "raw-shard", "nbytes": 26542080, "records": [ { "name": "model.layers.9.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 0 }, { "name": "model.layers.9.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 10485760 }, { "name": "model.layers.3.self_attn.c_attn.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 11796480 }, { "name": "model.layers.3.self_attn.c_attn.q_scale", "shape": [ 10240, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 24903680 } ], "md5sum": "3383fc7ea931f99f21511ccd6e5385c0" }, { "dataPath": "params_shard_59.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.30.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 0 }, { "name": "model.layers.30.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 10485760 }, { "name": "model.layers.16.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 11796480 }, { "name": "model.layers.16.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 22282240 } ], "md5sum": "1d01f2d56b4d602d2296544fb5a185d7" }, { "dataPath": "params_shard_60.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.17.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 0 }, { "name": "model.layers.17.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 10485760 }, { "name": "model.layers.24.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 11796480 }, { "name": "model.layers.24.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 22282240 } ], "md5sum": "6810d7af8a6fe94fd99276d63f5f6eeb" }, { "dataPath": "params_shard_61.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.25.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 0 }, { "name": "model.layers.25.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 10485760 }, { "name": "model.layers.5.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 11796480 }, { "name": "model.layers.5.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 22282240 } ], "md5sum": "620332db5a808452ae2e4acf9b393a33" }, { "dataPath": "params_shard_62.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.6.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 0 }, { "name": "model.layers.6.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 10485760 }, { "name": "model.layers.20.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 11796480 }, { "name": "model.layers.20.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 22282240 } ], "md5sum": "3732b2843dbefbd8eccd0c84093f3331" }, { "dataPath": "params_shard_63.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.21.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 0 }, { "name": "model.layers.21.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 10485760 }, { "name": "model.layers.26.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 11796480 }, { "name": "model.layers.26.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 22282240 } ], "md5sum": "f68db20d93cf8b19851c229a307bff8d" }, { "dataPath": "params_shard_64.bin", "format": "raw-shard", "nbytes": 26542080, "records": [ { "name": "model.layers.27.self_attn.c_attn.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "model.layers.27.self_attn.c_attn.q_scale", "shape": [ 10240, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "model.layers.28.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 14745600 }, { "name": "model.layers.28.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 25231360 } ], "md5sum": "b71035491e20f95017ddb563bb9a1915" }, { "dataPath": "params_shard_65.bin", "format": "raw-shard", "nbytes": 23592960, "records": [ { "name": "model.layers.29.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 0 }, { "name": "model.layers.29.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 10485760 }, { "name": "model.layers.0.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 11796480 }, { "name": "model.layers.0.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 22282240 } ], "md5sum": "6310c7d6e366b4ede7066ae8f66a712a" }, { "dataPath": "params_shard_66.bin", "format": "raw-shard", "nbytes": 29491200, "records": [ { "name": "model.layers.1.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 0 }, { "name": "model.layers.1.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 10485760 }, { "name": "model.layers.10.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 11796480 }, { "name": "model.layers.10.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 17039360 }, { "name": "model.layers.10.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 17694720 }, { "name": "model.layers.10.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 22937600 }, { "name": "model.layers.11.self_attn.o_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 23592960 }, { "name": "model.layers.11.self_attn.o_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "685fd07348ef9b153cf2106b455c68fa" }, { "dataPath": "params_shard_67.bin", "format": "raw-shard", "nbytes": 29491200, "records": [ { "name": "model.layers.18.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 0 }, { "name": "model.layers.18.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 5242880 }, { "name": "model.layers.18.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 5898240 }, { "name": "model.layers.18.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 11141120 }, { "name": "model.layers.19.self_attn.o_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 11796480 }, { "name": "model.layers.19.self_attn.o_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 17039360 }, { "name": "model.layers.2.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 17694720 }, { "name": "model.layers.2.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 22937600 }, { "name": "model.layers.12.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 23592960 }, { "name": "model.layers.12.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "eec9bb64554d03224dad51816937a6d0" }, { "dataPath": "params_shard_68.bin", "format": "raw-shard", "nbytes": 29491200, "records": [ { "name": "model.layers.12.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 0 }, { "name": "model.layers.12.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 5242880 }, { "name": "model.layers.13.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 5898240 }, { "name": "model.layers.13.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 11141120 }, { "name": "model.layers.13.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 11796480 }, { "name": "model.layers.13.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 17039360 }, { "name": "model.layers.7.self_attn.o_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 17694720 }, { "name": "model.layers.7.self_attn.o_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 22937600 }, { "name": "model.layers.8.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 23592960 }, { "name": "model.layers.8.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "3119f6e91a9bd3bdaa50f1d493cb3a22" }, { "dataPath": "params_shard_69.bin", "format": "raw-shard", "nbytes": 29491200, "records": [ { "name": "model.layers.8.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 0 }, { "name": "model.layers.8.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 5242880 }, { "name": "model.layers.21.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 5898240 }, { "name": "model.layers.21.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 11141120 }, { "name": "model.layers.22.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 11796480 }, { "name": "model.layers.22.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 17039360 }, { "name": "model.layers.22.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 17694720 }, { "name": "model.layers.22.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 22937600 }, { "name": "model.layers.23.self_attn.o_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 23592960 }, { "name": "model.layers.23.self_attn.o_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "fec503509ba0877c849fede27c4ef630" }, { "dataPath": "params_shard_70.bin", "format": "raw-shard", "nbytes": 29491200, "records": [ { "name": "model.layers.31.self_attn.o_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 0 }, { "name": "model.layers.31.self_attn.o_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 5242880 }, { "name": "model.layers.4.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 5898240 }, { "name": "model.layers.4.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 11141120 }, { "name": "model.layers.4.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 11796480 }, { "name": "model.layers.4.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 17039360 }, { "name": "model.layers.14.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 17694720 }, { "name": "model.layers.14.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 22937600 }, { "name": "model.layers.14.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 23592960 }, { "name": "model.layers.14.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "e63d6086631e3a94fc7a8165b2b350e9" }, { "dataPath": "params_shard_71.bin", "format": "raw-shard", "nbytes": 29491200, "records": [ { "name": "model.layers.15.self_attn.o_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 0 }, { "name": "model.layers.15.self_attn.o_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 5242880 }, { "name": "model.layers.9.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 5898240 }, { "name": "model.layers.9.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 11141120 }, { "name": "model.layers.9.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 11796480 }, { "name": "model.layers.9.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 17039360 }, { "name": "model.layers.3.self_attn.o_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 17694720 }, { "name": "model.layers.3.self_attn.o_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 22937600 }, { "name": "model.layers.30.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 23592960 }, { "name": "model.layers.30.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "5b4de635c92c3eb002eefe231d03a513" }, { "dataPath": "params_shard_72.bin", "format": "raw-shard", "nbytes": 29491200, "records": [ { "name": "model.layers.30.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 0 }, { "name": "model.layers.30.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 5242880 }, { "name": "model.layers.16.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 5898240 }, { "name": "model.layers.16.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 11141120 }, { "name": "model.layers.16.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 11796480 }, { "name": "model.layers.16.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 17039360 }, { "name": "model.layers.17.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 17694720 }, { "name": "model.layers.17.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 22937600 }, { "name": "model.layers.17.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 23592960 }, { "name": "model.layers.17.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "43d7e4199b31445d488b480e86395ffb" }, { "dataPath": "params_shard_73.bin", "format": "raw-shard", "nbytes": 29491200, "records": [ { "name": "model.layers.24.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 0 }, { "name": "model.layers.24.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 5242880 }, { "name": "model.layers.24.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 5898240 }, { "name": "model.layers.24.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 11141120 }, { "name": "model.layers.25.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 11796480 }, { "name": "model.layers.25.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 17039360 }, { "name": "model.layers.25.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 17694720 }, { "name": "model.layers.25.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 22937600 }, { "name": "model.layers.5.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 23592960 }, { "name": "model.layers.5.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "5cd9ab41a5f96e68f118bb3eebc86e4b" }, { "dataPath": "params_shard_74.bin", "format": "raw-shard", "nbytes": 29491200, "records": [ { "name": "model.layers.5.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 0 }, { "name": "model.layers.5.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 5242880 }, { "name": "model.layers.6.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 5898240 }, { "name": "model.layers.6.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 11141120 }, { "name": "model.layers.6.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 11796480 }, { "name": "model.layers.6.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 17039360 }, { "name": "model.layers.2.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 17694720 }, { "name": "model.layers.2.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 22937600 }, { "name": "model.layers.20.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 23592960 }, { "name": "model.layers.20.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "b269707abfc9f5b3f04bf5a4c31ad172" }, { "dataPath": "params_shard_75.bin", "format": "raw-shard", "nbytes": 29491200, "records": [ { "name": "model.layers.20.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 0 }, { "name": "model.layers.20.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 5242880 }, { "name": "model.layers.21.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 5898240 }, { "name": "model.layers.21.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 11141120 }, { "name": "model.layers.26.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 11796480 }, { "name": "model.layers.26.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 17039360 }, { "name": "model.layers.26.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 17694720 }, { "name": "model.layers.26.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 22937600 }, { "name": "model.layers.27.self_attn.o_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 23592960 }, { "name": "model.layers.27.self_attn.o_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "77ab720b1f8e16e1b64c19c49ca64307" }, { "dataPath": "params_shard_76.bin", "format": "raw-shard", "nbytes": 29491200, "records": [ { "name": "model.layers.28.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 0 }, { "name": "model.layers.28.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 5242880 }, { "name": "model.layers.28.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 5898240 }, { "name": "model.layers.28.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 11141120 }, { "name": "model.layers.29.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 11796480 }, { "name": "model.layers.29.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 17039360 }, { "name": "model.layers.29.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 17694720 }, { "name": "model.layers.29.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 22937600 }, { "name": "model.layers.0.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 23592960 }, { "name": "model.layers.0.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "4e09ab588a78fa8a1c2f1200dea0d306" }, { "dataPath": "params_shard_77.bin", "format": "raw-shard", "nbytes": 21829632, "records": [ { "name": "model.layers.0.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 0 }, { "name": "model.layers.0.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 5242880 }, { "name": "model.layers.1.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 5898240 }, { "name": "model.layers.1.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 11141120 }, { "name": "model.layers.1.linear_attn.out_proj.q_weight", "shape": [ 2560, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 5242880, "byteOffset": 11796480 }, { "name": "model.layers.1.linear_attn.out_proj.q_scale", "shape": [ 2560, 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 17039360 }, { "name": "model.layers.10.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 17694720 }, { "name": "model.layers.10.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 17735680 }, { "name": "model.layers.10.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 17740800 }, { "name": "model.layers.10.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 17781760 }, { "name": "model.layers.18.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 17786880 }, { "name": "model.layers.18.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 17827840 }, { "name": "model.layers.18.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 17832960 }, { "name": "model.layers.18.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 17873920 }, { "name": "model.layers.2.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 17879040 }, { "name": "model.layers.2.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 17920000 }, { "name": "model.layers.2.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 17925120 }, { "name": "model.layers.2.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 17966080 }, { "name": "model.layers.12.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 17971200 }, { "name": "model.layers.12.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18012160 }, { "name": "model.layers.12.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18017280 }, { "name": "model.layers.12.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18058240 }, { "name": "model.layers.13.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18063360 }, { "name": "model.layers.13.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18104320 }, { "name": "model.layers.13.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18109440 }, { "name": "model.layers.13.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18150400 }, { "name": "model.layers.8.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18155520 }, { "name": "model.layers.8.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18196480 }, { "name": "model.layers.8.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18201600 }, { "name": "model.layers.8.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18242560 }, { "name": "model.layers.22.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18247680 }, { "name": "model.layers.22.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18288640 }, { "name": "model.layers.22.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18293760 }, { "name": "model.layers.22.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18334720 }, { "name": "model.layers.4.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18339840 }, { "name": "model.layers.4.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18380800 }, { "name": "model.layers.4.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18385920 }, { "name": "model.layers.4.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18426880 }, { "name": "model.layers.14.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18432000 }, { "name": "model.layers.14.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18472960 }, { "name": "model.layers.14.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18478080 }, { "name": "model.layers.14.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18519040 }, { "name": "model.layers.9.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18524160 }, { "name": "model.layers.9.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18565120 }, { "name": "model.layers.9.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18570240 }, { "name": "model.layers.9.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18611200 }, { "name": "model.layers.30.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18616320 }, { "name": "model.layers.30.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18657280 }, { "name": "model.layers.30.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18662400 }, { "name": "model.layers.30.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18703360 }, { "name": "model.layers.16.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18708480 }, { "name": "model.layers.16.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18749440 }, { "name": "model.layers.16.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18754560 }, { "name": "model.layers.16.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18795520 }, { "name": "model.layers.17.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18800640 }, { "name": "model.layers.17.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18841600 }, { "name": "model.layers.17.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18846720 }, { "name": "model.layers.17.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18887680 }, { "name": "model.layers.24.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18892800 }, { "name": "model.layers.24.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18933760 }, { "name": "model.layers.24.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18938880 }, { "name": "model.layers.24.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 18979840 }, { "name": "model.layers.25.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 18984960 }, { "name": "model.layers.25.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19025920 }, { "name": "model.layers.25.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19031040 }, { "name": "model.layers.25.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19072000 }, { "name": "model.layers.5.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19077120 }, { "name": "model.layers.5.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19118080 }, { "name": "model.layers.5.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19123200 }, { "name": "model.layers.5.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19164160 }, { "name": "model.layers.6.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19169280 }, { "name": "model.layers.6.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19210240 }, { "name": "model.layers.6.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19215360 }, { "name": "model.layers.6.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19256320 }, { "name": "model.layers.20.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19261440 }, { "name": "model.layers.20.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19302400 }, { "name": "model.layers.20.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19307520 }, { "name": "model.layers.20.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19348480 }, { "name": "model.layers.21.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19353600 }, { "name": "model.layers.21.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19394560 }, { "name": "model.layers.21.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19399680 }, { "name": "model.layers.21.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19440640 }, { "name": "model.layers.26.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19445760 }, { "name": "model.layers.26.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19486720 }, { "name": "model.layers.26.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19491840 }, { "name": "model.layers.26.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19532800 }, { "name": "model.layers.28.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19537920 }, { "name": "model.layers.28.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19578880 }, { "name": "model.layers.28.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19584000 }, { "name": "model.layers.28.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19624960 }, { "name": "model.layers.29.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19630080 }, { "name": "model.layers.29.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19671040 }, { "name": "model.layers.29.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19676160 }, { "name": "model.layers.29.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19717120 }, { "name": "model.layers.0.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19722240 }, { "name": "model.layers.0.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19763200 }, { "name": "model.layers.0.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19768320 }, { "name": "model.layers.0.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19809280 }, { "name": "model.layers.1.linear_attn.in_proj_b.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19814400 }, { "name": "model.layers.1.linear_attn.in_proj_b.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19855360 }, { "name": "model.layers.1.linear_attn.in_proj_a.q_weight", "shape": [ 32, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 40960, "byteOffset": 19860480 }, { "name": "model.layers.1.linear_attn.in_proj_a.q_scale", "shape": [ 32, 80 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19901440 }, { "name": "model.layers.10.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19906560 }, { "name": "model.layers.2.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19972096 }, { "name": "model.layers.12.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20037632 }, { "name": "model.layers.13.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20103168 }, { "name": "model.layers.8.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20168704 }, { "name": "model.layers.9.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20234240 }, { "name": "model.layers.22.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20299776 }, { "name": "model.layers.4.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20365312 }, { "name": "model.layers.5.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20430848 }, { "name": "model.layers.14.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20496384 }, { "name": "model.layers.16.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20561920 }, { "name": "model.layers.30.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20627456 }, { "name": "model.layers.17.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20692992 }, { "name": "model.layers.18.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20758528 }, { "name": "model.layers.24.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20824064 }, { "name": "model.layers.25.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20889600 }, { "name": "model.layers.6.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20955136 }, { "name": "model.layers.20.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21020672 }, { "name": "model.layers.21.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21086208 }, { "name": "model.layers.26.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21151744 }, { "name": "model.layers.28.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21217280 }, { "name": "model.layers.29.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21282816 }, { "name": "model.layers.0.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21348352 }, { "name": "model.layers.1.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21413888 }, { "name": "model.layers.10.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21479424 }, { "name": "model.layers.10.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21484544 }, { "name": "model.layers.11.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21489664 }, { "name": "model.layers.11.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21494784 }, { "name": "model.layers.18.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21499904 }, { "name": "model.layers.19.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21505024 }, { "name": "model.layers.19.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21510144 }, { "name": "model.layers.2.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21515264 }, { "name": "model.layers.12.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21520384 }, { "name": "model.layers.12.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21525504 }, { "name": "model.layers.13.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21530624 }, { "name": "model.layers.13.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21535744 }, { "name": "model.layers.7.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21540864 }, { "name": "model.layers.8.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21545984 }, { "name": "model.layers.8.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21551104 }, { "name": "model.layers.9.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21556224 }, { "name": "model.layers.21.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21561344 }, { "name": "model.layers.22.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21566464 }, { "name": "model.layers.22.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21571584 }, { "name": "model.layers.23.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21576704 }, { "name": "model.layers.23.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21581824 }, { "name": "model.layers.31.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21586944 }, { "name": "model.layers.4.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21592064 }, { "name": "model.layers.4.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21597184 }, { "name": "model.layers.14.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21602304 }, { "name": "model.layers.14.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21607424 }, { "name": "model.layers.15.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21612544 }, { "name": "model.layers.15.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21617664 }, { "name": "model.layers.16.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21622784 }, { "name": "model.layers.9.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21627904 }, { "name": "model.layers.3.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21633024 }, { "name": "model.layers.3.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21638144 }, { "name": "model.layers.30.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21643264 }, { "name": "model.layers.30.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21648384 }, { "name": "model.layers.31.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21653504 }, { "name": "model.layers.16.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21658624 }, { "name": "model.layers.17.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21663744 }, { "name": "model.layers.17.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21668864 }, { "name": "model.layers.18.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21673984 }, { "name": "model.layers.24.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21679104 }, { "name": "model.layers.24.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21684224 }, { "name": "model.layers.25.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21689344 }, { "name": "model.layers.25.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21694464 }, { "name": "model.layers.5.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21699584 }, { "name": "model.layers.5.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21704704 }, { "name": "model.layers.6.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21709824 }, { "name": "model.layers.6.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21714944 }, { "name": "model.layers.7.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21720064 }, { "name": "model.layers.2.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21725184 }, { "name": "model.layers.20.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21730304 }, { "name": "model.layers.20.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21735424 }, { "name": "model.layers.21.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21740544 }, { "name": "model.layers.26.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21745664 }, { "name": "model.layers.26.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21750784 }, { "name": "model.layers.27.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21755904 }, { "name": "model.layers.27.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21761024 }, { "name": "model.layers.28.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21766144 }, { "name": "model.layers.28.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21771264 }, { "name": "model.layers.29.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21776384 }, { "name": "model.layers.29.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21781504 }, { "name": "model.layers.0.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21786624 }, { "name": "model.layers.0.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21791744 }, { "name": "model.layers.1.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21796864 }, { "name": "model.layers.1.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21801984 }, { "name": "model.norm.weight", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21807104 }, { "name": "model.layers.10.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21812224 }, { "name": "model.layers.11.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21812480 }, { "name": "model.layers.11.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21812992 }, { "name": "model.layers.18.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21813504 }, { "name": "model.layers.19.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21813760 }, { "name": "model.layers.19.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21814272 }, { "name": "model.layers.2.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21814784 }, { "name": "model.layers.12.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21815040 }, { "name": "model.layers.13.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21815296 }, { "name": "model.layers.7.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21815552 }, { "name": "model.layers.7.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21816064 }, { "name": "model.layers.8.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21816576 }, { "name": "model.layers.22.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21816832 }, { "name": "model.layers.23.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21817088 }, { "name": "model.layers.31.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21817600 }, { "name": "model.layers.31.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21818112 }, { "name": "model.layers.4.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21818624 }, { "name": "model.layers.14.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21818880 }, { "name": "model.layers.15.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21819136 }, { "name": "model.layers.15.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21819648 }, { "name": "model.layers.9.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21820160 }, { "name": "model.layers.3.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21820416 }, { "name": "model.layers.3.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21820928 }, { "name": "model.layers.30.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21821440 }, { "name": "model.layers.16.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21821696 }, { "name": "model.layers.17.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21821952 }, { "name": "model.layers.23.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21822208 }, { "name": "model.layers.24.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21822720 }, { "name": "model.layers.25.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21822976 }, { "name": "model.layers.5.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21823232 }, { "name": "model.layers.6.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21823488 }, { "name": "model.layers.20.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21823744 }, { "name": "model.layers.21.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21824000 }, { "name": "model.layers.26.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21824256 }, { "name": "model.layers.27.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21824512 }, { "name": "model.layers.27.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21825024 }, { "name": "model.layers.28.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21825536 }, { "name": "model.layers.29.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21825792 }, { "name": "model.layers.0.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21826048 }, { "name": "model.layers.1.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21826304 }, { "name": "model.layers.10.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21826560 }, { "name": "model.layers.12.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21826624 }, { "name": "model.layers.2.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21826688 }, { "name": "model.layers.13.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21826752 }, { "name": "model.layers.14.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21826816 }, { "name": "model.layers.8.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21826880 }, { "name": "model.layers.9.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21826944 }, { "name": "model.layers.22.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827008 }, { "name": "model.layers.4.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827072 }, { "name": "model.layers.5.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827136 }, { "name": "model.layers.16.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827200 }, { "name": "model.layers.30.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827264 }, { "name": "model.layers.17.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827328 }, { "name": "model.layers.18.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827392 }, { "name": "model.layers.24.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827456 }, { "name": "model.layers.25.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827520 }, { "name": "model.layers.6.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827584 }, { "name": "model.layers.20.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827648 }, { "name": "model.layers.21.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827712 }, { "name": "model.layers.26.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827776 }, { "name": "model.layers.28.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827840 }, { "name": "model.layers.29.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827904 }, { "name": "model.layers.0.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21827968 }, { "name": "model.layers.1.linear_attn.A_log", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828032 }, { "name": "model.layers.10.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828096 }, { "name": "model.layers.2.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828160 }, { "name": "model.layers.12.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828224 }, { "name": "model.layers.13.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828288 }, { "name": "model.layers.8.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828352 }, { "name": "model.layers.9.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828416 }, { "name": "model.layers.22.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828480 }, { "name": "model.layers.4.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828544 }, { "name": "model.layers.5.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828608 }, { "name": "model.layers.14.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828672 }, { "name": "model.layers.16.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828736 }, { "name": "model.layers.30.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828800 }, { "name": "model.layers.17.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828864 }, { "name": "model.layers.18.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828928 }, { "name": "model.layers.24.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21828992 }, { "name": "model.layers.25.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21829056 }, { "name": "model.layers.6.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21829120 }, { "name": "model.layers.20.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21829184 }, { "name": "model.layers.21.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21829248 }, { "name": "model.layers.26.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21829312 }, { "name": "model.layers.28.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21829376 }, { "name": "model.layers.29.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21829440 }, { "name": "model.layers.0.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21829504 }, { "name": "model.layers.1.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 64, "byteOffset": 21829568 } ], "md5sum": "305657b1e61de49da51e5ed6439f5f76" } ] }