{ "metadata": { "ParamSize": 579, "ParamBytes": 2367120384.0, "BitsPerParam": 4.063843633749424 }, "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": "b87f9eaaf4d7a4061b8756d74f2c248e" }, { "dataPath": "params_shard_1.bin", "format": "raw-shard", "nbytes": 39731200, "records": [ { "name": "model.embed_tokens.q_scale", "shape": [ 248320, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 39731200, "byteOffset": 0 } ], "md5sum": "f3e9b86a58de53d8633f0907a3769ae1" }, { "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": "7a3d7d7128adaca8c04622952741547b" }, { "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": "1bf4cb134b715ec7fafcd5deb4368b19" }, { "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.1.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.10.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "383b3d70bde6a4f978a7b7c16f57af19" }, { "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": "a77e7b946f8e8c25f3dbfbec7df7cc4d" }, { "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": "3936af22b5248daec78b4ba6b971e716" }, { "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.11.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.18.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "b3a5a3eab05370d38543653d087c1173" }, { "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": "b8c43d94b72b000dbec4a51511160703" }, { "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": "221e3c255336eb74b1609b0ad649975e" }, { "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.19.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.12.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "0425c57f0ead941fe8dbe9bdac2040e4" }, { "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": "00ff3a02f95a008fc329376db17ed0a6" }, { "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.13.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 } ], "md5sum": "58d116c527fe2d7691dbdc563292a4b3" }, { "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": "5bbc32fba2cfa11c254b79ec830e193d" }, { "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": "025ba9308858d0dc93838bfbfe1d5075" }, { "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.8.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.21.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "062bbd3af1a4dec75c0d3553c76bee07" }, { "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": "1f6000b9f49c8bbb06af124cdfc662a4" }, { "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": "37e1d517964e8c0f606ecf5b3cae6260" }, { "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.22.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.23.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "05b0bb608f3989def4f0578eae422351" }, { "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": "739ee205658ec985ad46955ada26258d" }, { "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": "aa161e99c3d2f0dca988c0efffa1bd02" }, { "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.31.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.4.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "24f7d3f47423d481dad023cf9eee4d49" }, { "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": "68aa682d60ce9a48fd8738e6228d4047" }, { "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": "3575d45932922091681e09522183d115" }, { "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.14.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.15.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "f19c598671a96897abbc82850abfe3bc" }, { "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": "f4ebe713f4c5a99a298d67e64d0691c5" }, { "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.9.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 } ], "md5sum": "d378c37ba18f39b5c9819afd0c7026d8" }, { "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": "86f9ce846732c390dbf2ef1b2c04fae6" }, { "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": "c96b8c206e4c0ef179cf822dbbdb4855" }, { "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.3.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.30.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "f99253e7ba9d24effd24def797d0db26" }, { "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": "7731e12293b90b564d70699256162449" }, { "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": "2fc72293f7235e448dbfe9f024d02513" }, { "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.16.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.17.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "04499cc8310cba17d278aadb9bec5c98" }, { "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": "6f0014a2d5e06beb58a8fae4697fc0a5" }, { "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": "395288ff85a0a2114e6b92cac28b08fe" }, { "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.24.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 28016640 }, { "name": "model.layers.5.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 29491200 } ], "md5sum": "830896265a3a7ab060b6050b355d431d" }, { "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": "36b075dedf15fd9826b5d5bf00798c46" }, { "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": "da0036de791b23768b668222bd947be1" }, { "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": "39e6d25294f16e5b491ff1740202a2e2" }, { "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.6.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 13271040 }, { "name": "model.layers.7.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 30965760 } ], "md5sum": "6d9c349cdcf639d8b5915c6fc16bb377" }, { "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": "bf26d1b7581176004cbb04670965a2a9" }, { "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": "1c6ffee7f54e965d2883607559c44cee" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 14745600 }, { "name": "model.layers.20.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 30965760 } ], "md5sum": "c4587de350a5fd287fe57dd9fc710ee6" }, { "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": "63bcb619361b9c8753ae8bff4db28742" }, { "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": "8ea35fada55beda174efc7ae52fb6438" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 14745600 }, { "name": "model.layers.26.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 30965760 } ], "md5sum": "558b68082601e1a77aa9602a21b9ca3a" }, { "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": "12003c18cc65f33fc0bb569ff65d8314" }, { "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": "98a79b4e87d48017bbcf28f07710039c" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 14745600 }, { "name": "model.layers.28.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 16220160 }, { "name": "model.layers.29.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2949120, "byteOffset": 19169280 } ], "md5sum": "5a064e740bce408fb34f8629d9b99591" }, { "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": "fa69217433430a14ebbf98842593c36c" }, { "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": "float16", "format": "f32-to-bf16", "nbytes": 1474560, "byteOffset": 11796480 }, { "name": "model.layers.0.mlp.gate_up_proj.q_scale", "shape": [ 18432, 80 ], "dtype": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 26705920 } ], "md5sum": "2ab233839fe18567680a5b1204c3b6d4" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 25231360 } ], "md5sum": "5153fa38d433a4656459b9cd6815f0a2" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 25231360 } ], "md5sum": "4b50d72a7255c54da3a4969e7431e90c" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 22282240 } ], "md5sum": "fac9fd52cfd121601318abd2f31f32aa" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 25231360 } ], "md5sum": "c4357153895c142891db3884fc3ecbd9" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 24903680 } ], "md5sum": "66577e4eabd1048f7c71296523e16ef9" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 25231360 } ], "md5sum": "5f4700deaf8a64a79bb3ba4d11afcf8e" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 24903680 } ], "md5sum": "ec47608e1806ca9879244c2c080b40fa" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 24903680 } ], "md5sum": "f2ca326e9762922fb30b8961965b09ea" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 22282240 } ], "md5sum": "a08237bdce6982f0611b19ff77ad480d" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 22282240 } ], "md5sum": "9a716a28ec9c1800c9527d366e2b0ebf" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 22282240 } ], "md5sum": "fbe4368b4505ed5306860471a32631f1" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 22282240 } ], "md5sum": "c343e7e54c25cad03f1bd547ccab18f9" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 22282240 } ], "md5sum": "3962426d05ad36ff4e678508c5c4d169" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 25231360 } ], "md5sum": "2b38539d40821a9106c55fed4162b466" }, { "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 1310720, "byteOffset": 22282240 } ], "md5sum": "3fc73192578f17669aee6f88544318cb" }, { "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": "float16", "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": "float16", "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "298d632a6c09616682bcfc3a50174016" }, { "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "e6445dae830bc6a6b5fed81ccf56d5ce" }, { "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "7db882f956056709af87f2b908e1caa6" }, { "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "5248ba801d6060dee1f411a208f5dc55" }, { "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "210901807e73b52021d9b40c3353d51d" }, { "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "02291ec03772a0c30008877f8a828e60" }, { "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "0e0d26ecbb3ef901d07825d39bcb1d78" }, { "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "c1bd69a009a249cbce07bf0c5c2f66dc" }, { "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "279c1de74ecc3377177b54106d5c9514" }, { "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "84d318c50d6794efcde73dfce541dda6" }, { "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 655360, "byteOffset": 28835840 } ], "md5sum": "e978802625504f7272b92a0378c4104b" }, { "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "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": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 19901440 }, { "name": "model.layers.10.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19906560 }, { "name": "model.layers.2.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19972096 }, { "name": "model.layers.12.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20037632 }, { "name": "model.layers.13.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20103168 }, { "name": "model.layers.8.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20168704 }, { "name": "model.layers.9.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20234240 }, { "name": "model.layers.22.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20299776 }, { "name": "model.layers.4.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20365312 }, { "name": "model.layers.5.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20430848 }, { "name": "model.layers.14.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20496384 }, { "name": "model.layers.16.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20561920 }, { "name": "model.layers.30.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20627456 }, { "name": "model.layers.17.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20692992 }, { "name": "model.layers.18.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20758528 }, { "name": "model.layers.24.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20824064 }, { "name": "model.layers.25.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20889600 }, { "name": "model.layers.6.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20955136 }, { "name": "model.layers.20.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21020672 }, { "name": "model.layers.21.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21086208 }, { "name": "model.layers.26.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21151744 }, { "name": "model.layers.28.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21217280 }, { "name": "model.layers.29.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21282816 }, { "name": "model.layers.0.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21348352 }, { "name": "model.layers.1.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21413888 }, { "name": "model.layers.10.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21479424 }, { "name": "model.layers.10.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21484544 }, { "name": "model.layers.11.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21489664 }, { "name": "model.layers.11.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21494784 }, { "name": "model.layers.18.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21499904 }, { "name": "model.layers.19.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21505024 }, { "name": "model.layers.19.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21510144 }, { "name": "model.layers.2.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21515264 }, { "name": "model.layers.12.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21520384 }, { "name": "model.layers.12.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21525504 }, { "name": "model.layers.13.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21530624 }, { "name": "model.layers.13.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21535744 }, { "name": "model.layers.7.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21540864 }, { "name": "model.layers.8.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21545984 }, { "name": "model.layers.8.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21551104 }, { "name": "model.layers.9.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21556224 }, { "name": "model.layers.21.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21561344 }, { "name": "model.layers.22.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21566464 }, { "name": "model.layers.22.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21571584 }, { "name": "model.layers.23.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21576704 }, { "name": "model.layers.23.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21581824 }, { "name": "model.layers.31.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21586944 }, { "name": "model.layers.4.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21592064 }, { "name": "model.layers.4.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21597184 }, { "name": "model.layers.14.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21602304 }, { "name": "model.layers.14.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21607424 }, { "name": "model.layers.15.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21612544 }, { "name": "model.layers.15.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21617664 }, { "name": "model.layers.16.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21622784 }, { "name": "model.layers.9.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21627904 }, { "name": "model.layers.3.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21633024 }, { "name": "model.layers.3.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21638144 }, { "name": "model.layers.30.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21643264 }, { "name": "model.layers.30.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21648384 }, { "name": "model.layers.31.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21653504 }, { "name": "model.layers.16.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21658624 }, { "name": "model.layers.17.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21663744 }, { "name": "model.layers.17.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21668864 }, { "name": "model.layers.18.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21673984 }, { "name": "model.layers.24.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21679104 }, { "name": "model.layers.24.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21684224 }, { "name": "model.layers.25.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21689344 }, { "name": "model.layers.25.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21694464 }, { "name": "model.layers.5.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21699584 }, { "name": "model.layers.5.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21704704 }, { "name": "model.layers.6.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21709824 }, { "name": "model.layers.6.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21714944 }, { "name": "model.layers.7.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21720064 }, { "name": "model.layers.2.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21725184 }, { "name": "model.layers.20.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21730304 }, { "name": "model.layers.20.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21735424 }, { "name": "model.layers.21.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21740544 }, { "name": "model.layers.26.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21745664 }, { "name": "model.layers.26.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21750784 }, { "name": "model.layers.27.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21755904 }, { "name": "model.layers.27.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21761024 }, { "name": "model.layers.28.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21766144 }, { "name": "model.layers.28.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21771264 }, { "name": "model.layers.29.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21776384 }, { "name": "model.layers.29.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21781504 }, { "name": "model.layers.0.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21786624 }, { "name": "model.layers.0.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21791744 }, { "name": "model.layers.1.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21796864 }, { "name": "model.layers.1.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21801984 }, { "name": "model.norm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 21807104 }, { "name": "model.layers.10.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21812224 }, { "name": "model.layers.11.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21812480 }, { "name": "model.layers.11.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21812992 }, { "name": "model.layers.18.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21813504 }, { "name": "model.layers.19.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21813760 }, { "name": "model.layers.19.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21814272 }, { "name": "model.layers.2.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21814784 }, { "name": "model.layers.12.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21815040 }, { "name": "model.layers.13.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21815296 }, { "name": "model.layers.7.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21815552 }, { "name": "model.layers.7.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21816064 }, { "name": "model.layers.8.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21816576 }, { "name": "model.layers.22.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21816832 }, { "name": "model.layers.23.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21817088 }, { "name": "model.layers.31.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21817600 }, { "name": "model.layers.31.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21818112 }, { "name": "model.layers.4.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21818624 }, { "name": "model.layers.14.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21818880 }, { "name": "model.layers.15.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21819136 }, { "name": "model.layers.15.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21819648 }, { "name": "model.layers.9.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21820160 }, { "name": "model.layers.3.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21820416 }, { "name": "model.layers.3.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21820928 }, { "name": "model.layers.30.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21821440 }, { "name": "model.layers.16.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21821696 }, { "name": "model.layers.17.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21821952 }, { "name": "model.layers.23.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21822208 }, { "name": "model.layers.24.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21822720 }, { "name": "model.layers.25.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21822976 }, { "name": "model.layers.5.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21823232 }, { "name": "model.layers.6.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21823488 }, { "name": "model.layers.20.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21823744 }, { "name": "model.layers.21.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21824000 }, { "name": "model.layers.26.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21824256 }, { "name": "model.layers.27.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21824512 }, { "name": "model.layers.27.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 21825024 }, { "name": "model.layers.28.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21825536 }, { "name": "model.layers.29.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21825792 }, { "name": "model.layers.0.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21826048 }, { "name": "model.layers.1.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 21826304 }, { "name": "model.layers.10.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21826560 }, { "name": "model.layers.12.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21826624 }, { "name": "model.layers.2.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21826688 }, { "name": "model.layers.13.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21826752 }, { "name": "model.layers.14.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21826816 }, { "name": "model.layers.8.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21826880 }, { "name": "model.layers.9.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21826944 }, { "name": "model.layers.22.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827008 }, { "name": "model.layers.4.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827072 }, { "name": "model.layers.5.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827136 }, { "name": "model.layers.16.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827200 }, { "name": "model.layers.30.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827264 }, { "name": "model.layers.17.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827328 }, { "name": "model.layers.18.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827392 }, { "name": "model.layers.24.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827456 }, { "name": "model.layers.25.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827520 }, { "name": "model.layers.6.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827584 }, { "name": "model.layers.20.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827648 }, { "name": "model.layers.21.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827712 }, { "name": "model.layers.26.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827776 }, { "name": "model.layers.28.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827840 }, { "name": "model.layers.29.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827904 }, { "name": "model.layers.0.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21827968 }, { "name": "model.layers.1.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828032 }, { "name": "model.layers.10.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828096 }, { "name": "model.layers.2.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828160 }, { "name": "model.layers.12.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828224 }, { "name": "model.layers.13.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828288 }, { "name": "model.layers.8.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828352 }, { "name": "model.layers.9.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828416 }, { "name": "model.layers.22.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828480 }, { "name": "model.layers.4.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828544 }, { "name": "model.layers.5.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828608 }, { "name": "model.layers.14.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828672 }, { "name": "model.layers.16.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828736 }, { "name": "model.layers.30.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828800 }, { "name": "model.layers.17.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828864 }, { "name": "model.layers.18.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828928 }, { "name": "model.layers.24.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21828992 }, { "name": "model.layers.25.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21829056 }, { "name": "model.layers.6.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21829120 }, { "name": "model.layers.20.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21829184 }, { "name": "model.layers.21.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21829248 }, { "name": "model.layers.26.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21829312 }, { "name": "model.layers.28.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21829376 }, { "name": "model.layers.29.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21829440 }, { "name": "model.layers.0.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21829504 }, { "name": "model.layers.1.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 21829568 } ], "md5sum": "b5e407ecfa7988932cad28f8713722d0" } ] }