diff --git a/added_tokens.json b/added_tokens.json new file mode 100644 index 0000000000000000000000000000000000000000..f6093c86014c94348f18bf42c2e404e6fbde995d --- /dev/null +++ b/added_tokens.json @@ -0,0 +1,7 @@ +{ + "": 2, + "": 1, + "": 0, + "<|im_end|>": 32000, + "<|im_start|>": 32001 +} diff --git a/ndarray-cache.json b/ndarray-cache.json new file mode 100644 index 0000000000000000000000000000000000000000..e7fb4d685ff586fba25578e236b8eda85d7ff0eb --- /dev/null +++ b/ndarray-cache.json @@ -0,0 +1,5611 @@ +{ + "metadata": { + "ParamSize": 453 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 262160384, + "records": [ + { + "name": "param_0", + "shape": [ + 32002, + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262160384, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_7", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_10", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_1", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_2", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_3", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_4", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_5", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_6", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_8", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_9", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_11", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_12", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_13", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_14", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_21", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_24", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_15", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_16", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_17", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_18", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_19", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_20", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_22", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_23", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_25", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_26", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_27", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_28", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_35", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_38", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_29", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_30", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_31", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_32", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_33", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_34", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_36", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_37", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_39", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_40", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_41", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_42", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_49", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_52", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_43", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_44", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_45", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_46", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_47", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_48", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_50", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_51", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_53", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_54", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_55", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_56", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_63", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_66", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_57", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_58", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_59", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_60", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_61", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_62", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_64", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_65", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_67", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_68", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_69", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_70", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_77", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_80", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_71", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_72", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_73", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_74", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_75", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_76", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_78", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_79", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_81", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_82", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_83", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_84", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_91", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_94", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_85", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_86", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_87", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_88", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_89", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_90", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_92", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_93", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_95", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_96", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_97", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_98", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_105", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_108", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_99", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_100", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_101", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_102", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_103", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_104", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_106", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_107", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_109", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_110", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_111", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_112", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_119", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_122", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_113", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_114", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_115", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_116", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_117", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_118", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_120", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_121", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_123", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_124", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_125", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_126", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_133", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_136", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_127", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_128", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_129", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_130", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_131", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_132", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_134", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_135", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_137", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_138", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_139", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_140", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_147", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_150", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_141", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_142", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_143", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_144", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_145", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_146", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_148", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_149", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_151", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_152", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_153", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_154", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_161", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_164", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_155", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_156", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_157", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_158", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_159", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_160", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_162", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_163", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_165", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_166", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_167", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_168", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_175", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_178", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_169", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_170", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_171", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_172", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_173", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_174", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_176", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_177", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_179", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_180", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_181", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_182", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_189", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_192", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_183", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_184", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_185", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_186", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_187", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_188", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_190", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_191", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_193", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_194", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_195", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_196", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_203", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_206", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_197", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_198", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_199", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_200", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_201", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_202", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_204", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_205", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_207", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_208", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_209", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_210", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_217", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_220", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_211", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_212", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_213", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_214", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_215", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_216", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_218", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_219", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_221", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_222", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_223", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_224", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_231", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_234", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_51.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_225", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_226", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_227", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_228", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_229", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_230", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_232", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_233", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_235", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_236", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_237", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_238", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_52.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_245", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_53.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_248", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_54.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_239", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_240", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_241", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_242", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_243", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_244", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_246", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_247", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_249", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_250", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_251", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_252", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_55.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_259", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_56.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_262", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_57.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_253", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_254", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_255", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_256", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_257", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_258", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_260", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_261", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_263", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_264", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_265", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_266", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_58.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_273", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_59.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_276", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_60.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_267", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_268", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_269", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_270", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_271", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_272", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_274", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_275", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_277", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_278", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_279", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_280", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_61.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_287", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_62.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_290", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_63.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_281", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_282", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_283", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_284", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_285", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_286", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_288", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_289", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_291", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_292", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_293", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_294", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_64.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_301", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_65.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_304", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_66.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_295", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_296", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_297", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_298", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_299", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_300", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_302", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_303", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_305", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_306", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_307", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_308", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_67.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_315", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_68.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_318", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_69.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_309", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_310", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_311", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_312", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_313", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_314", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_316", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_317", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_319", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_320", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_321", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_322", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_70.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_329", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_71.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_332", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_72.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_323", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_324", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_325", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_326", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_327", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_328", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_330", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_331", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_333", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_334", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_335", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_336", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_73.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_343", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_74.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_346", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_75.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_337", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_338", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_339", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_340", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_341", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_342", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_344", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_345", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_347", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_348", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_349", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_350", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_76.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_357", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_77.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_360", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_78.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_351", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_352", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_353", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_354", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_355", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_356", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_358", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_359", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_361", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_362", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_363", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_364", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_79.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_371", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_80.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_374", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_81.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_365", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_366", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_367", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_368", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_369", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_370", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_372", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_373", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_375", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_376", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_377", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_378", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_82.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_385", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_83.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_388", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_84.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_379", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_380", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_381", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_382", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_383", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_384", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_386", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_387", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_389", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_390", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_391", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_392", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_85.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_399", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_86.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_402", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_87.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_393", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_394", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_395", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_396", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_397", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_398", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_400", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_401", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_403", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_404", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_405", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_406", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_88.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_413", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_89.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_416", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_90.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_407", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_408", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_409", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_410", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_411", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_412", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_414", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_415", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_417", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_418", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_419", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_420", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_91.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_427", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_92.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_430", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_93.bin", + "format": "raw-shard", + "nbytes": 27803648, + "records": [ + { + "name": "param_421", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_422", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_423", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_424", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_425", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_426", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_428", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_429", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_431", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_432", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_433", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_434", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + } + ] + }, + { + "dataPath": "params_shard_94.bin", + "format": "raw-shard", + "nbytes": 58720256, + "records": [ + { + "name": "param_441", + "shape": [ + 28672, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 58720256, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_95.bin", + "format": "raw-shard", + "nbytes": 29360128, + "records": [ + { + "name": "param_444", + "shape": [ + 4096, + 1792 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 29360128, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_96.bin", + "format": "raw-shard", + "nbytes": 262160384, + "records": [ + { + "name": "param_450", + "shape": [ + 32002, + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262160384, + "byteOffset": 0 + } + ] + }, + { + "dataPath": "params_shard_97.bin", + "format": "raw-shard", + "nbytes": 28860416, + "records": [ + { + "name": "param_435", + "shape": [ + 6144, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "param_436", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "param_437", + "shape": [ + 6144, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 393216, + "byteOffset": 12976128 + }, + { + "name": "param_438", + "shape": [ + 4096, + 512 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 13369344 + }, + { + "name": "param_439", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 21757952 + }, + { + "name": "param_440", + "shape": [ + 4096, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 22020096 + }, + { + "name": "param_442", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 22282240 + }, + { + "name": "param_443", + "shape": [ + 28672, + 32 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1835008, + "byteOffset": 24117248 + }, + { + "name": "param_445", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 25952256 + }, + { + "name": "param_446", + "shape": [ + 4096, + 112 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 917504, + "byteOffset": 26869760 + }, + { + "name": "param_447", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27787264 + }, + { + "name": "param_448", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27795456 + }, + { + "name": "param_449", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 8192, + "byteOffset": 27803648 + }, + { + "name": "param_451", + "shape": [ + 2048, + 128 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 524288, + "byteOffset": 27811840 + }, + { + "name": "param_452", + "shape": [ + 2048, + 128 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 524288, + "byteOffset": 28336128 + } + ] + } + ] +} \ No newline at end of file diff --git a/params_shard_0.bin b/params_shard_0.bin new file mode 100644 index 0000000000000000000000000000000000000000..ab09bf5d862c3616d6e8e8d9047d185af831de71 --- /dev/null +++ b/params_shard_0.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e4a87b8549f176212b09d684ea41ac02e0e11eff7eb708c9a1bf46c39f24d579 +size 262160384 diff --git a/params_shard_1.bin b/params_shard_1.bin new file mode 100644 index 0000000000000000000000000000000000000000..5b5930fdb1d0d93d6f8fbf51d58a75882b2dfc62 --- /dev/null +++ b/params_shard_1.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bec95677d7c3c42be59b2194ee5a58045ac25d68194d3f9c3f5d0a66c5e11271 +size 58720256 diff --git a/params_shard_10.bin b/params_shard_10.bin new file mode 100644 index 0000000000000000000000000000000000000000..5a23b6ba3b0547a89337afac1b7b68e9d556f300 --- /dev/null +++ b/params_shard_10.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fee519319854337662494d1d7bda42aaece98e6997506ebadf3caa3576e7d2bf +size 58720256 diff --git a/params_shard_11.bin b/params_shard_11.bin new file mode 100644 index 0000000000000000000000000000000000000000..f8a60335dae24f1823624d50ef923971ec5d0b5b --- /dev/null +++ b/params_shard_11.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2653df9433d3235f146ed08f27be81515d5cb0d3ebeded4442d457bcacb9557 +size 29360128 diff --git a/params_shard_12.bin b/params_shard_12.bin new file mode 100644 index 0000000000000000000000000000000000000000..5c6ccdbd35008dd2b02cf712f0d95467da4ea5d8 --- /dev/null +++ b/params_shard_12.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7a7f5a3752491f86cb7db10c58ce328f10eaef73da645c56ae3ac6fba7db2b15 +size 27803648 diff --git a/params_shard_13.bin b/params_shard_13.bin new file mode 100644 index 0000000000000000000000000000000000000000..e296904196628b34a3f117cc4f9002e0fd2ec3ad --- /dev/null +++ b/params_shard_13.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9fd02f3c175c8b27d03c03b70e0c5a99272da581f7d7d499a1c63b52a2d7714f +size 58720256 diff --git a/params_shard_14.bin b/params_shard_14.bin new file mode 100644 index 0000000000000000000000000000000000000000..62a9dae6ef16436c4418b43c89015d872b40616f --- /dev/null +++ b/params_shard_14.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:544015729880d212293a4c4f3bd3317b857dfaabc96161f0249e02a7748a5293 +size 29360128 diff --git a/params_shard_15.bin b/params_shard_15.bin new file mode 100644 index 0000000000000000000000000000000000000000..a85c9eb7af0163a94e0c06a1e8089b27bf0ddfbf --- /dev/null +++ b/params_shard_15.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:594432e1b6f7943ce2f3be1f7f132620011f64b33f1263d985116cf4ce733895 +size 27803648 diff --git a/params_shard_16.bin b/params_shard_16.bin new file mode 100644 index 0000000000000000000000000000000000000000..e455db0bee40045825f91345b76508ce2eea7518 --- /dev/null +++ b/params_shard_16.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c03c657034a0168c674335f680a60b5d68dcbb6bc967a2da30c4c7fa368db04e +size 58720256 diff --git a/params_shard_17.bin b/params_shard_17.bin new file mode 100644 index 0000000000000000000000000000000000000000..5e3033f4803038225771ee766c02f58e0c5c493f --- /dev/null +++ b/params_shard_17.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4acb2283b0b4057d8b9486f2a48a0d1ef7cfd5a476c7d7db9b367a2148445b87 +size 29360128 diff --git a/params_shard_18.bin b/params_shard_18.bin new file mode 100644 index 0000000000000000000000000000000000000000..4e035ab76c554115a957f22880282bd513de9f02 --- /dev/null +++ b/params_shard_18.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b656588a752c6fb8eeb91a3309dd626ed940162be854abc45a6b4008e96fa77 +size 27803648 diff --git a/params_shard_19.bin b/params_shard_19.bin new file mode 100644 index 0000000000000000000000000000000000000000..b39776f51bcc6a6c0f76b6bd8d26f18008870ceb --- /dev/null +++ b/params_shard_19.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a82467d506ebc76cea9c930ef9d4b7bf468740ef05f991847bab107b1194bad1 +size 58720256 diff --git a/params_shard_2.bin b/params_shard_2.bin new file mode 100644 index 0000000000000000000000000000000000000000..bc60e932462b549da2e8fc8e04f06cbafad396b7 --- /dev/null +++ b/params_shard_2.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:93f1d839775173ed3ebdbea06dd8d9af1d0430698c1dc50f78f5e54d8f11566f +size 29360128 diff --git a/params_shard_20.bin b/params_shard_20.bin new file mode 100644 index 0000000000000000000000000000000000000000..d5d8ae68453e3b82c1026f1f868387bc014bf073 --- /dev/null +++ b/params_shard_20.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bf081917c8df945b03a9c19ee29d6c31c41f8add164759239de800a7d9da841f +size 29360128 diff --git a/params_shard_21.bin b/params_shard_21.bin new file mode 100644 index 0000000000000000000000000000000000000000..4f0b728a97003b6a68046d7e3b4f5584d1a4b9ec --- /dev/null +++ b/params_shard_21.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:357282645abcb5bbe1a9e82e7de5303c6572cf1a0c9ff3621c73f241208de1fb +size 27803648 diff --git a/params_shard_22.bin b/params_shard_22.bin new file mode 100644 index 0000000000000000000000000000000000000000..5005e64be2d4a07de0d5404fd5fefe9760f529cb --- /dev/null +++ b/params_shard_22.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7ab72253941c84ca6c2470ac9455a2b914d4884905878bbe0d183f74b2081016 +size 58720256 diff --git a/params_shard_23.bin b/params_shard_23.bin new file mode 100644 index 0000000000000000000000000000000000000000..d0b0a0c712d08a5c7c7d21e0a2dc6413d46d750d --- /dev/null +++ b/params_shard_23.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f3ac17bad2e008ea1ce7e2bda31bcd412fe2316e77d06a056ad8633d08137e92 +size 29360128 diff --git a/params_shard_24.bin b/params_shard_24.bin new file mode 100644 index 0000000000000000000000000000000000000000..b9a7688a7ff8bd739adec67865a27e79539e163e --- /dev/null +++ b/params_shard_24.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:76ca896266dd8c3032863d55f0c588247c9d50573255984531df55458aa7e330 +size 27803648 diff --git a/params_shard_25.bin b/params_shard_25.bin new file mode 100644 index 0000000000000000000000000000000000000000..d88bf1ab383aee9d30f626fecb081fdf961cb37e --- /dev/null +++ b/params_shard_25.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:976fd523669d64a995b04ed6c8b467ea240f7aad09c8d80731287aa47ba14fd4 +size 58720256 diff --git a/params_shard_26.bin b/params_shard_26.bin new file mode 100644 index 0000000000000000000000000000000000000000..b4a893c3b59ce273e53700234a5a125ed0f10633 --- /dev/null +++ b/params_shard_26.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:387d271b10545142a09ce2737a3fee533f62eb0eb0bed178cdee694176a8aee4 +size 29360128 diff --git a/params_shard_27.bin b/params_shard_27.bin new file mode 100644 index 0000000000000000000000000000000000000000..6c2ac767a38e493b6efd3d4eb8cb2bb223e67465 --- /dev/null +++ b/params_shard_27.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:654b2a9d4bac17f094e21a3c86694826c74f26bbe9903f655c991808275584f2 +size 27803648 diff --git a/params_shard_28.bin b/params_shard_28.bin new file mode 100644 index 0000000000000000000000000000000000000000..633ebf5a9cb091092110baf72d83cc3fd56fe12e --- /dev/null +++ b/params_shard_28.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5df6198fdc76fc85cf8df3679f97cb8a3a62bdda6ad71798fbb994e48ed7b3ea +size 58720256 diff --git a/params_shard_29.bin b/params_shard_29.bin new file mode 100644 index 0000000000000000000000000000000000000000..b0b5db2a5c1bc9dadcc7689a2d243e033cf4af1d --- /dev/null +++ b/params_shard_29.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b39fc324917234738d87efe16511a4917e84f2c9e126a6023ab78a55718c38d +size 29360128 diff --git a/params_shard_3.bin b/params_shard_3.bin new file mode 100644 index 0000000000000000000000000000000000000000..b0b7276aa13cf72a0d50dc6503d30c4dc1e22193 --- /dev/null +++ b/params_shard_3.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0c10eacb5c17fe44f3de7b453083aaca14938a5c57e6aedcea2c488e4b52c200 +size 27803648 diff --git a/params_shard_30.bin b/params_shard_30.bin new file mode 100644 index 0000000000000000000000000000000000000000..f46cae0f7b1b5012a69371ffd6a509542879eba0 --- /dev/null +++ b/params_shard_30.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3624068b873e0bd6e63199dadf92e7c3803816a1090d923954fd45ec7d6f52c6 +size 27803648 diff --git a/params_shard_31.bin b/params_shard_31.bin new file mode 100644 index 0000000000000000000000000000000000000000..a5d6461e21298fcb9b22b6fe1ffde5d3a7a3f364 --- /dev/null +++ b/params_shard_31.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:090509c41ff889ab42bf276cb273c6503668b6fb59e47ae4456f825d84e8e339 +size 58720256 diff --git a/params_shard_32.bin b/params_shard_32.bin new file mode 100644 index 0000000000000000000000000000000000000000..e92c715864ae71091008e74df19b24ba0e29df41 --- /dev/null +++ b/params_shard_32.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8c45577f8e9e7927a7fa444ff16282b3cc55a913c57d904cd864869d0eb8b5d8 +size 29360128 diff --git a/params_shard_33.bin b/params_shard_33.bin new file mode 100644 index 0000000000000000000000000000000000000000..0ccb2bfccbabfb0a3381d55962c473c0281442ac --- /dev/null +++ b/params_shard_33.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e91bb5459c2197d4face9f99943742e527b0297fa00c9a8d99f0929aa0595048 +size 27803648 diff --git a/params_shard_34.bin b/params_shard_34.bin new file mode 100644 index 0000000000000000000000000000000000000000..6500a06907c21354777855277746ca4566e8b8c7 --- /dev/null +++ b/params_shard_34.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5bb694cc4f6d65025a9184de59cf1beaee9b3d9c55ffeeeb89ec375efd1d3f24 +size 58720256 diff --git a/params_shard_35.bin b/params_shard_35.bin new file mode 100644 index 0000000000000000000000000000000000000000..f49f0f9b02d1c1017bd24000c7a3cb6419bf1d13 --- /dev/null +++ b/params_shard_35.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c2509c19f66912360afec22beb07100389fd26fa0460fd71016489e967b2a4e +size 29360128 diff --git a/params_shard_36.bin b/params_shard_36.bin new file mode 100644 index 0000000000000000000000000000000000000000..fa1eeb983bbe82f9ac457cde839de94e068e17c5 --- /dev/null +++ b/params_shard_36.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b01c7d972a18423ccc4d659a1e0a8a4d026019607ef71d807e085e80c9e90c6 +size 27803648 diff --git a/params_shard_37.bin b/params_shard_37.bin new file mode 100644 index 0000000000000000000000000000000000000000..6e06d8b1b671dd1bcc48cd4f854e812f232b2068 --- /dev/null +++ b/params_shard_37.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dd5845a5880186ad1904e356c04327340fbc7f7037c90176f5cd985d05285b02 +size 58720256 diff --git a/params_shard_38.bin b/params_shard_38.bin new file mode 100644 index 0000000000000000000000000000000000000000..6f42953ad645daf3aad1f7f47d134568d26ac815 --- /dev/null +++ b/params_shard_38.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d5e7429bad0b4cd4c4068adad8b4c417eb294e39262272815c89049312005f2c +size 29360128 diff --git a/params_shard_39.bin b/params_shard_39.bin new file mode 100644 index 0000000000000000000000000000000000000000..55e7ba5633d7fdcfe55ea75c0075805b5ea5df2f --- /dev/null +++ b/params_shard_39.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ecc949600506a2fb73021b8fc842b4743e03a6f5bcfecd5b06849950bf44d4be +size 27803648 diff --git a/params_shard_4.bin b/params_shard_4.bin new file mode 100644 index 0000000000000000000000000000000000000000..6fc85aa48ca921f73d07569b826583c8d95d687c --- /dev/null +++ b/params_shard_4.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e0b50412968302829ca72630561f272391da5e5885de191782d300916269d766 +size 58720256 diff --git a/params_shard_40.bin b/params_shard_40.bin new file mode 100644 index 0000000000000000000000000000000000000000..321a4610fa8eb98885ad300750eef8497a1951cf --- /dev/null +++ b/params_shard_40.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea7f4b8e77afa2aed756903fce3ec306479292e3dce89d34a06b9a7c44ab004a +size 58720256 diff --git a/params_shard_41.bin b/params_shard_41.bin new file mode 100644 index 0000000000000000000000000000000000000000..8b5cbe76a151be708a4e6aabe7fc94ff99182491 --- /dev/null +++ b/params_shard_41.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b562c232fb9b30dd5da00a09d33b5b552a6f6fab0a4144d4ab74eccaaae12d99 +size 29360128 diff --git a/params_shard_42.bin b/params_shard_42.bin new file mode 100644 index 0000000000000000000000000000000000000000..4044991772b17b6999deeb83492b64a6c2778550 --- /dev/null +++ b/params_shard_42.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d3be92c5853a60db384da673bf478bc093038b52eb3a98e205d9f54224d55d37 +size 27803648 diff --git a/params_shard_43.bin b/params_shard_43.bin new file mode 100644 index 0000000000000000000000000000000000000000..934ccf42b5061b253eb15f0d7be8e4ac623d1507 --- /dev/null +++ b/params_shard_43.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f7f440eb74d4a64e7d9e0df1432a182540c5ff983a4c7b8d5a643287448f16d5 +size 58720256 diff --git a/params_shard_44.bin b/params_shard_44.bin new file mode 100644 index 0000000000000000000000000000000000000000..ca435bac61f2ed18ad4f3da620eac26a23d78488 --- /dev/null +++ b/params_shard_44.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a6954d63f876cb22e8fd55ff2290a8613f63577f45fc147f42fe85bea75c94f +size 29360128 diff --git a/params_shard_45.bin b/params_shard_45.bin new file mode 100644 index 0000000000000000000000000000000000000000..ba274af863bebec980fcacae38d9287fa65bedf4 --- /dev/null +++ b/params_shard_45.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c5721ccbe69b4bf30a766538225434754cbc7e6a321ab018ebdfa3704f375f01 +size 27803648 diff --git a/params_shard_46.bin b/params_shard_46.bin new file mode 100644 index 0000000000000000000000000000000000000000..4b697f81eb18b79df28fd082ebedf8b42ad9cb97 --- /dev/null +++ b/params_shard_46.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:81864bc41b04e69f87831513e5dcd5759071d7373a2d489e2e2ffa106abe1505 +size 58720256 diff --git a/params_shard_47.bin b/params_shard_47.bin new file mode 100644 index 0000000000000000000000000000000000000000..4d5e32a71215e8abd4fa2602b360abdaceeb8d55 --- /dev/null +++ b/params_shard_47.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ce3529246d75ab3525513a5ab18d63a51778945a7f125376f87a387cb44cf196 +size 29360128 diff --git a/params_shard_48.bin b/params_shard_48.bin new file mode 100644 index 0000000000000000000000000000000000000000..5a475cb1bbeabf07c4c6ed635d0bf9684e7b265e --- /dev/null +++ b/params_shard_48.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f621ac4b865ca1fa0afd7c1e44b07b262e066a225f6146d3b279d452682d3527 +size 27803648 diff --git a/params_shard_49.bin b/params_shard_49.bin new file mode 100644 index 0000000000000000000000000000000000000000..4e59aaeff467590799846b83129c2841e5509eaa --- /dev/null +++ b/params_shard_49.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:744c1f053e7280b2b358e089ec6a2e005b8cee93d0e0cf85506114393e76b402 +size 58720256 diff --git a/params_shard_5.bin b/params_shard_5.bin new file mode 100644 index 0000000000000000000000000000000000000000..eca3f7ec1acd5ecdf5848538ae8edd49d025ad89 --- /dev/null +++ b/params_shard_5.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4d0f3b6d7213d1e4aee481eb0c889a1f93596ffb97462d8816a29bdd2cdf037c +size 29360128 diff --git a/params_shard_50.bin b/params_shard_50.bin new file mode 100644 index 0000000000000000000000000000000000000000..9d4e152e212ff7766f102398ac5e1a91accf24c8 --- /dev/null +++ b/params_shard_50.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5103c12a2fe24d80160cf871ff1d30fd5664b47a71439f7be5b40b61d5ff6799 +size 29360128 diff --git a/params_shard_51.bin b/params_shard_51.bin new file mode 100644 index 0000000000000000000000000000000000000000..bff4cf4e904e19d063e39ea276ed4e6a4f44648c --- /dev/null +++ b/params_shard_51.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e464ba71775d0c46efcf57c1c37cacd296d5dee8a45659e0e6a88bd59729e17c +size 27803648 diff --git a/params_shard_52.bin b/params_shard_52.bin new file mode 100644 index 0000000000000000000000000000000000000000..5fe123cc0d2ddce051cad4f0a5614d1db4ee78d5 --- /dev/null +++ b/params_shard_52.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:836bdb6374da29cf0ad345d88b1e9b2bf97d1361a3370d2f151617a99b989d73 +size 58720256 diff --git a/params_shard_53.bin b/params_shard_53.bin new file mode 100644 index 0000000000000000000000000000000000000000..447879240b6da7f4c68a4c711b072c0c9978d6b1 --- /dev/null +++ b/params_shard_53.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6225141732be89ff8793a9f1db396c734ec01b30ed2a20b8e69dc40950dbb44a +size 29360128 diff --git a/params_shard_54.bin b/params_shard_54.bin new file mode 100644 index 0000000000000000000000000000000000000000..f9d28854357ff7a4e891a850c80f074d29ef2327 --- /dev/null +++ b/params_shard_54.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:060afe5bb7a78a7667f62194f3008c393b7a51cf06752cf167bdc2df8ce7c850 +size 27803648 diff --git a/params_shard_55.bin b/params_shard_55.bin new file mode 100644 index 0000000000000000000000000000000000000000..20a42d447fc458ae72e1712b029280e4a347594d --- /dev/null +++ b/params_shard_55.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1b02f9dab7f7a5049f1494c14752e60a17b90ea1501935a44811c72cb9f2540b +size 58720256 diff --git a/params_shard_56.bin b/params_shard_56.bin new file mode 100644 index 0000000000000000000000000000000000000000..1b867f6499367f066d76e4f357642751c012fb3d --- /dev/null +++ b/params_shard_56.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:24bc472825c5f849b0baafe206df1e7b1a6b1c9e513a076c3236f40680ac1411 +size 29360128 diff --git a/params_shard_57.bin b/params_shard_57.bin new file mode 100644 index 0000000000000000000000000000000000000000..082054a9464276a0f09eccad6500d45decb3e92d --- /dev/null +++ b/params_shard_57.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fd703b24163b2ee1a268f4356cf8e27816f64584088a3afb7a0bd743524439f2 +size 27803648 diff --git a/params_shard_58.bin b/params_shard_58.bin new file mode 100644 index 0000000000000000000000000000000000000000..a6191a1d441a22b4c1cf21ead46d850c37ac34ea --- /dev/null +++ b/params_shard_58.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b0b82d1129df2d11fd638312ca3d389e7d8cc03d62d46cf5b7d5d05660d10461 +size 58720256 diff --git a/params_shard_59.bin b/params_shard_59.bin new file mode 100644 index 0000000000000000000000000000000000000000..98b8a4c457cf199dbc13e36e26d1db86249a89db --- /dev/null +++ b/params_shard_59.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4bc36c5e0d6b33df7ab053a2b566a0f4f528c7a7f7ece740771d07a0e6f3b123 +size 29360128 diff --git a/params_shard_6.bin b/params_shard_6.bin new file mode 100644 index 0000000000000000000000000000000000000000..67a1692ac984c6ab36d9d9bb0e0738f2d44fea6c --- /dev/null +++ b/params_shard_6.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:206bf7456388fce7a2024d5c53789b988ba3f7bf1cad0f513cabbca81af09cc2 +size 27803648 diff --git a/params_shard_60.bin b/params_shard_60.bin new file mode 100644 index 0000000000000000000000000000000000000000..544b36ecba8abf107a122accdad6d49028a6b9c4 --- /dev/null +++ b/params_shard_60.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c084a365123a4e14deaa11eb5acc3d88ecbec5a8ef4b07d23dc2e9653ed9e0cb +size 27803648 diff --git a/params_shard_61.bin b/params_shard_61.bin new file mode 100644 index 0000000000000000000000000000000000000000..36c3b96a9825e077a0285a42b4dc21bcb8d8f3eb --- /dev/null +++ b/params_shard_61.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18d3073db767b636cde94c77a3e8572cf09eed7bf271562e64f2cff495b3529d +size 58720256 diff --git a/params_shard_62.bin b/params_shard_62.bin new file mode 100644 index 0000000000000000000000000000000000000000..4f8f310f14f394ac4946be43bcc10ceafdc176ce --- /dev/null +++ b/params_shard_62.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:790bd91e23e45546b673f229a76c2f6c039ecfbffe214d51dfa34f99dc9ca5a5 +size 29360128 diff --git a/params_shard_63.bin b/params_shard_63.bin new file mode 100644 index 0000000000000000000000000000000000000000..8d3ae72851bec8bfecbafd72fae0926a1f3815f3 --- /dev/null +++ b/params_shard_63.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2df2b939aba2a17c48240d338db621ecd6fd05cd93771591a983fc7214d15725 +size 27803648 diff --git a/params_shard_64.bin b/params_shard_64.bin new file mode 100644 index 0000000000000000000000000000000000000000..c53c1c98d14232c71c1ecbdc4f5ac40df6eb1a97 --- /dev/null +++ b/params_shard_64.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:683de80eb659ebef86885638f4dfe3ad7afd33600b75663c7b773af7bd576be6 +size 58720256 diff --git a/params_shard_65.bin b/params_shard_65.bin new file mode 100644 index 0000000000000000000000000000000000000000..76b8b12b3c23c68cec13fdc7ac03c9c4a9bc83f0 --- /dev/null +++ b/params_shard_65.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:35a075ab912e1c68f92eda21b6328e9810cea220646a77beb7f0650a24faf8f8 +size 29360128 diff --git a/params_shard_66.bin b/params_shard_66.bin new file mode 100644 index 0000000000000000000000000000000000000000..d6bad8fe3f675193075cb7621e31a4ac225bf38b --- /dev/null +++ b/params_shard_66.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c72c62b49147a54f30abba371127ef84785922bd732ea01c64fe235ef89942a6 +size 27803648 diff --git a/params_shard_67.bin b/params_shard_67.bin new file mode 100644 index 0000000000000000000000000000000000000000..1175ec88aca80b2655cf7aef1282883313f9aaff --- /dev/null +++ b/params_shard_67.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e0d4ab2bfb9c20654c9d8489edc4739c189f67d8dd8d426e5c6e5ec9b8ef9874 +size 58720256 diff --git a/params_shard_68.bin b/params_shard_68.bin new file mode 100644 index 0000000000000000000000000000000000000000..a625fbc7122102a815ce952ee08c6034583a5072 --- /dev/null +++ b/params_shard_68.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:893f2f33f8e609e8a983b7487e87a0bd94612de71edda16517e6722ccab8705c +size 29360128 diff --git a/params_shard_69.bin b/params_shard_69.bin new file mode 100644 index 0000000000000000000000000000000000000000..b035fd4de89709cbed6d10dfe976953eb8329730 --- /dev/null +++ b/params_shard_69.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ebba388f38078dd643919006acea488d848d0b3d37a24906b2db6c0fd128fcd4 +size 27803648 diff --git a/params_shard_7.bin b/params_shard_7.bin new file mode 100644 index 0000000000000000000000000000000000000000..e4d28a9024028d7e19d4634792756ce926f95788 --- /dev/null +++ b/params_shard_7.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6266b07116f3a52dd859478010231a43a240696ff64968b12f8dfdd081924891 +size 58720256 diff --git a/params_shard_70.bin b/params_shard_70.bin new file mode 100644 index 0000000000000000000000000000000000000000..42719c6b40dd4fb702ad6074a1e9077349fb8667 --- /dev/null +++ b/params_shard_70.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9ac0516b5017c12f58ee7ec7bf5553e84cc69da50fe4b4695ffb82e7b8714029 +size 58720256 diff --git a/params_shard_71.bin b/params_shard_71.bin new file mode 100644 index 0000000000000000000000000000000000000000..5208f9e1a20bba8725bdd8267d409312c2fc90b5 --- /dev/null +++ b/params_shard_71.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cac68bb5a132574bb12645332b70e5b599ddf4e1d95f755b925b2823b3496ace +size 29360128 diff --git a/params_shard_72.bin b/params_shard_72.bin new file mode 100644 index 0000000000000000000000000000000000000000..4e3fe8660a309eaf18cae9a2f1ded81db4b18b08 --- /dev/null +++ b/params_shard_72.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d56e2d8e16b55ab12e9bd4f35b9f0ff5b9afd74f252f8ee7019ab1c6e391583a +size 27803648 diff --git a/params_shard_73.bin b/params_shard_73.bin new file mode 100644 index 0000000000000000000000000000000000000000..44a69bdd834798e94c6a1d4e8e97e33615a77b28 --- /dev/null +++ b/params_shard_73.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:339c98111050f7a1314c5b1d9bb99a9cda8b38af1fadba471d0541e4bab4ad27 +size 58720256 diff --git a/params_shard_74.bin b/params_shard_74.bin new file mode 100644 index 0000000000000000000000000000000000000000..07a590dd0c7e0cb9747a3996e570c510225f40f2 --- /dev/null +++ b/params_shard_74.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:feebe8d50898253e6bb14c728ba1b323bcc708312b8fcfd1521d8bce75996d45 +size 29360128 diff --git a/params_shard_75.bin b/params_shard_75.bin new file mode 100644 index 0000000000000000000000000000000000000000..740d2eab1235d4490b1203cedc732ef8d40585f9 --- /dev/null +++ b/params_shard_75.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c53504e63ba44b8ea42bab42fc0344649eadeb89da0a7686da3e3ffbc4c26b13 +size 27803648 diff --git a/params_shard_76.bin b/params_shard_76.bin new file mode 100644 index 0000000000000000000000000000000000000000..10e8c39cd6be25f5cb98dd9ef8f7b6bb008e832a --- /dev/null +++ b/params_shard_76.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ab65cada22d63a48829437f231aafaa7e527b96c073642151156db494d1b218 +size 58720256 diff --git a/params_shard_77.bin b/params_shard_77.bin new file mode 100644 index 0000000000000000000000000000000000000000..2e69a00081fbb44f6473aacac3aa893ba452ca11 --- /dev/null +++ b/params_shard_77.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b53b386e28eb516f892ec2e6c34da741479e5af332ad1c9613132c69edd17531 +size 29360128 diff --git a/params_shard_78.bin b/params_shard_78.bin new file mode 100644 index 0000000000000000000000000000000000000000..743bc79da232c57fe337b31433b34acd76e7bebd --- /dev/null +++ b/params_shard_78.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d65df3e176595bf9594bc7c6ea9be4c3e9454efebf08ccd428d3f27b326d920d +size 27803648 diff --git a/params_shard_79.bin b/params_shard_79.bin new file mode 100644 index 0000000000000000000000000000000000000000..8cc1e830302e032b6821acabac7c8b0a71e1ad53 --- /dev/null +++ b/params_shard_79.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6d13fab9552ff0f75a8ad4191129fdc49251546326b9f4efa06ba1a3e31cd74f +size 58720256 diff --git a/params_shard_8.bin b/params_shard_8.bin new file mode 100644 index 0000000000000000000000000000000000000000..7e8cefc4d690f024e7865a3379e50ab9072a268c --- /dev/null +++ b/params_shard_8.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6fe1dee6469569d30ea3d58d0d33e79105c94e91bcff0703bfb65cd01ec3d90 +size 29360128 diff --git a/params_shard_80.bin b/params_shard_80.bin new file mode 100644 index 0000000000000000000000000000000000000000..42a1347d1b1a8485fe50e820cf94da8af3a920c1 --- /dev/null +++ b/params_shard_80.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a26a1cc2f1f703b301cc16b2b1cce41335fdf28e6ac144c91b9949fa16bcf457 +size 29360128 diff --git a/params_shard_81.bin b/params_shard_81.bin new file mode 100644 index 0000000000000000000000000000000000000000..f2d034b02bbce7c849b237594764dca083a9594c --- /dev/null +++ b/params_shard_81.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e6d7fa49d7fd1a4fbdeda2389748669349b1ef6c5d97c0f9379466b8982344a +size 27803648 diff --git a/params_shard_82.bin b/params_shard_82.bin new file mode 100644 index 0000000000000000000000000000000000000000..bd51bfdb0eb74bd37099f3d8f812efa275f74d33 --- /dev/null +++ b/params_shard_82.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de0899da68ed31e9d10c32c2c66765480f7c7e808a02c5e4eeeb685bf4064286 +size 58720256 diff --git a/params_shard_83.bin b/params_shard_83.bin new file mode 100644 index 0000000000000000000000000000000000000000..c4744995ce6a581ea029ce6872a5c69ae53de00f --- /dev/null +++ b/params_shard_83.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:896a1da08539489ef787cc6f3fd6a00a601b251bf36f32f1cbe872ce159f417c +size 29360128 diff --git a/params_shard_84.bin b/params_shard_84.bin new file mode 100644 index 0000000000000000000000000000000000000000..59026295c9dd389d6ca5ffdec2edfa7f62c4d21c --- /dev/null +++ b/params_shard_84.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9057bdacc0c7eeccf286901d44e329db3c6b269df33515893af236ccb240732b +size 27803648 diff --git a/params_shard_85.bin b/params_shard_85.bin new file mode 100644 index 0000000000000000000000000000000000000000..3bd9a4c9dafdc83767be95454d3d73423d0d34eb --- /dev/null +++ b/params_shard_85.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1c3ff171abb8261f99837c8d22eccf5783be62224861e7ba3824ccafe52e85f +size 58720256 diff --git a/params_shard_86.bin b/params_shard_86.bin new file mode 100644 index 0000000000000000000000000000000000000000..dd7e9e586a519943e05b540e64526081c66e1903 --- /dev/null +++ b/params_shard_86.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:978a13671c2d7458ea1edc33ad0f23f93e792dd47da6b8c21f3b613e8d53f3b9 +size 29360128 diff --git a/params_shard_87.bin b/params_shard_87.bin new file mode 100644 index 0000000000000000000000000000000000000000..36318de0e4107d07c80377b6c7aeb1e34ec93f5a --- /dev/null +++ b/params_shard_87.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:465fe94f14e1438f1b85111c3c289c87f3ee23b1e2efbed723b5604041b785d2 +size 27803648 diff --git a/params_shard_88.bin b/params_shard_88.bin new file mode 100644 index 0000000000000000000000000000000000000000..c8f1bd36de6992f63bbcd8440e06995d91f4e902 --- /dev/null +++ b/params_shard_88.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:15d3b0ce77a4cdc6c5e87eeececda312f40e06be15fd1b091af0bf9b6f876aaa +size 58720256 diff --git a/params_shard_89.bin b/params_shard_89.bin new file mode 100644 index 0000000000000000000000000000000000000000..954d7861c0ca8ae5046b4805a5b545ace802f4a8 --- /dev/null +++ b/params_shard_89.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9a75ff2d8085652a15651f77967a4cf159f1601617b9112fe725f8bc4ba052d +size 29360128 diff --git a/params_shard_9.bin b/params_shard_9.bin new file mode 100644 index 0000000000000000000000000000000000000000..f4a6a7c1aa5f0b1603d045af3b18b8af2ab8dcaa --- /dev/null +++ b/params_shard_9.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87f3343d2f459bbd54cef4656901cae2be0864bb0959eb4718cfa0fec3e6ba19 +size 27803648 diff --git a/params_shard_90.bin b/params_shard_90.bin new file mode 100644 index 0000000000000000000000000000000000000000..dc5eeb9aabf333772fc8233fd08ccc55432b1d3b --- /dev/null +++ b/params_shard_90.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f1755cda92e91f510a186b68d10b51dd07c3903628dbaacea84b559403fb8b1 +size 27803648 diff --git a/params_shard_91.bin b/params_shard_91.bin new file mode 100644 index 0000000000000000000000000000000000000000..76deae500e3b12cb3901f8e5823abdf10bb54561 --- /dev/null +++ b/params_shard_91.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1e64a1e019bc4862535e7512f5238118c8d44b7eb4e8cc9a6396d3eb3fce1a62 +size 58720256 diff --git a/params_shard_92.bin b/params_shard_92.bin new file mode 100644 index 0000000000000000000000000000000000000000..57628bc4212fdd709ddceb1502c9824f506b287b --- /dev/null +++ b/params_shard_92.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6881377a87bf7adeb73126d45a0187b8a1ff612e96dde502258a8caeb5b402df +size 29360128 diff --git a/params_shard_93.bin b/params_shard_93.bin new file mode 100644 index 0000000000000000000000000000000000000000..ddf7f8a9d988aa1a5434e6d2bc85f49525a25df0 --- /dev/null +++ b/params_shard_93.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b4c943963f91d5c20888fece63b335d10c56fa57dcd15551d32b0f33f017e2b +size 27803648 diff --git a/params_shard_94.bin b/params_shard_94.bin new file mode 100644 index 0000000000000000000000000000000000000000..b5bb7ef0644054e3b8cf26d44cb3c2e16883bee2 --- /dev/null +++ b/params_shard_94.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:75c229ff693f73e169c64d8f1ae4468673e554299bb03a2fde16aa7456b8ee67 +size 58720256 diff --git a/params_shard_95.bin b/params_shard_95.bin new file mode 100644 index 0000000000000000000000000000000000000000..58c10bf8dfaf6efd1939f53098e01d8db5fb3fbb --- /dev/null +++ b/params_shard_95.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7bfc76530d2cf44b4679cdc18fe9d6d5fe30194c58aa782e2f9b50fd2e998a5c +size 29360128 diff --git a/params_shard_96.bin b/params_shard_96.bin new file mode 100644 index 0000000000000000000000000000000000000000..edee300d65e36d3b9626e949995c2cf7f3906050 --- /dev/null +++ b/params_shard_96.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f4040a16dbab7cab694eb85995efd03602d1ad9b03a7ef2d37953586780ce848 +size 262160384 diff --git a/params_shard_97.bin b/params_shard_97.bin new file mode 100644 index 0000000000000000000000000000000000000000..d3164eeeb312e8a0bab4d9458440e127997c5fff --- /dev/null +++ b/params_shard_97.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2a47555a3c3f7e533c302c4f2c923c50130fff6193783bbdf71b5b3628172bf0 +size 28860416 diff --git a/private-llm-config.json b/private-llm-config.json new file mode 100644 index 0000000000000000000000000000000000000000..83fd4ad4adca2c0808e46035cf4a0d4b75bb2bd6 --- /dev/null +++ b/private-llm-config.json @@ -0,0 +1,21 @@ +{ + "model_lib": "OpenHermes-2-Mistral-7B-w4a16g128asym", + "local_id": "OpenHermes-2-Mistral-7B-w4a16g128asym", + "conv_template": "mistral_default", + "temperature": 0.7, + "repetition_penalty": 1.0, + "top_p": 0.95, + "mean_gen_len": 128, + "max_gen_len": 512, + "num_shards": 1, + "shift_fill_factor": 0.3, + "tokenizer_files": [ + "added_tokens.json", + "tokenizer.model" + ], + "model_category": "mistral", + "model_name": "OpenHermes-2-Mistral-7B", + "vocab_size": 32002, + "sliding_window": 4096, + "chunk_size": 4096 +} \ No newline at end of file diff --git a/tokenizer.model b/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..8b443ef19c2a19acc3ac64fb9c3db4a72921dff6 --- /dev/null +++ b/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055 +size 493443 diff --git a/tokenizer_config.json b/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..2f1b059ec8dbdafbdbc27dd345ca5526cf84bb14 --- /dev/null +++ b/tokenizer_config.json @@ -0,0 +1,65 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "32000": { + "content": "<|im_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "32001": { + "content": "<|im_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + } + }, + "additional_special_tokens": [ + "", + "", + "" + ], + "bos_token": "", + "clean_up_tokenization_spaces": false, + "eos_token": "<|im_end|>", + "legacy": false, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "tokenizer_class": "LlamaTokenizer", + "tokenizer_file": null, + "trust_remote_code": false, + "unk_token": "", + "use_default_system_prompt": true, + "use_fast": true +}