diff --git a/added_tokens.json b/added_tokens.json new file mode 100644 index 0000000000000000000000000000000000000000..cbce74e5c64b97114098962fa58454a57d7fb532 --- /dev/null +++ b/added_tokens.json @@ -0,0 +1,5 @@ +{ + "": 2, + "": 1, + "": 0 +} diff --git a/ndarray-cache.json b/ndarray-cache.json new file mode 100644 index 0000000000000000000000000000000000000000..fd0fd9b733259e8e472c0f16ce909d8600f2e3ce --- /dev/null +++ b/ndarray-cache.json @@ -0,0 +1,5611 @@ +{ + "metadata": { + "ParamSize": 453 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 262144000, + "records": [ + { + "name": "param_0", + "shape": [ + 32000, + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144000, + "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": 262144000, + "records": [ + { + "name": "param_450", + "shape": [ + 32000, + 4096 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144000, + "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..2ca93e3e3d01bcf9bd38699056091bd6e657d803 --- /dev/null +++ b/params_shard_0.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cb2d0a34f6a78534d08505f9d7f32dd42499a5c160fb793c003fa0bb35cf88a4 +size 262144000 diff --git a/params_shard_1.bin b/params_shard_1.bin new file mode 100644 index 0000000000000000000000000000000000000000..761177818f55954814c7c612b73782dcef5bbe5b --- /dev/null +++ b/params_shard_1.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:14ba564791d74b5f0314caf71660d9463607fe9ad4d0566b3ddd6008d8c8190f +size 58720256 diff --git a/params_shard_10.bin b/params_shard_10.bin new file mode 100644 index 0000000000000000000000000000000000000000..9c602dbb3f3a329500a836ad29a679b70027822a --- /dev/null +++ b/params_shard_10.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:801374e287ca8c020774e40875c51a54dea39ffa78fb62d678cf31119fbbb4a9 +size 58720256 diff --git a/params_shard_11.bin b/params_shard_11.bin new file mode 100644 index 0000000000000000000000000000000000000000..495d75db922910f117043af53d8ebbd2e32a13c5 --- /dev/null +++ b/params_shard_11.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f6c2872c41dd59ab5d8770f87617c4cad1e19c0e2c5d8acc372e9b9d48c066e +size 29360128 diff --git a/params_shard_12.bin b/params_shard_12.bin new file mode 100644 index 0000000000000000000000000000000000000000..151c2ac28be44a9435065b66066c36009c308b31 --- /dev/null +++ b/params_shard_12.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5dfaf9be81377d1bbd53521e7f1ac8a44c8260218332cbb2cba6b7fe27777d35 +size 27803648 diff --git a/params_shard_13.bin b/params_shard_13.bin new file mode 100644 index 0000000000000000000000000000000000000000..839774c2d2a43b74d23e1cd2e26ac5f57881526d --- /dev/null +++ b/params_shard_13.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:12381d06f7feafee5444df6a8315479a4a954984c62c117ccb6563df38a1b99d +size 58720256 diff --git a/params_shard_14.bin b/params_shard_14.bin new file mode 100644 index 0000000000000000000000000000000000000000..d9463f8df38a819316ec9be7014af469e36e737d --- /dev/null +++ b/params_shard_14.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a1a288c85550070042196931e0d95f7ec7c1ff302edc1ab0b1c67606b68b739 +size 29360128 diff --git a/params_shard_15.bin b/params_shard_15.bin new file mode 100644 index 0000000000000000000000000000000000000000..3e61d60a0e79a20f87ee336be5edffe7d9113e26 --- /dev/null +++ b/params_shard_15.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:134788ea7af7f14837aafb79da544f06e096daaa364a33367e81cc3c1bfe3951 +size 27803648 diff --git a/params_shard_16.bin b/params_shard_16.bin new file mode 100644 index 0000000000000000000000000000000000000000..43ce4d846f57c684566c01b241876f53405991ee --- /dev/null +++ b/params_shard_16.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86ce2e5dd6ab186c0b6771f460bed6ebd024ecb3512c7a1e9fa30a8df4266474 +size 58720256 diff --git a/params_shard_17.bin b/params_shard_17.bin new file mode 100644 index 0000000000000000000000000000000000000000..4ffa78f3aba5c206290a2f2362bc1ef244a139eb --- /dev/null +++ b/params_shard_17.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aadf076572209e9ffb1790010f77d7649c6d8f47a4d100eab332c075bbcc10ba +size 29360128 diff --git a/params_shard_18.bin b/params_shard_18.bin new file mode 100644 index 0000000000000000000000000000000000000000..ea9d2b1b18725783e6e7ef28f33624894e381496 --- /dev/null +++ b/params_shard_18.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b83d0259525293d6e60dbf85c3ebddaff6a0565559043d9a0e2b9de9f1a96e84 +size 27803648 diff --git a/params_shard_19.bin b/params_shard_19.bin new file mode 100644 index 0000000000000000000000000000000000000000..357b3d9aadca8be5a4bf2e6286b81a0c13b207cd --- /dev/null +++ b/params_shard_19.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e8f9b0cf2cc78ffc999b22439fe1439f95c606e62d9f7b4ee38f2e20d9c63a26 +size 58720256 diff --git a/params_shard_2.bin b/params_shard_2.bin new file mode 100644 index 0000000000000000000000000000000000000000..cf15da465b6377701138f7eb242fc55704e9d561 --- /dev/null +++ b/params_shard_2.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d77e70292f770da6786f8dc5c8d096798ab4a6b908167994d05b9699cd7e69c6 +size 29360128 diff --git a/params_shard_20.bin b/params_shard_20.bin new file mode 100644 index 0000000000000000000000000000000000000000..1df5a74f21acba18ca544eb490939f98d3d9dd3e --- /dev/null +++ b/params_shard_20.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:baa68db5aa6707b275294701735cefcb0093305ad1df491c511b341fb7e6fdb8 +size 29360128 diff --git a/params_shard_21.bin b/params_shard_21.bin new file mode 100644 index 0000000000000000000000000000000000000000..5d05830cf9277d5f93beef43fbd00068470bf7ea --- /dev/null +++ b/params_shard_21.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f3b699fb646b47830721d707005ed29aed234b1d34f6b7e988e1e24e8ac8406 +size 27803648 diff --git a/params_shard_22.bin b/params_shard_22.bin new file mode 100644 index 0000000000000000000000000000000000000000..9fa21db20ae6ec9620068bfe6d2ea7870793e794 --- /dev/null +++ b/params_shard_22.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:33402fb787a291dae4e6d0dab09a57e98291e5d396b9f925e54dfecd941908ab +size 58720256 diff --git a/params_shard_23.bin b/params_shard_23.bin new file mode 100644 index 0000000000000000000000000000000000000000..c12948d75b1bb141e221c7104eb9de05b23f7b5d --- /dev/null +++ b/params_shard_23.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ad7a51131c8cfe940e679c79689113be2011e46cd2ace8ff55623b11f45b8e2 +size 29360128 diff --git a/params_shard_24.bin b/params_shard_24.bin new file mode 100644 index 0000000000000000000000000000000000000000..7d0ce95c97a6aa0ea8894d888b3a455aad52ae08 --- /dev/null +++ b/params_shard_24.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:33c6d8297dc53490713eb379e3bcde6620c6c9a055a7b5b9541a329e2d2373c1 +size 27803648 diff --git a/params_shard_25.bin b/params_shard_25.bin new file mode 100644 index 0000000000000000000000000000000000000000..afebacc5f91cca5a9cb93d8b5e92cd06ffbe5353 --- /dev/null +++ b/params_shard_25.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a771b5c725a1b9b7cea1b9827b2d3744132db69f002680429cabcdc15ff44be +size 58720256 diff --git a/params_shard_26.bin b/params_shard_26.bin new file mode 100644 index 0000000000000000000000000000000000000000..2e542088db90a423b1aa884396b5e89c8b9939f1 --- /dev/null +++ b/params_shard_26.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6232c1b9a56f7f6a4d05c8d9cb8d4d34c2e18752858021ea5c01ca63365a669e +size 29360128 diff --git a/params_shard_27.bin b/params_shard_27.bin new file mode 100644 index 0000000000000000000000000000000000000000..9e0b93873c690c885dcb844d015b3efb566d165f --- /dev/null +++ b/params_shard_27.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8cdc8524ba4c76218802078ae778c4a6272b338b2307cc971382d1b5d620c35b +size 27803648 diff --git a/params_shard_28.bin b/params_shard_28.bin new file mode 100644 index 0000000000000000000000000000000000000000..8d3de2659325e55db96c816d84a0cd3da1649a8a --- /dev/null +++ b/params_shard_28.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea97b0179f0126ca3424511b9c0a86e97cc38d4c51a6ead91ab52fb2dec66fe9 +size 58720256 diff --git a/params_shard_29.bin b/params_shard_29.bin new file mode 100644 index 0000000000000000000000000000000000000000..97fdbb04d64c0b2b682f2b34ab69f80499f08c80 --- /dev/null +++ b/params_shard_29.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f82a72a4e205ee7b142ffe5be10be2376a6b4791a16ed15c437fea5977bf7d00 +size 29360128 diff --git a/params_shard_3.bin b/params_shard_3.bin new file mode 100644 index 0000000000000000000000000000000000000000..585a4a1ac66e40043cb1747424a574e7fc2eddb9 --- /dev/null +++ b/params_shard_3.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:06185e9a9c23033d69a882bcbdaaa79834beec6747e076a1c533138687caa793 +size 27803648 diff --git a/params_shard_30.bin b/params_shard_30.bin new file mode 100644 index 0000000000000000000000000000000000000000..ba2839f861e96f0032be98a606f46848e37894e7 --- /dev/null +++ b/params_shard_30.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:02ed608d7db9595e2cacbc6a6c90547b28296104f9aa31749dc60b16f872fc4b +size 27803648 diff --git a/params_shard_31.bin b/params_shard_31.bin new file mode 100644 index 0000000000000000000000000000000000000000..a1aebce738d7f754ea4e78a370c154c918523c03 --- /dev/null +++ b/params_shard_31.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a46f7be2afb5464ff0bd7b26e15ddacc5dbeee2bc20b57c64be844ccbe7360a7 +size 58720256 diff --git a/params_shard_32.bin b/params_shard_32.bin new file mode 100644 index 0000000000000000000000000000000000000000..3ad3aa2cb3795efbe0d0b3efedadcfed5469ae20 --- /dev/null +++ b/params_shard_32.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6256358f091dae21504f9966bd3ce2075a36dae2aab660ced93ef9a589b62459 +size 29360128 diff --git a/params_shard_33.bin b/params_shard_33.bin new file mode 100644 index 0000000000000000000000000000000000000000..71b99c912b4f3367ab9472dc93660d443a79be19 --- /dev/null +++ b/params_shard_33.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0c0ba5c71af73a2274e716da236c3b10eebfd03145724fbf52a2afdaba12c3ff +size 27803648 diff --git a/params_shard_34.bin b/params_shard_34.bin new file mode 100644 index 0000000000000000000000000000000000000000..7f62403d1c992788b64a79aac5ca21e7aaed3e73 --- /dev/null +++ b/params_shard_34.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7a42596f9db459244c67c4fa0f2a992cf37358c61b2dc80fdf5c46f79bf2681 +size 58720256 diff --git a/params_shard_35.bin b/params_shard_35.bin new file mode 100644 index 0000000000000000000000000000000000000000..57de85808c77b403fcbe85cdfa2e0e7ce5814aff --- /dev/null +++ b/params_shard_35.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ce5aa0f619ae1f902becc7c1dddd3d75e2ad33861a3136ee2fd02177c7952e62 +size 29360128 diff --git a/params_shard_36.bin b/params_shard_36.bin new file mode 100644 index 0000000000000000000000000000000000000000..864127f7b15a9288721aebdf4b6878fdf3df9928 --- /dev/null +++ b/params_shard_36.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8966ae6f84908d09932577abe6e6300c28ac5448102cc556bea3289cc9df71d2 +size 27803648 diff --git a/params_shard_37.bin b/params_shard_37.bin new file mode 100644 index 0000000000000000000000000000000000000000..7994fe1dab3b3f5e0361fffa32604595d70ca8f0 --- /dev/null +++ b/params_shard_37.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3d35fec12ea117a96c34012e0d6567c893be21a1c28bedd83acd356b55a20245 +size 58720256 diff --git a/params_shard_38.bin b/params_shard_38.bin new file mode 100644 index 0000000000000000000000000000000000000000..2aa3550331e9f447f67842b45075761d72db7f4a --- /dev/null +++ b/params_shard_38.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4010c4cc4b6795da7e02029dc17f3a69409cf2e27c756a3d8e4107f00488241f +size 29360128 diff --git a/params_shard_39.bin b/params_shard_39.bin new file mode 100644 index 0000000000000000000000000000000000000000..147643b9ecfe4858ca1db417e58f8d199e65faac --- /dev/null +++ b/params_shard_39.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:737c749c9efc205a36e6b6067633db8fa07ef0b6de6aedc89bbdf666838d59bc +size 27803648 diff --git a/params_shard_4.bin b/params_shard_4.bin new file mode 100644 index 0000000000000000000000000000000000000000..a3827b9398a82bcb97e5436de6175596c8ec5718 --- /dev/null +++ b/params_shard_4.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e33fd64fa3e24188d8abcc114a43593f2f0fd0ab130f6febd0caf3319d2076e +size 58720256 diff --git a/params_shard_40.bin b/params_shard_40.bin new file mode 100644 index 0000000000000000000000000000000000000000..8a8107be942ed5e426b726899e00ac3d89d117da --- /dev/null +++ b/params_shard_40.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5389779a205c5e4e53e2f226f98707abaef704cc1633e20b645165187ffd5548 +size 58720256 diff --git a/params_shard_41.bin b/params_shard_41.bin new file mode 100644 index 0000000000000000000000000000000000000000..10177130e333b0c5ea29a839c6386d71654d8b04 --- /dev/null +++ b/params_shard_41.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:47cc1838f24af4928d6cd62a08abb4cfb03675f59c4894989a65cd2bc43675fb +size 29360128 diff --git a/params_shard_42.bin b/params_shard_42.bin new file mode 100644 index 0000000000000000000000000000000000000000..5bd9ce2a14475f378e478ccd9d12615c54aca6bc --- /dev/null +++ b/params_shard_42.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a24a21a1c9bc2af5484283d37bd0af618a13f47688572d4ef6784789f8986bca +size 27803648 diff --git a/params_shard_43.bin b/params_shard_43.bin new file mode 100644 index 0000000000000000000000000000000000000000..819dece9d11f84aa01448fa0b841be1e77bc7dd9 --- /dev/null +++ b/params_shard_43.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f201694ef11aa19c634cae1953062f22b23881a1a56ded4d413bf147e658721 +size 58720256 diff --git a/params_shard_44.bin b/params_shard_44.bin new file mode 100644 index 0000000000000000000000000000000000000000..14e18958e42ae6ae98682b56cc564d4a5bdd1cb4 --- /dev/null +++ b/params_shard_44.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f09d58664b70ecaf6f00eed4054527c4e18fc49806215779b8f664493169b51 +size 29360128 diff --git a/params_shard_45.bin b/params_shard_45.bin new file mode 100644 index 0000000000000000000000000000000000000000..6c26c187f20dbafbdd1187f3242cd59bfec6fffd --- /dev/null +++ b/params_shard_45.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b33f5e8169ff2a799590c1c762c780a6d958ed132c0f3f69fa5745e435ccf438 +size 27803648 diff --git a/params_shard_46.bin b/params_shard_46.bin new file mode 100644 index 0000000000000000000000000000000000000000..e90187aa63cda1f9585f1dcccb9f054783cf0533 --- /dev/null +++ b/params_shard_46.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f74e045282eb49109b6328a301fcaf3143fd50f7e2580cbdbcf9cfa90e19999d +size 58720256 diff --git a/params_shard_47.bin b/params_shard_47.bin new file mode 100644 index 0000000000000000000000000000000000000000..085484e9c1dcc67ff541bee31d65f1e9bb6e38ef --- /dev/null +++ b/params_shard_47.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8195d5bd12b7a55562faa0121449f13db515677c087b1d57353e32745bb7bafc +size 29360128 diff --git a/params_shard_48.bin b/params_shard_48.bin new file mode 100644 index 0000000000000000000000000000000000000000..398d0c3a6dc6389d17177aec1e441ccbed328346 --- /dev/null +++ b/params_shard_48.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65a843f4ba0e08a6afc101e6aa7468e9517729dd8b34be545d3ce41ee359047e +size 27803648 diff --git a/params_shard_49.bin b/params_shard_49.bin new file mode 100644 index 0000000000000000000000000000000000000000..8fdb9fd6d9f8ecbf7018f1894799ab1e68543cdb --- /dev/null +++ b/params_shard_49.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96d63054da297f693885dd57fefa2a74dcd92ae4ca4715af075a959e5c478b0f +size 58720256 diff --git a/params_shard_5.bin b/params_shard_5.bin new file mode 100644 index 0000000000000000000000000000000000000000..884ea873c001e9b56fce4dc085a0c06b48533dc4 --- /dev/null +++ b/params_shard_5.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b798e3498041835f282eecb68b988b78815587fe3d2aa1e501b5fe723a8a96ec +size 29360128 diff --git a/params_shard_50.bin b/params_shard_50.bin new file mode 100644 index 0000000000000000000000000000000000000000..66b0dfb74dd8c544640e7b2f933c8cedfaa8c897 --- /dev/null +++ b/params_shard_50.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d404296823312d96493c6f33b75797e3872c42e6dd7f746d1c38d37dec8ed984 +size 29360128 diff --git a/params_shard_51.bin b/params_shard_51.bin new file mode 100644 index 0000000000000000000000000000000000000000..aaa975f53a88a8d01ca4ab33fc930076f2a8e71c --- /dev/null +++ b/params_shard_51.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d5c6bf2309fc400cf4ee0430a452b8cfefb6e77924c20e99c5593a3222806371 +size 27803648 diff --git a/params_shard_52.bin b/params_shard_52.bin new file mode 100644 index 0000000000000000000000000000000000000000..03cf6b20e2d01c5bac86b4d00baa42e7fee880ce --- /dev/null +++ b/params_shard_52.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:31efc55bf009094a8fb148ba4805b2ae575cb002ce4a940c11b68fb7db2d1c97 +size 58720256 diff --git a/params_shard_53.bin b/params_shard_53.bin new file mode 100644 index 0000000000000000000000000000000000000000..d430ed3c308405b3d9e9c4968e218bf08d7ef646 --- /dev/null +++ b/params_shard_53.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3dc9af7947add6bd3db31482115a88fa63e4ea0a1030ed9cf2d55be766ad311c +size 29360128 diff --git a/params_shard_54.bin b/params_shard_54.bin new file mode 100644 index 0000000000000000000000000000000000000000..37c630e15eb7a75b00e40fab0ab65b5aebe561d3 --- /dev/null +++ b/params_shard_54.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:adf05843485cf1677d60bd31587a1387cef69debbd0074c505a2bf4f69c3eb0d +size 27803648 diff --git a/params_shard_55.bin b/params_shard_55.bin new file mode 100644 index 0000000000000000000000000000000000000000..44f03634f3346a35013d7863a1f39b3b29ae2308 --- /dev/null +++ b/params_shard_55.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ffe0954a4a74a3ca41351f4936ad18685a8f17bb4fc7210b81d376bb32596e3b +size 58720256 diff --git a/params_shard_56.bin b/params_shard_56.bin new file mode 100644 index 0000000000000000000000000000000000000000..6c54102b2296e842f0410fa3ee48f4dccd72e119 --- /dev/null +++ b/params_shard_56.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da55c0123c3b7fcaf4a5cea430b01e72f52b99c854e167f4c87446cc6eb7d9a3 +size 29360128 diff --git a/params_shard_57.bin b/params_shard_57.bin new file mode 100644 index 0000000000000000000000000000000000000000..3b5aa8834afb8c6c19673a4a3aca11bb94d0fc7b --- /dev/null +++ b/params_shard_57.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5049e68be18a0f0d6dd1b8175fdd9e77232efd7e4cd05ccd7ebec7c90031001c +size 27803648 diff --git a/params_shard_58.bin b/params_shard_58.bin new file mode 100644 index 0000000000000000000000000000000000000000..4710977121eb12b8caaf17a0c1dcc1bbd0a66ff3 --- /dev/null +++ b/params_shard_58.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82a5914c4f24a2b63926e68c4bb7ea5d920a19039e3f1bb1be773e1c1fe88e74 +size 58720256 diff --git a/params_shard_59.bin b/params_shard_59.bin new file mode 100644 index 0000000000000000000000000000000000000000..2438946a27ccd2861e8db91f8ee7383231bc28e0 --- /dev/null +++ b/params_shard_59.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d3499873b3207fb047a6d4a272bf39ce2e9710c35c427d338213de1d0ad2844e +size 29360128 diff --git a/params_shard_6.bin b/params_shard_6.bin new file mode 100644 index 0000000000000000000000000000000000000000..aa3cfd259f780fdbc2f55f6b2c49d2c6fc241a8e --- /dev/null +++ b/params_shard_6.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a78fe389f6cdb2c82cb2166d2e6c60d2ecc11ad5c4e5f35f1f11f6c6c1eb6725 +size 27803648 diff --git a/params_shard_60.bin b/params_shard_60.bin new file mode 100644 index 0000000000000000000000000000000000000000..514c427b034e09f8947b2433e52454d87c490e3a --- /dev/null +++ b/params_shard_60.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:890e2e5ebea18d852ea4d4ab9edf6f072fef028824bc955f5ad88f8801e010a8 +size 27803648 diff --git a/params_shard_61.bin b/params_shard_61.bin new file mode 100644 index 0000000000000000000000000000000000000000..9936c4d3d6c5628092bf76a74db6417d71a8c9cc --- /dev/null +++ b/params_shard_61.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ca161f191a17d88a40bff1f77cb195da7398237c65ba10b50b5aeb62c07d032 +size 58720256 diff --git a/params_shard_62.bin b/params_shard_62.bin new file mode 100644 index 0000000000000000000000000000000000000000..5674bd343fdeab4576f70d25ecda979eaf943b9b --- /dev/null +++ b/params_shard_62.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:93ca1f43cdf9fb4c6995b6449f110e21e7fad59ccd5deed0e0e6db2fa8961bce +size 29360128 diff --git a/params_shard_63.bin b/params_shard_63.bin new file mode 100644 index 0000000000000000000000000000000000000000..86cf6481cad0ecf6ff43e2b431f18cb01ef24495 --- /dev/null +++ b/params_shard_63.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82cc4392a109d951c2d6cec78e2a10004582e123aa170654971d8219482cf783 +size 27803648 diff --git a/params_shard_64.bin b/params_shard_64.bin new file mode 100644 index 0000000000000000000000000000000000000000..9c8527e51d02ec6b46c93a9341332c2e4295fc51 --- /dev/null +++ b/params_shard_64.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:46e8afa17ae0b1613970992739e45abf8566f33e69593109e4918deda28499d9 +size 58720256 diff --git a/params_shard_65.bin b/params_shard_65.bin new file mode 100644 index 0000000000000000000000000000000000000000..9db2e7cf502180562457b86bb5703930189d9e35 --- /dev/null +++ b/params_shard_65.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38068c460536ff0616b2edb533ae916583038698197209273197f41661a499fe +size 29360128 diff --git a/params_shard_66.bin b/params_shard_66.bin new file mode 100644 index 0000000000000000000000000000000000000000..c9782a1e92b034a2f94b71b489b6b3ba6bff590f --- /dev/null +++ b/params_shard_66.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ae7e7d51640d62872012097a0581434730ebdc87155c919228920fd2bb6f704 +size 27803648 diff --git a/params_shard_67.bin b/params_shard_67.bin new file mode 100644 index 0000000000000000000000000000000000000000..371e564477e21682ad28b07c4b55988a142ff506 --- /dev/null +++ b/params_shard_67.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7d86d2252aff7985f95644bac4dac3532293de4dfde1297bf8d6bbd7a9eb8302 +size 58720256 diff --git a/params_shard_68.bin b/params_shard_68.bin new file mode 100644 index 0000000000000000000000000000000000000000..3602df7af1c65a9855297cface2cc3b305df895e --- /dev/null +++ b/params_shard_68.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:016aabe8fee8329bb6fa70001bb227247ba7e2e99676a8fbd383e1e150abd92c +size 29360128 diff --git a/params_shard_69.bin b/params_shard_69.bin new file mode 100644 index 0000000000000000000000000000000000000000..015bb9d368679060b93ffbf0c42d71bfab1599d1 --- /dev/null +++ b/params_shard_69.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:392790ed99b0e2068cd2078605e485a977064305ec61f8bda0ba3c11c8cf2692 +size 27803648 diff --git a/params_shard_7.bin b/params_shard_7.bin new file mode 100644 index 0000000000000000000000000000000000000000..cd5ec3b840013d7cf3b58340e63f82555a68264a --- /dev/null +++ b/params_shard_7.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b69997e715dc0d9ee718732b3c4684dd5e16dac38373b31cc85c57f2626fc619 +size 58720256 diff --git a/params_shard_70.bin b/params_shard_70.bin new file mode 100644 index 0000000000000000000000000000000000000000..e824246b5e844db6bddb46e120574549bdb85791 --- /dev/null +++ b/params_shard_70.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a2c3a88359789ec095b998b9be37a23d7066cf735fe237461dffc2085bb12bd +size 58720256 diff --git a/params_shard_71.bin b/params_shard_71.bin new file mode 100644 index 0000000000000000000000000000000000000000..6a919c4311b46c932d77e27d4edcd736d41ec277 --- /dev/null +++ b/params_shard_71.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c411e9149ed3c3a6afa5959a7071139e0218d5e0793f8845fb6256c9a2a5e486 +size 29360128 diff --git a/params_shard_72.bin b/params_shard_72.bin new file mode 100644 index 0000000000000000000000000000000000000000..79cd045b39aaedfb2e3a6eab1384faae6ec59b3f --- /dev/null +++ b/params_shard_72.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50e17d3ee92396f86b63b5b4dc2917c4c14a1d2952ad1e181fae3f6fa7313836 +size 27803648 diff --git a/params_shard_73.bin b/params_shard_73.bin new file mode 100644 index 0000000000000000000000000000000000000000..182356837810de3915bbb1b8221f72aaa44bb1f5 --- /dev/null +++ b/params_shard_73.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f18bce41a6bd6b3a69f3f69c95b5a04f34edd0f6788247d795d6f4190e5003c7 +size 58720256 diff --git a/params_shard_74.bin b/params_shard_74.bin new file mode 100644 index 0000000000000000000000000000000000000000..402eb54442460e196c8c10973b9a0fe9751be109 --- /dev/null +++ b/params_shard_74.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3102beaafd83c34ce1db68a859ad3b9315057c36061aeaece91092591a0fd4bf +size 29360128 diff --git a/params_shard_75.bin b/params_shard_75.bin new file mode 100644 index 0000000000000000000000000000000000000000..240e2459abda6db49d0c79f1d1b6d2518d116331 --- /dev/null +++ b/params_shard_75.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7ff18a62d9b69f2f77420d1aa8e086df4de9f6b2bd0b130e9455f053923f3809 +size 27803648 diff --git a/params_shard_76.bin b/params_shard_76.bin new file mode 100644 index 0000000000000000000000000000000000000000..ad1b3c1fba5c538a29e4748c6f53ee31eb55440f --- /dev/null +++ b/params_shard_76.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:49a69ded9a08148044cb210f8bc06d55f6383d67959c47550d3539088e90cb12 +size 58720256 diff --git a/params_shard_77.bin b/params_shard_77.bin new file mode 100644 index 0000000000000000000000000000000000000000..846931e000ed0342c35af7a41c2f31e26a1826c7 --- /dev/null +++ b/params_shard_77.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7954e6ffba48a2347221790faced4ed7240978aa02032fe3dfc891f6a4bc275b +size 29360128 diff --git a/params_shard_78.bin b/params_shard_78.bin new file mode 100644 index 0000000000000000000000000000000000000000..40508ac0b69fe9ab3d5ff73cb23bb241a6012e42 --- /dev/null +++ b/params_shard_78.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1233e00bfa06847c68cd8d366c497b3224822cf2721b7b5e6dca46c63d76883e +size 27803648 diff --git a/params_shard_79.bin b/params_shard_79.bin new file mode 100644 index 0000000000000000000000000000000000000000..bf1ed749f9be0a12c57842f8114e38ac6fa5f65b --- /dev/null +++ b/params_shard_79.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0787da5f54e90dfd7b8f568bba09fe6a0cd23af9df6e74465ad05677e8cbc8a2 +size 58720256 diff --git a/params_shard_8.bin b/params_shard_8.bin new file mode 100644 index 0000000000000000000000000000000000000000..2c60c2b5473695df2733b5d38ffa209eaad1e2bc --- /dev/null +++ b/params_shard_8.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b2709dbed96d11bd962743af043da44043234bf8fb067da0c4c571da1725b06 +size 29360128 diff --git a/params_shard_80.bin b/params_shard_80.bin new file mode 100644 index 0000000000000000000000000000000000000000..63440808dc4b836063f91267d347273998ff76fc --- /dev/null +++ b/params_shard_80.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a416467f2650f20eecf1b0da1f01c57b579712ff75cc7a69b9326165e3f923e +size 29360128 diff --git a/params_shard_81.bin b/params_shard_81.bin new file mode 100644 index 0000000000000000000000000000000000000000..114c11905589b2ad2f6799d3fadafa308405a924 --- /dev/null +++ b/params_shard_81.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:03e520215128556bfca5dfee2d6e1e50d74cc2293f0ebbebf00a5ac3cb1c8eeb +size 27803648 diff --git a/params_shard_82.bin b/params_shard_82.bin new file mode 100644 index 0000000000000000000000000000000000000000..4561f1b243a9ff08b1c7db2ee282810119c24931 --- /dev/null +++ b/params_shard_82.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:76243848d5d0f54bb1d7a133a2a115c2ba856020f5d55aa7834f6542cb68340b +size 58720256 diff --git a/params_shard_83.bin b/params_shard_83.bin new file mode 100644 index 0000000000000000000000000000000000000000..3ca512b1e5d33b5b670360abcb334797911ce59f --- /dev/null +++ b/params_shard_83.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:229d3cdf1e08f5808b499aadadbfcdf0974a1bc5468f4d77c81fbf8c13506e3e +size 29360128 diff --git a/params_shard_84.bin b/params_shard_84.bin new file mode 100644 index 0000000000000000000000000000000000000000..a62ba92861474eb9f0299f01043dd7051b09511a --- /dev/null +++ b/params_shard_84.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7a7a39c7ad593676fdb4357d840caf9be944ee95e3a178245d86afc60aa19d2d +size 27803648 diff --git a/params_shard_85.bin b/params_shard_85.bin new file mode 100644 index 0000000000000000000000000000000000000000..665425643e862641d255b69c9108fb86fc9aad27 --- /dev/null +++ b/params_shard_85.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aafdb277915310e8d042d5345df13d20428903b7feada27b9dbf4a2e0d3ffc9b +size 58720256 diff --git a/params_shard_86.bin b/params_shard_86.bin new file mode 100644 index 0000000000000000000000000000000000000000..9866f1b534555342d8766b77e52db2603ba6b1ec --- /dev/null +++ b/params_shard_86.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3565ceb890df4ce500801c616aea435c6bcc073e09eb739541f8e334a2fbf26b +size 29360128 diff --git a/params_shard_87.bin b/params_shard_87.bin new file mode 100644 index 0000000000000000000000000000000000000000..b0a29b5878b54b6292848747b442d004b2137cd6 --- /dev/null +++ b/params_shard_87.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4f728c4af8a42d356048051418311168fbd10e3a8cae99f05789db3529af57e9 +size 27803648 diff --git a/params_shard_88.bin b/params_shard_88.bin new file mode 100644 index 0000000000000000000000000000000000000000..53d692f8c2c70d4262ce40939f7fd6d59988d555 --- /dev/null +++ b/params_shard_88.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:83c9941fa7e949e2807d957cb8b684b5f3477d717e91b74584d4dcc74dbcc659 +size 58720256 diff --git a/params_shard_89.bin b/params_shard_89.bin new file mode 100644 index 0000000000000000000000000000000000000000..6c61621ba0408852f0a3b74c2518fb2d0791475f --- /dev/null +++ b/params_shard_89.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd5f186276337f977c4987f4389cbda4ad79a0b6dc06d5c5387c02f227ad7582 +size 29360128 diff --git a/params_shard_9.bin b/params_shard_9.bin new file mode 100644 index 0000000000000000000000000000000000000000..ce4670443954da9244356d0f2114b0ab2ea20bde --- /dev/null +++ b/params_shard_9.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f3ad8afd0dee00e0e2a4ce621613a3c00fb0aac3b37db26af2bc5575fb7a2c03 +size 27803648 diff --git a/params_shard_90.bin b/params_shard_90.bin new file mode 100644 index 0000000000000000000000000000000000000000..e0191a11ff2bcd79c0ddda95a485b6ece8b034c5 --- /dev/null +++ b/params_shard_90.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bbbb6eb015846fe3e4e134ed3cd61ea636f85e558c52fa32e4f0117a9d17f1af +size 27803648 diff --git a/params_shard_91.bin b/params_shard_91.bin new file mode 100644 index 0000000000000000000000000000000000000000..e760f4e7a2ccc9ebf8fdb30fa6a1f1db44477dfc --- /dev/null +++ b/params_shard_91.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ec5ccab366793d575560f790dc64b75f686b65609779b8a295cee7f8c39fb60 +size 58720256 diff --git a/params_shard_92.bin b/params_shard_92.bin new file mode 100644 index 0000000000000000000000000000000000000000..9a53969364f17cb8159dbc492adb65ee1f8affbf --- /dev/null +++ b/params_shard_92.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65a3ccaae22b540865b967346ac74d12de2f45911f9495b4c30b051a3d40e722 +size 29360128 diff --git a/params_shard_93.bin b/params_shard_93.bin new file mode 100644 index 0000000000000000000000000000000000000000..7161b23c2660f5ef2320c6c6125b237e4e74bdca --- /dev/null +++ b/params_shard_93.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:13c51794560915d48e3248a8f859edae6771adcd9d71d443ea37678911356a96 +size 27803648 diff --git a/params_shard_94.bin b/params_shard_94.bin new file mode 100644 index 0000000000000000000000000000000000000000..363eb48b5b588165fecd223077fc8fc90127deb2 --- /dev/null +++ b/params_shard_94.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b03940e9d4493b3be2d1cb515b0dcf13b6b69c5cf2878cae41bcad49cadd7394 +size 58720256 diff --git a/params_shard_95.bin b/params_shard_95.bin new file mode 100644 index 0000000000000000000000000000000000000000..cb984b799b56191bc903e75ef9fbc66963c8f843 --- /dev/null +++ b/params_shard_95.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:48e7bfbe634ec408f9cca219be6a55c3a11d7d003db86ce0003f0a9b67f97c82 +size 29360128 diff --git a/params_shard_96.bin b/params_shard_96.bin new file mode 100644 index 0000000000000000000000000000000000000000..94a93bf7e92905f0edbd2106fbbcbc6d47bdc38c --- /dev/null +++ b/params_shard_96.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:23652030d6bf224058acde07eb0c1e3fc565b4ef16c0e9587a237c4f3f087eba +size 262144000 diff --git a/params_shard_97.bin b/params_shard_97.bin new file mode 100644 index 0000000000000000000000000000000000000000..078604712b909e119615f8c69fad7a28dab7c5ba --- /dev/null +++ b/params_shard_97.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:73d82f1b803f88a63132b672976bbd74cc2617c23308d0cd417ed84397991bfa +size 28860416 diff --git a/private-llm-config.json b/private-llm-config.json new file mode 100644 index 0000000000000000000000000000000000000000..16834fad65d54bba2fe36088ba07ac7733b8d437 --- /dev/null +++ b/private-llm-config.json @@ -0,0 +1,21 @@ +{ + "model_lib": "Mistral-Trismegistus-7B-w4a16g128asym", + "local_id": "Mistral-Trismegistus-7B-w4a16g128asym", + "conv_template": "vicuna_v1.1", + "temperature": 0.7, + "repetition_penalty": 1.1, + "top_p": 0.95, + "mean_gen_len": 512, + "max_gen_len": 2048, + "num_shards": 1, + "shift_fill_factor": 0.3, + "tokenizer_files": [ + "added_tokens.json", + "tokenizer.model" + ], + "model_category": "mistral", + "model_name": "Mistral-Trismegistus-7B", + "vocab_size": 32000, + "sliding_window": 4096, + "sliding_window_chunk_size": 4096 +} 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..f5b3416bc161b8bdcce471d2a306cccb5d54389d --- /dev/null +++ b/tokenizer_config.json @@ -0,0 +1,48 @@ +{ + "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 + } + }, + "additional_special_tokens": [ + "", + "", + "" + ], + "bos_token": "", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": false, + "model_max_length": 1000000000000000019884624838656, + "pad_token": null, + "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 +}