diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,5153 @@ +{ + "metadata": { + "ParamSize": 339, + "ParamBytes": 29540067328.0, + "BitsPerParam": 16.0 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 1557135360, + "records": [ + { + "name": "lm_head.weight", + "shape": [ + 152064, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1557135360, + "byteOffset": 0 + } + ], + "md5sum": "021ce94ca7f373f89ac030932f4319ad" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.47.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "4b7afa187c64b584f1d20102510d3e73" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 1557135360, + "records": [ + { + "name": "model.embed_tokens.weight", + "shape": [ + 152064, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1557135360, + "byteOffset": 0 + } + ], + "md5sum": "4b5d6fc9be16ec13d5e6433b229eb977" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.0.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "dad037ae5bd7fad586e0308ad8f2e3da" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "caf1c0801505d206587d1c343531b1ee" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.0.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "6e927c2db682d8f86737abfc64c43572" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.0.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "2e354604e0c2cb99464faa66b239b585" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.1.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "bebd4062b50997fecd8e96b01ff243df" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.1.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "d70ea65b2d5fb6c14e40c2bae62a0ff1" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.1.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "6b4a8a954f0149211aec3e9a7c342bef" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.1.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "0cc7d8c9d7160ed412d40dac20fe3190" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.2.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "793e871f8cfa68d2f9d9f4449aac0c7c" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "f71bac16d6ff791f50696df9ea7b749d" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.2.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "8900f688f90bdd725dc3b0608d036c8e" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.2.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "fbafa253be592962a03da0d11ccbd8cf" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.3.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "cbcc08ca09d3a9aa47168df0260d36b6" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.3.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "28154774ff302928aa2e3574fdd5ec04" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.3.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "4912d112bc5cfbec77645726643a261e" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.3.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "50a7220b93cdbb1a0bf88173d4c56665" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.4.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "070ec8645c4795cd28355dfa1b962f65" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.4.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "6db0b230d594ae9df3f842e9fffdf46e" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.10.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "d673ae5d66eb9622b3de738f677e6507" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "514729955cd96e1004aa57b771a66f78" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.10.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "30f4138bfd35954e34e4db680d5c4865" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.10.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "a821065615b908cb954d18e4e04287db" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.11.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "d4fc91addfaa04dfecc8445c95b9fac9" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.11.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "8f0937be75d61442056e618bd15764f8" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.11.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "901d2ede49491894dcba236f66ce87bc" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.4.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "e954c7ea1fd76e1e89e8681a02aa29ff" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "c38e40106e42b8bfad621078cd13414f" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.5.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "a8433658f86e25c8e386c53caa1ccf82" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.5.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "4ccc9d6a7a74e8cabaa1ad80e379815f" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.5.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "a183c950a18840e053c345e8009f2a18" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.5.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "73d40765f4625eb2a613c8e692f26b3d" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.6.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "24ed28b0dca38fbb912b7f5b054f1568" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "6cbe66f33d522d4dad76c4b8492954a4" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.6.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "f8916e96acb8bc2f4e9e1f13ca708469" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.6.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "fdb2ba5bd0f077a2b74bfadff91cd9c7" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.7.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "6adafcb6a5415dad968a4cb88676528b" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "75200cfe0b6dc6a0a6a2b64abbe97bdb" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.7.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "6325542b871bb0a32682c56d0f6a85cb" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.7.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "c87157e5a438d7e77517eaded08ab53f" + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.8.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "ef0356883b7c36ea0a864fe8bb3fe1ed" + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "85dbf830b818d7b2b3718d8e0e7184ff" + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.8.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "b82670f151e3ddfb4d327b67eab2c7bd" + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.8.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "c257cbf9735f4e508c6cde733fe1dea8" + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.9.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "182183f0d8802cc2824c1a8987d1229e" + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "4bbb83542b3ba1a5f039208e439638ee" + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.9.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "ce56e34475e95297b80bb8eac58ee878" + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.9.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "4de81e8132995062b56df9ba9d7aac67" + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.11.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "3332f1a8395e9d4cec2e64fd5e34afe4" + }, + { + "dataPath": "params_shard_51.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.12.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "3bc88aefff102a6788880737bf3a46ce" + }, + { + "dataPath": "params_shard_52.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "1c97e7e1b1443e4879ca3a31bdf17216" + }, + { + "dataPath": "params_shard_53.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.12.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "bfeface08bc6d944f6fdf922d7b1c3cf" + }, + { + "dataPath": "params_shard_54.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.12.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "1273cd79e9af708516122c1db14f3721" + }, + { + "dataPath": "params_shard_55.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.13.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "5450339345c0a440c0f64eaf4f774de1" + }, + { + "dataPath": "params_shard_56.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.13.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "b15e93fd6f08351ddc7ce4b06913864b" + }, + { + "dataPath": "params_shard_57.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.13.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "4a262f5d89808c634ed0e16ceed7f5d0" + }, + { + "dataPath": "params_shard_58.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.13.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "4b7435362b536f8eadacd9b3718d6c6b" + }, + { + "dataPath": "params_shard_59.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.14.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "1d15029f65d66889bce09fe009d85624" + }, + { + "dataPath": "params_shard_60.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "5705e7d2f63a9871f6522ff505b484e7" + }, + { + "dataPath": "params_shard_61.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.14.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "9bbbfd9fc69e4d6bf42ecba5330cc652" + }, + { + "dataPath": "params_shard_62.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.14.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "9ac10484fcb746a8c3fd0effdc1a0fd6" + }, + { + "dataPath": "params_shard_63.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.15.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "6fbe01a58d00fc7feafec63f043c84b9" + }, + { + "dataPath": "params_shard_64.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.15.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "e14108fef80465f71f3969ecb8722bb8" + }, + { + "dataPath": "params_shard_65.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.15.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "bbadf4fc8f806ea28c5a356ace7dfc26" + }, + { + "dataPath": "params_shard_66.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.15.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "3e43dfd1587165b46aaee3ae66b890ef" + }, + { + "dataPath": "params_shard_67.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.16.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "2a2e8346597007aabf0bb371d9642f18" + }, + { + "dataPath": "params_shard_68.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "40229bbc1d0279af3e0d76fd6a66ca3e" + }, + { + "dataPath": "params_shard_69.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.16.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "9506c2f099bb98775dce50b58d92a123" + }, + { + "dataPath": "params_shard_70.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.16.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "cb9d4d9ac1b81509a136b7ff8403f827" + }, + { + "dataPath": "params_shard_71.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.17.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "73b2ca7471dbeecdc2b6d54218f9f599" + }, + { + "dataPath": "params_shard_72.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.17.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "d3b5ef3ad520bee59b4c457914de6ac1" + }, + { + "dataPath": "params_shard_73.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.17.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "bfb882b4389871b928dc9a65706a269c" + }, + { + "dataPath": "params_shard_74.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.17.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "32ab01ab76aba073dfb1f4e9d0f944ef" + }, + { + "dataPath": "params_shard_75.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "fa071cec182f5b49b0b43771208277f9" + }, + { + "dataPath": "params_shard_76.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.18.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "5c82256dcb0c10066d82cd347ea6db09" + }, + { + "dataPath": "params_shard_77.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.18.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "4c9706a113499cad30a5a8c7fe0d8458" + }, + { + "dataPath": "params_shard_78.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.18.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "dbcddf02bb3b1efc051a41bf00be5137" + }, + { + "dataPath": "params_shard_79.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.19.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "810f09132e0ea695a626389a3c340e82" + }, + { + "dataPath": "params_shard_80.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.19.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "aca9503b00dd2046e024c99b7189fcf9" + }, + { + "dataPath": "params_shard_81.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.19.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "acc0eacc2cba26631bade954c6acfc23" + }, + { + "dataPath": "params_shard_82.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.19.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "d5f8ef5dd925a50a81ffdba4d45b4d9d" + }, + { + "dataPath": "params_shard_83.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.20.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "2a58285b2058e39b323bcd5e0574eea6" + }, + { + "dataPath": "params_shard_84.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "e013295deac53d7c7d8ca72f3123dfaf" + }, + { + "dataPath": "params_shard_85.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.20.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "46e8051d33aa001f3ae0e2b30f234e12" + }, + { + "dataPath": "params_shard_86.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.20.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "03851849fab268dfb22ceea832d0b18e" + }, + { + "dataPath": "params_shard_87.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.21.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "57909ae8d0e3f0528216dbe76037e910" + }, + { + "dataPath": "params_shard_88.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "9b9d6985c2d7e7a1d7a9e0bac399013a" + }, + { + "dataPath": "params_shard_89.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.21.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "bb0a1a6cbaa6f3342664820c7cbe3003" + }, + { + "dataPath": "params_shard_90.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.21.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "bba4a3968ac59aac8480f5ea9bb04b75" + }, + { + "dataPath": "params_shard_91.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.22.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "4b4786617c48407784e921ba30fd8ecc" + }, + { + "dataPath": "params_shard_92.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.22.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "334262e15b861f9d0fa857ee99368932" + }, + { + "dataPath": "params_shard_93.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.22.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "7c80380bdb8bc894ca794ab1b7af1d8b" + }, + { + "dataPath": "params_shard_94.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.22.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "cd93c24afd38fcd6d59d9360e0227eb9" + }, + { + "dataPath": "params_shard_95.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.23.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "a36b754e3e4b3769e6e406ff6db48031" + }, + { + "dataPath": "params_shard_96.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "225989dba07260506212d7f066b30c46" + }, + { + "dataPath": "params_shard_97.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.23.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "a37489f097acbf205e3ca7b885152b49" + }, + { + "dataPath": "params_shard_98.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.23.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "c9cde7a9bcfa2cdf2624658ef12166d9" + }, + { + "dataPath": "params_shard_99.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.24.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "cec69c4e7454288b444bc76b0eb82845" + }, + { + "dataPath": "params_shard_100.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "5dd2044683f9793b6e52c5b6f10b02fb" + }, + { + "dataPath": "params_shard_101.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.24.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "9271398bee66af28f7ea04e139842060" + }, + { + "dataPath": "params_shard_102.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.24.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "ea4cf7a3f39d3368a0359e34cb56e0a1" + }, + { + "dataPath": "params_shard_103.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.25.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "1648c92ebb47bd2b2d58e74f2395f634" + }, + { + "dataPath": "params_shard_104.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "b98260d9bbab71594186d27528d99670" + }, + { + "dataPath": "params_shard_105.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.25.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "7f1192c1ce75d54ee3b22005154bb65f" + }, + { + "dataPath": "params_shard_106.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.25.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "ec8afc9972e5d41e81c7ad7421d1e34d" + }, + { + "dataPath": "params_shard_107.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.26.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "9688426f59aa18d0c79237141c5f51d5" + }, + { + "dataPath": "params_shard_108.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.26.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "cef00e2b0187f1f3abce81e8c296dd32" + }, + { + "dataPath": "params_shard_109.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.26.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "a3a7e858fdcbd655f1e062f6003f2de6" + }, + { + "dataPath": "params_shard_110.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.26.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "94504b5733fe91ba745fd6db29d44eae" + }, + { + "dataPath": "params_shard_111.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.27.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "bef33e47b9a49037697fe40f08b934b9" + }, + { + "dataPath": "params_shard_112.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.27.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "5113c2d8696fca5d5d47e4ad70a5039a" + }, + { + "dataPath": "params_shard_113.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.27.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "6aa33063700ce7cdf82dd7f422ecbfe0" + }, + { + "dataPath": "params_shard_114.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.27.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "69105557791084dcde6ecc661700b367" + }, + { + "dataPath": "params_shard_115.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.28.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "107ef74a33b145b6fa7c3285c38ecc18" + }, + { + "dataPath": "params_shard_116.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.28.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "ebf825e44aa711597e1762f9c5a4bd6a" + }, + { + "dataPath": "params_shard_117.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.28.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "365561046f84e25b7f53d98c3e490155" + }, + { + "dataPath": "params_shard_118.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.28.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "08378eca42fd01b0e5d5fd0351973ff1" + }, + { + "dataPath": "params_shard_119.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.29.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "15ab41885c6dbab0049428709ad1014f" + }, + { + "dataPath": "params_shard_120.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.29.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "b97b538722994fec761c88b92f3bba83" + }, + { + "dataPath": "params_shard_121.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.29.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "8b0f2e4e56821c4523661c8a42dac6da" + }, + { + "dataPath": "params_shard_122.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.29.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "21288d03f10d6541a52ccd6475e53abb" + }, + { + "dataPath": "params_shard_123.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.30.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "0a59e198c139c5d692b69d68940c2aeb" + }, + { + "dataPath": "params_shard_124.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.30.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "ad34b77d83c8cea42a3069ee9f24ebea" + }, + { + "dataPath": "params_shard_125.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.30.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "744028d0c2ebe4975752cfcf7d24d28a" + }, + { + "dataPath": "params_shard_126.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.30.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "a0b8e99b5c3e5d5eaa22c107eb7894c4" + }, + { + "dataPath": "params_shard_127.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.31.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "99d564b8a8386c1ef164f1e75155ca93" + }, + { + "dataPath": "params_shard_128.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.31.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "6904aff80c8ab4a7b1c5617cbab469c0" + }, + { + "dataPath": "params_shard_129.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.31.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "df907562fe366e0bc93cebec95206d46" + }, + { + "dataPath": "params_shard_130.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.31.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "74c3418dd2e0ee6a44527c4b47f86e03" + }, + { + "dataPath": "params_shard_131.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.32.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "637e6d7a0a2e4d87df509013968681ca" + }, + { + "dataPath": "params_shard_132.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.32.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "f13397f7e197c2cc2e9e27fa6c739d32" + }, + { + "dataPath": "params_shard_133.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.32.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "c49d275c59e41b8ee5adf567aaa0bb37" + }, + { + "dataPath": "params_shard_134.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.32.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "f7db0352d5f75dbf924ba45f92dc71fb" + }, + { + "dataPath": "params_shard_135.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.33.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "d4db8ba7ecc06c5375dc801fd3c18143" + }, + { + "dataPath": "params_shard_136.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.33.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "d34fd3223e4f7821e948dfc3046d66f4" + }, + { + "dataPath": "params_shard_137.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.33.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "2c48238e7819bf9e512b48fb2c0fd4a2" + }, + { + "dataPath": "params_shard_138.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.33.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "c0e86382ea9d37727e458a9ed71b44bd" + }, + { + "dataPath": "params_shard_139.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.34.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "270b661858cf0f993d6a927304fe5a66" + }, + { + "dataPath": "params_shard_140.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.34.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "de8fa95e427cb9bf5b6ed30977cbd071" + }, + { + "dataPath": "params_shard_141.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.34.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "c02405f5a67b148880b102687def560b" + }, + { + "dataPath": "params_shard_142.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.34.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "c2c57246b0d010b4467053015ab3265c" + }, + { + "dataPath": "params_shard_143.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.35.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "4a190ec51d7e41c5f3f5c6bdf90991d3" + }, + { + "dataPath": "params_shard_144.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.35.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "52776cb5a8f899ad51d8e6b3b5011e7b" + }, + { + "dataPath": "params_shard_145.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.35.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "52046da70c115e0d5ee5034f60d01730" + }, + { + "dataPath": "params_shard_146.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.35.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "189191254ad18943f50ffb364395afa7" + }, + { + "dataPath": "params_shard_147.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.36.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "f7aae2dc764f204c12b45433e2288c3c" + }, + { + "dataPath": "params_shard_148.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.36.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "64446817c857e04f28d04e24d0c22dcd" + }, + { + "dataPath": "params_shard_149.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.36.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "4cf27041fce37a9ba97d6e63fd2fb431" + }, + { + "dataPath": "params_shard_150.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.36.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "8559ccc24db21ebf4da39adf26a4331d" + }, + { + "dataPath": "params_shard_151.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.37.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "b05592423179d142d9f152e7afa759fd" + }, + { + "dataPath": "params_shard_152.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.37.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "1243dbdcbfa46183990538b6e8d2482c" + }, + { + "dataPath": "params_shard_153.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.37.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "42ed430d037cde970d61158647cd84f7" + }, + { + "dataPath": "params_shard_154.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.37.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "31eec2af1e5803d53cacd66d99524c3b" + }, + { + "dataPath": "params_shard_155.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.38.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "51753f049e6aa397e530a88c75fe3974" + }, + { + "dataPath": "params_shard_156.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.38.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "da8236d8206dc4c63be9ca886b195648" + }, + { + "dataPath": "params_shard_157.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.38.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "c117dcc81aa4eba245d74e44976c8943" + }, + { + "dataPath": "params_shard_158.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.38.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "bde69b718f5b1d39eebd0c8ad36e68c0" + }, + { + "dataPath": "params_shard_159.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.39.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "72d83c5c6d0f7e5c6e5d46fd76ab2cfc" + }, + { + "dataPath": "params_shard_160.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.39.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "a61632fac6a7721ecd918cdeb93f6251" + }, + { + "dataPath": "params_shard_161.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.39.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "f7920816682fecad407630e84af82a0c" + }, + { + "dataPath": "params_shard_162.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.39.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "165d875b332c7bce1cea4082add760bd" + }, + { + "dataPath": "params_shard_163.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.40.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "92475e17dd24a59ea07853f1ae36d791" + }, + { + "dataPath": "params_shard_164.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.40.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "f0c14c435b6bfe664e9093c1a8237a4b" + }, + { + "dataPath": "params_shard_165.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.40.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "95582f3253d8523931ad1b7bc714a25d" + }, + { + "dataPath": "params_shard_166.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.40.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "c4df2a486897705202f68cad86075b77" + }, + { + "dataPath": "params_shard_167.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.41.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "066b5c82459730c85ea3c97a29df742a" + }, + { + "dataPath": "params_shard_168.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.41.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "076fb1afdf88d8f0f1231f1bf1cfd84b" + }, + { + "dataPath": "params_shard_169.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.41.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "22fbe937c883c942624d82076348aa61" + }, + { + "dataPath": "params_shard_170.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.41.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "9179c2b5b4d641886e04091ebdde38ef" + }, + { + "dataPath": "params_shard_171.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.42.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "2fe5671226265b1d054a941ea2ed74d7" + }, + { + "dataPath": "params_shard_172.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.42.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "89bba482f19f8298e257d0dc72c30ace" + }, + { + "dataPath": "params_shard_173.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.42.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "f884b78e95ffedc0c8ad85f5ef29a391" + }, + { + "dataPath": "params_shard_174.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.42.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "c222e5c7d6dd0c68d1ab57aea000babf" + }, + { + "dataPath": "params_shard_175.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.43.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "9628c3d2d25556f524e73f4401c619e8" + }, + { + "dataPath": "params_shard_176.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.43.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "fcf93ed29b0d9501e44c2511425f3e2d" + }, + { + "dataPath": "params_shard_177.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.43.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "a5df71c1aadaff964ca80896bd33e49d" + }, + { + "dataPath": "params_shard_178.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.43.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "ecd4928cd9c2d9f4ee9048fed00a6438" + }, + { + "dataPath": "params_shard_179.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.44.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "d9c123e209792e52e2504b1f0e6d2ca6" + }, + { + "dataPath": "params_shard_180.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.44.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "2ece5010c7b10f1cd187d56dbba8f234" + }, + { + "dataPath": "params_shard_181.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.44.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "35effac46990a316b859e908ae6fcdf2" + }, + { + "dataPath": "params_shard_182.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.44.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "246639b2a385988ca4fae4fbb754b491" + }, + { + "dataPath": "params_shard_183.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.45.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "d2749095009eba712c43788abfb2b613" + }, + { + "dataPath": "params_shard_184.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.45.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "449be869d2c22c6a3385c0b17abd4ec8" + }, + { + "dataPath": "params_shard_185.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.45.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "bc087bccbcb8786d9e98e060278573f6" + }, + { + "dataPath": "params_shard_186.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.45.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "6b4f5bfa4cd205ad6d55d57e182514eb" + }, + { + "dataPath": "params_shard_187.bin", + "format": "raw-shard", + "nbytes": 141557760, + "records": [ + { + "name": "model.layers.46.mlp.down_proj.weight", + "shape": [ + 5120, + 13824 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 141557760, + "byteOffset": 0 + } + ], + "md5sum": "31562a32302e48003185b2b3800e83be" + }, + { + "dataPath": "params_shard_188.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.46.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "8489b9488f636db396e6ba224606f122" + }, + { + "dataPath": "params_shard_189.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.46.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "6c1672ad93811fb6d9bef70128ef4517" + }, + { + "dataPath": "params_shard_190.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.46.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "beb0d0f7e5f468fdbbccf76eafad0648" + }, + { + "dataPath": "params_shard_191.bin", + "format": "raw-shard", + "nbytes": 283115520, + "records": [ + { + "name": "model.layers.47.mlp.gate_up_proj.weight", + "shape": [ + 27648, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 283115520, + "byteOffset": 0 + } + ], + "md5sum": "e040f852c70be0ab69846ff44e85b043" + }, + { + "dataPath": "params_shard_192.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.47.self_attn.c_attn.weight", + "shape": [ + 7168, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "74796385a54f9e62bca96b51f56e3465" + }, + { + "dataPath": "params_shard_193.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "model.layers.47.self_attn.o_proj.weight", + "shape": [ + 5120, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "147fa005531b476c6f69afccf1df453b" + }, + { + "dataPath": "params_shard_194.bin", + "format": "raw-shard", + "nbytes": 1681408, + "records": [ + { + "name": "model.norm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 0 + }, + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 10240 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 20480 + }, + { + "name": "model.layers.0.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 30720 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 45056 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 55296 + }, + { + "name": "model.layers.1.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 65536 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 79872 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 90112 + }, + { + "name": "model.layers.2.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 100352 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 114688 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 124928 + }, + { + "name": "model.layers.3.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 135168 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 149504 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 159744 + }, + { + "name": "model.layers.4.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 169984 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 184320 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 194560 + }, + { + "name": "model.layers.10.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 204800 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 219136 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 229376 + }, + { + "name": "model.layers.11.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 239616 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 253952 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 264192 + }, + { + "name": "model.layers.5.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 274432 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 288768 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 299008 + }, + { + "name": "model.layers.6.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 309248 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 323584 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 333824 + }, + { + "name": "model.layers.7.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 344064 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 358400 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 368640 + }, + { + "name": "model.layers.8.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 378880 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 393216 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 403456 + }, + { + "name": "model.layers.9.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 413696 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 428032 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 438272 + }, + { + "name": "model.layers.12.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 448512 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 462848 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 473088 + }, + { + "name": "model.layers.13.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 483328 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 497664 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 507904 + }, + { + "name": "model.layers.14.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 518144 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 532480 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 542720 + }, + { + "name": "model.layers.15.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 552960 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 567296 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 577536 + }, + { + "name": "model.layers.16.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 587776 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 602112 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 612352 + }, + { + "name": "model.layers.17.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 622592 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 636928 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 647168 + }, + { + "name": "model.layers.18.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 657408 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 671744 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 681984 + }, + { + "name": "model.layers.19.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 692224 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 706560 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 716800 + }, + { + "name": "model.layers.20.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 727040 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 741376 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 751616 + }, + { + "name": "model.layers.21.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 761856 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 776192 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 786432 + }, + { + "name": "model.layers.22.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 796672 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 811008 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 821248 + }, + { + "name": "model.layers.23.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 831488 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 845824 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 856064 + }, + { + "name": "model.layers.24.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 866304 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 880640 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 890880 + }, + { + "name": "model.layers.25.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 901120 + }, + { + "name": "model.layers.26.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 915456 + }, + { + "name": "model.layers.26.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 925696 + }, + { + "name": "model.layers.26.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 935936 + }, + { + "name": "model.layers.27.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 950272 + }, + { + "name": "model.layers.27.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 960512 + }, + { + "name": "model.layers.27.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 970752 + }, + { + "name": "model.layers.28.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 985088 + }, + { + "name": "model.layers.28.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 995328 + }, + { + "name": "model.layers.28.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1005568 + }, + { + "name": "model.layers.29.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1019904 + }, + { + "name": "model.layers.29.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1030144 + }, + { + "name": "model.layers.29.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1040384 + }, + { + "name": "model.layers.30.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1054720 + }, + { + "name": "model.layers.30.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1064960 + }, + { + "name": "model.layers.30.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1075200 + }, + { + "name": "model.layers.31.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1089536 + }, + { + "name": "model.layers.31.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1099776 + }, + { + "name": "model.layers.31.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1110016 + }, + { + "name": "model.layers.32.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1124352 + }, + { + "name": "model.layers.32.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1134592 + }, + { + "name": "model.layers.32.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1144832 + }, + { + "name": "model.layers.33.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1159168 + }, + { + "name": "model.layers.33.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1169408 + }, + { + "name": "model.layers.33.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1179648 + }, + { + "name": "model.layers.34.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1193984 + }, + { + "name": "model.layers.34.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1204224 + }, + { + "name": "model.layers.34.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1214464 + }, + { + "name": "model.layers.35.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1228800 + }, + { + "name": "model.layers.35.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1239040 + }, + { + "name": "model.layers.35.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1249280 + }, + { + "name": "model.layers.36.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1263616 + }, + { + "name": "model.layers.36.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1273856 + }, + { + "name": "model.layers.36.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1284096 + }, + { + "name": "model.layers.37.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1298432 + }, + { + "name": "model.layers.37.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1308672 + }, + { + "name": "model.layers.37.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1318912 + }, + { + "name": "model.layers.38.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1333248 + }, + { + "name": "model.layers.38.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1343488 + }, + { + "name": "model.layers.38.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1353728 + }, + { + "name": "model.layers.39.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1368064 + }, + { + "name": "model.layers.39.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1378304 + }, + { + "name": "model.layers.39.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1388544 + }, + { + "name": "model.layers.40.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1402880 + }, + { + "name": "model.layers.40.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1413120 + }, + { + "name": "model.layers.40.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1423360 + }, + { + "name": "model.layers.41.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1437696 + }, + { + "name": "model.layers.41.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1447936 + }, + { + "name": "model.layers.41.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1458176 + }, + { + "name": "model.layers.42.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1472512 + }, + { + "name": "model.layers.42.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1482752 + }, + { + "name": "model.layers.42.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1492992 + }, + { + "name": "model.layers.43.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1507328 + }, + { + "name": "model.layers.43.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1517568 + }, + { + "name": "model.layers.43.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1527808 + }, + { + "name": "model.layers.44.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1542144 + }, + { + "name": "model.layers.44.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1552384 + }, + { + "name": "model.layers.44.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1562624 + }, + { + "name": "model.layers.45.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1576960 + }, + { + "name": "model.layers.45.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1587200 + }, + { + "name": "model.layers.45.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1597440 + }, + { + "name": "model.layers.46.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1611776 + }, + { + "name": "model.layers.46.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1622016 + }, + { + "name": "model.layers.46.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1632256 + }, + { + "name": "model.layers.47.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1646592 + }, + { + "name": "model.layers.47.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 1656832 + }, + { + "name": "model.layers.47.self_attn.c_attn.bias", + "shape": [ + 7168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336, + "byteOffset": 1667072 + } + ], + "md5sum": "55444d4291d56dac5badb1c09ecd2240" + } + ] +} \ No newline at end of file