diff --git "a/tensor-cache.json" "b/tensor-cache.json" new file mode 100644--- /dev/null +++ "b/tensor-cache.json" @@ -0,0 +1,13937 @@ +{ + "metadata": { + "ParamSize": 1119, + "ParamBytes": 15194704384.0, + "BitsPerParam": 4.431169188588552 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 704815104, + "records": [ + { + "name": "language_model.model.embed_tokens.q_weight", + "shape": [ + 262208, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 704815104, + "byteOffset": 0 + } + ], + "md5sum": "ce2814e6ad13ffb16e9be092acc67608" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 88101888, + "records": [ + { + "name": "language_model.model.embed_tokens.q_scale", + "shape": [ + 262208, + 168 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 88101888, + "byteOffset": 0 + } + ], + "md5sum": "c10f794772334957d727081ee5c0d299" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "0c2a84c37ad79d55e2000dc078e1f82b" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "43960e2b8e1331fb0e555b7c6aa12659" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 27912448, + "records": [ + { + "name": "language_model.model.layers.0.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 10752 + }, + { + "name": "language_model.model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 7236096 + }, + { + "name": "language_model.model.layers.0.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21686784 + }, + { + "name": "language_model.model.layers.0.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21697536 + }, + { + "name": "language_model.model.layers.0.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21708288 + }, + { + "name": "language_model.model.layers.0.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21719040 + }, + { + "name": "language_model.model.layers.0.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 21719296 + }, + { + "name": "language_model.model.layers.0.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 27224320 + } + ], + "md5sum": "690a7b977eeb40da3d7c8a1d06174dc0" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "b21856ff7e866e965cd28e43265b2297" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.0.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.0.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 12386560 + }, + { + "name": "language_model.model.layers.0.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 23396608 + }, + { + "name": "language_model.model.layers.0.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 24772864 + }, + { + "name": "language_model.model.layers.0.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 30277888 + } + ], + "md5sum": "47344e89d9116dd4fe9a6e1535ab602e" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 33030656, + "records": [ + { + "name": "language_model.model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.1.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.1.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14450944 + }, + { + "name": "language_model.model.layers.1.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19955968 + }, + { + "name": "language_model.model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20644096 + }, + { + "name": "language_model.model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31654144 + }, + { + "name": "language_model.model.layers.1.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030400 + } + ], + "md5sum": "26044e5e784d102a54ef85687168f1f5" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "6bef082a632c41d69afb9164efe93b55" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "6caeb682dc40fb5c36f9624806c1569c" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "c0d302e5536b5330c2031cc9eb37b2d2" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 33083904, + "records": [ + { + "name": "language_model.model.layers.1.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.1.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.1.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.1.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.1.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + }, + { + "name": "language_model.model.layers.1.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25815552 + }, + { + "name": "language_model.model.layers.1.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25826304 + }, + { + "name": "language_model.model.layers.1.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25837056 + }, + { + "name": "language_model.model.layers.2.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25847808 + }, + { + "name": "language_model.model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 25858560 + } + ], + "md5sum": "3df0e0b3278edd754c24e27abe8f809e" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.2.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.2.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.2.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.2.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.2.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.2.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.2.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "ea33c60a682b3d6eb756a63b04cc0727" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "d61506edd3bf1bc847b19fe1243993f7" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "3e099ce8413391df0be228d85ce04c67" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.2.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.2.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.2.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.2.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.3.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "10e76d4a66b04626400852fd13c994b2" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.3.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.3.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.3.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.3.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.3.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.3.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.3.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "18b48f6d4ecbfeaa98ecc659d1663838" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "cebb718a876503de5ca2419bb12a094e" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "75fabb4377b453b0b93f8037295f98d3" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.3.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.3.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.3.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.3.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.4.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "d8d40c31a5d08cf8bf5bfa94b1059beb" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.4.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.4.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.4.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.4.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.4.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.4.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.4.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "683997c0b718ed0041ba6f69bc47ec38" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "f030b36007ee5d462f03a6bb60fa85ee" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "0a5f6b247966e4463dd49c8769216cd6" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.4.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.4.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.4.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.4.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.5.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "1340675d8e922363d1192c8213ca49b4" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.5.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.5.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.5.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.5.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.5.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.5.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.5.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "9eeefc15236aba868a0e522b47c874d5" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "d8e65bad88844c4f726731eb7d036c28" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "64dc43e4b61697054bd9c13381c96ba9" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.5.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.5.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.5.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.5.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.6.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "c9267149013e6dffe8d85bcc64ec127f" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.6.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.6.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.6.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.6.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.6.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.6.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.6.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "c0c66dd90c90225acc9ece671fc2ab5b" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "1fd99a8c79e6b0c85efb04a24d63bc56" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.6.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.6.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.6.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.6.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.7.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "c0dd2c10bf48deb15fc861a4d7362936" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.7.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.7.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.7.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.7.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.7.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "7df705066afe2ba84e590c0e4a04a57e" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "e714035e4056b0553598c9f794a040b3" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "f3d152165fb501014cf5ff04946916f7" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 33417472, + "records": [ + { + "name": "language_model.model.layers.7.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.7.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.10.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.10.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 27879936 + }, + { + "name": "language_model.model.layers.10.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 27890688 + }, + { + "name": "language_model.model.layers.10.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 27901440 + }, + { + "name": "language_model.model.layers.10.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 27912192 + }, + { + "name": "language_model.model.layers.10.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 27912448 + } + ], + "md5sum": "c13809853787e62a0174f94c16ad1558" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "54c044513060e0de8c0f895fea90ba00" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 31664896, + "records": [ + { + "name": "language_model.model.layers.10.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 688128 + }, + { + "name": "language_model.model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11698176 + }, + { + "name": "language_model.model.layers.10.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 13074432 + }, + { + "name": "language_model.model.layers.10.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 13074688 + }, + { + "name": "language_model.model.layers.10.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 24084736 + }, + { + "name": "language_model.model.layers.10.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 25460992 + }, + { + "name": "language_model.model.layers.10.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 30966016 + }, + { + "name": "language_model.model.layers.11.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 31654144 + } + ], + "md5sum": "31d0f42bc3aadbb5e35288174c9799d8" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "a03e624ac13d7346f00c5f2c95139f83" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 27901696, + "records": [ + { + "name": "language_model.model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 7225344 + }, + { + "name": "language_model.model.layers.11.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21676032 + }, + { + "name": "language_model.model.layers.11.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21686784 + }, + { + "name": "language_model.model.layers.11.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21697536 + }, + { + "name": "language_model.model.layers.11.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21708288 + }, + { + "name": "language_model.model.layers.11.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 21708544 + }, + { + "name": "language_model.model.layers.11.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 27213568 + } + ], + "md5sum": "d277fe97f5730ff6ac369ca5358e9af4" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "6e0ba17e2d1b26a67b1a63ec72a04df7" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 30976768, + "records": [ + { + "name": "language_model.model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.11.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.11.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 12386560 + }, + { + "name": "language_model.model.layers.11.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 23396608 + }, + { + "name": "language_model.model.layers.11.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 24772864 + }, + { + "name": "language_model.model.layers.11.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 30277888 + }, + { + "name": "language_model.model.layers.12.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 30966016 + } + ], + "md5sum": "d8af36b207718a40c98b8ce8ca9c928e" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "e16a785d5b4c405184da660c92ff1717" + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 27901696, + "records": [ + { + "name": "language_model.model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 7225344 + }, + { + "name": "language_model.model.layers.12.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21676032 + }, + { + "name": "language_model.model.layers.12.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21686784 + }, + { + "name": "language_model.model.layers.12.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21697536 + }, + { + "name": "language_model.model.layers.12.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21708288 + }, + { + "name": "language_model.model.layers.12.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 21708544 + }, + { + "name": "language_model.model.layers.12.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 27213568 + } + ], + "md5sum": "5483f28fc7d8d92e518339939402d5fb" + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "bb7f2d6d4719b5d5b2dd1c501547786f" + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.12.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.12.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 12386560 + }, + { + "name": "language_model.model.layers.12.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 23396608 + }, + { + "name": "language_model.model.layers.12.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 24772864 + }, + { + "name": "language_model.model.layers.12.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 30277888 + } + ], + "md5sum": "d30d46dcb3b955bbde6a75aa4053521b" + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 33030656, + "records": [ + { + "name": "language_model.model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.13.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.13.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14450944 + }, + { + "name": "language_model.model.layers.13.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19955968 + }, + { + "name": "language_model.model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20644096 + }, + { + "name": "language_model.model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31654144 + }, + { + "name": "language_model.model.layers.13.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030400 + } + ], + "md5sum": "281464fe74cf9eadd631cc86c6baea98" + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "84ab1fd108f412256bd63901b75def41" + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "c577225a563dcf348f9fdafd57e214ec" + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "068b3ada05aa55914bea04856397cd12" + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 33083904, + "records": [ + { + "name": "language_model.model.layers.13.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.13.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.13.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.13.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.7.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + }, + { + "name": "language_model.model.layers.7.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25815552 + }, + { + "name": "language_model.model.layers.7.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25826304 + }, + { + "name": "language_model.model.layers.7.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25837056 + }, + { + "name": "language_model.model.layers.8.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25847808 + }, + { + "name": "language_model.model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 25858560 + } + ], + "md5sum": "d1f4f03463504ac3e8adbf63ebc5f264" + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.8.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.8.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.8.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.8.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.8.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.8.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.8.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "3ef40759ddaa0620fe68ce22016a2daf" + }, + { + "dataPath": "params_shard_51.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "3c95400f90d2b0228cbabff4fd203c7c" + }, + { + "dataPath": "params_shard_52.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "be2832d9d7d22ce492171023c5f17071" + }, + { + "dataPath": "params_shard_53.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.8.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.8.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.8.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.8.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.9.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "75814765ccf53276c468be2e015cbb2b" + }, + { + "dataPath": "params_shard_54.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.9.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.9.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.9.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.9.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.9.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.9.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.9.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "71b4eb95ac9f258efa745c98009fbc85" + }, + { + "dataPath": "params_shard_55.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "ca27990790078f159ad00ef662a82a86" + }, + { + "dataPath": "params_shard_56.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "81d77fd6f59281c71bee6461f11e67b7" + }, + { + "dataPath": "params_shard_57.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "c4d5d56690e0d91ad6db1e90cb43995a" + }, + { + "dataPath": "params_shard_58.bin", + "format": "raw-shard", + "nbytes": 33083904, + "records": [ + { + "name": "language_model.model.layers.9.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.9.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.9.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.9.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.13.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + }, + { + "name": "language_model.model.layers.13.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25815552 + }, + { + "name": "language_model.model.layers.13.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25826304 + }, + { + "name": "language_model.model.layers.13.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25837056 + }, + { + "name": "language_model.model.layers.14.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25847808 + }, + { + "name": "language_model.model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 25858560 + } + ], + "md5sum": "eb72573821333f29e76bdf037ebd9d09" + }, + { + "dataPath": "params_shard_59.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.14.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.14.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.14.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.14.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.14.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.14.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.14.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "e56a90ff5f8af8fb501eb87f70f2179d" + }, + { + "dataPath": "params_shard_60.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "767aa022153c4fd5497b9f3bb35ebb04" + }, + { + "dataPath": "params_shard_61.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "ec70e3e468b849a3eddfc330b486cac6" + }, + { + "dataPath": "params_shard_62.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.14.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.14.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.14.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.14.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.15.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "502b23a2889f0ede738704a8e9d96921" + }, + { + "dataPath": "params_shard_63.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.15.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.15.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.15.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.15.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.15.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.15.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.15.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "142a30b501e2d38d10d9eb3f9dc8cf5f" + }, + { + "dataPath": "params_shard_64.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "c1caeb434fdebe3fcc7e2be5766564b3" + }, + { + "dataPath": "params_shard_65.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "0f2eb89dfa4e97c4f27052191645f0c5" + }, + { + "dataPath": "params_shard_66.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.15.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.15.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.15.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.15.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.16.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "e35a077e02a7aa2733738c82c219914e" + }, + { + "dataPath": "params_shard_67.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.16.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.16.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.16.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.16.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.16.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.16.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.16.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "a0fd5bb875a79b23c5223f636e5f2bcc" + }, + { + "dataPath": "params_shard_68.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "aa1a956c8fb92e18ba6e79ea98b12d69" + }, + { + "dataPath": "params_shard_69.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "dc1a8139a849887df05183df4871e873" + }, + { + "dataPath": "params_shard_70.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.16.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.16.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.16.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.16.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.17.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "1780cc59b5ee4d5cd5e7a6849051fe92" + }, + { + "dataPath": "params_shard_71.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.17.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.17.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.17.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.17.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.17.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.17.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.17.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "b7dcf3c6e586c369a5eef2b3288a17d2" + }, + { + "dataPath": "params_shard_72.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "3cadcc4390ddfc16d7ec902a75a14e0a" + }, + { + "dataPath": "params_shard_73.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "084f44e6b2f44bfbe515de2c4cd66d95" + }, + { + "dataPath": "params_shard_74.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.17.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.17.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.17.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.17.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.18.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "2ad785fab251fbe7fa3f7c1ff7d7d003" + }, + { + "dataPath": "params_shard_75.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.18.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.18.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.18.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.18.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.18.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.18.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.18.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "41124ebe88f806924ce216a61f5d7f4d" + }, + { + "dataPath": "params_shard_76.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "c7e0a6cf2fa65c298996373f37e1235e" + }, + { + "dataPath": "params_shard_77.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.18.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.18.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.18.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.18.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.19.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "60264bbec1abf9089182ff9d266b9dab" + }, + { + "dataPath": "params_shard_78.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.19.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.19.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.19.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.19.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.19.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "67e6f7294dafa661a028f283c2225b85" + }, + { + "dataPath": "params_shard_79.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "d274148da316fa6b116610e7bbb9d041" + }, + { + "dataPath": "params_shard_80.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "be6897bde31d9e4acdf4eefa1df0f06c" + }, + { + "dataPath": "params_shard_81.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "ae27f42bd1d1f7dfb657067aa434660b" + }, + { + "dataPath": "params_shard_82.bin", + "format": "raw-shard", + "nbytes": 20697600, + "records": [ + { + "name": "language_model.model.layers.19.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.19.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.19.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.19.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.19.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.19.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.layers.20.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + }, + { + "name": "language_model.model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 13472256 + } + ], + "md5sum": "1d690cfc2c4e2c07536a36f383617f62" + }, + { + "dataPath": "params_shard_83.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.20.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.20.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.20.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.20.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.20.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.20.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.20.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "6c76675b78ff16f0ed5a2bbc872dbe71" + }, + { + "dataPath": "params_shard_84.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "f0d2e33196c05e0ed8bee4a13bd905cc" + }, + { + "dataPath": "params_shard_85.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "10b3f661f4d638c9e04618f0406d66c9" + }, + { + "dataPath": "params_shard_86.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.20.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.20.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.20.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.20.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.21.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "4abe7b173ca54fa7a4860fc4e3aeb689" + }, + { + "dataPath": "params_shard_87.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.21.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.21.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.21.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.21.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.21.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.21.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.21.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "1df24835e819ca98008e88db5c73917b" + }, + { + "dataPath": "params_shard_88.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "4e46854f6387520ced782e15133ad00d" + }, + { + "dataPath": "params_shard_89.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "716f2537f7b9d14dd60f15fac4bd40d8" + }, + { + "dataPath": "params_shard_90.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.21.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.21.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.21.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.21.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.22.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "972ae9bcfc46a907e1c15acdda9be810" + }, + { + "dataPath": "params_shard_91.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.22.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.22.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.22.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.22.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.22.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.22.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.22.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "6f400619946032e949a30fdb16d8aa26" + }, + { + "dataPath": "params_shard_92.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "b5e7ebb37dad751f1a9816fab059deb4" + }, + { + "dataPath": "params_shard_93.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "bcd2f882dbe844f87d0969e15427ac83" + }, + { + "dataPath": "params_shard_94.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.22.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.22.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.22.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.22.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.23.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "c0f587a8c4dfde33a3ea9b14d6214367" + }, + { + "dataPath": "params_shard_95.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.23.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.23.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.23.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.23.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.23.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.23.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.23.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "e5107edf53690a319523105f178d3345" + }, + { + "dataPath": "params_shard_96.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "7f045b97b9f6583c19b1b607df729319" + }, + { + "dataPath": "params_shard_97.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "33df0300119175b27ea4485f80bb8458" + }, + { + "dataPath": "params_shard_98.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.23.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.23.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.23.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.23.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.24.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "467dd18574a6a47ae0f0f4de7f041c1c" + }, + { + "dataPath": "params_shard_99.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.24.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.24.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.24.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.24.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.24.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.24.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.24.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "0b075098c5d8034a7f5e03694fb7a6f0" + }, + { + "dataPath": "params_shard_100.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "091c53affd60b7f2bd79e5c615325942" + }, + { + "dataPath": "params_shard_101.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.24.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.24.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.24.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.24.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.25.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "88c7c7d20277369de35ed8c1ed8d662a" + }, + { + "dataPath": "params_shard_102.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.25.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.25.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.25.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.25.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.25.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "c42c5b664e180af1035a223cbac32774" + }, + { + "dataPath": "params_shard_103.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "5954e0d8bd03c61d8ac4d1f487a2ea89" + }, + { + "dataPath": "params_shard_104.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.26.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "6950cb1b3efe66f3ddee735fa39680d2" + }, + { + "dataPath": "params_shard_105.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.26.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "af014847f14f6aa645d03374fd11b136" + }, + { + "dataPath": "params_shard_106.bin", + "format": "raw-shard", + "nbytes": 20697600, + "records": [ + { + "name": "language_model.model.layers.25.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.25.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.25.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.25.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.25.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.25.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.layers.26.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + }, + { + "name": "language_model.model.layers.26.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 13472256 + } + ], + "md5sum": "b3cccc4b0d8de357cedafd74c1697cc8" + }, + { + "dataPath": "params_shard_107.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.26.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.26.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.26.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.26.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.26.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.26.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.26.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.26.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.26.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.26.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "2f8c2832f75c7e413630233446626b76" + }, + { + "dataPath": "params_shard_108.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.27.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "7d36436f4269d6e811baf2a3d25742d2" + }, + { + "dataPath": "params_shard_109.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.27.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "890c6fbf9d680140f522bd08eb7a17a4" + }, + { + "dataPath": "params_shard_110.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.26.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.26.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.26.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.26.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.27.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.27.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "fd839976a583399332bfbcfa39ffb214" + }, + { + "dataPath": "params_shard_111.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.27.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.27.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.27.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.27.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.27.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.27.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.27.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.27.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.27.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.27.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "d6ee7b0a2793311ed3d73727258a387f" + }, + { + "dataPath": "params_shard_112.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.28.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "3fa492cb5ab8b70e6b24e94e3180c920" + }, + { + "dataPath": "params_shard_113.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.28.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "bc74c645e76fed81c9bca93cf1deb8bd" + }, + { + "dataPath": "params_shard_114.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.27.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.27.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.27.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.27.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.28.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.28.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "13f85c0cd1d726bd0f2c14ba720b4fc7" + }, + { + "dataPath": "params_shard_115.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.28.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.28.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.28.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.28.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.28.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.28.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.28.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.28.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.28.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.28.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "9504b44b6e12b992185938e19e6d1e5e" + }, + { + "dataPath": "params_shard_116.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.29.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "92c63b4825a1c38bf3cf99507d20afa5" + }, + { + "dataPath": "params_shard_117.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.29.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "9127ae721d8e4def2acb108652e0cceb" + }, + { + "dataPath": "params_shard_118.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.28.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.28.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.28.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.28.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.29.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.29.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "00c3d1750fe29bb214b1d51ef5fdec87" + }, + { + "dataPath": "params_shard_119.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.29.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.29.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.29.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.29.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.29.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.29.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.29.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.29.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.29.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.29.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "043a93165211f8fd26da42861a6b7037" + }, + { + "dataPath": "params_shard_120.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.30.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "273de4ccad5bd118cc0312aba7ceb6b6" + }, + { + "dataPath": "params_shard_121.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.30.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "529bb3eddaf7a57499c99b3e7911afc6" + }, + { + "dataPath": "params_shard_122.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.29.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.29.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.29.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.29.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.30.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.30.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "a1d8296e6ed45570c1b6fff3744da1ea" + }, + { + "dataPath": "params_shard_123.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.30.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.30.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.30.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.30.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.30.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.30.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.30.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.30.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.30.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.30.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "dd8ba3951ef35165f4bd83c7fa2203ed" + }, + { + "dataPath": "params_shard_124.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.31.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "c1edd7b73ff1e471f8562f0d1c976836" + }, + { + "dataPath": "params_shard_125.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.30.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.30.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.30.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.30.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.31.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.31.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "e7fa1fc9f278a1cf4b958736677f3cf7" + }, + { + "dataPath": "params_shard_126.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.31.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.31.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.31.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.31.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.31.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.31.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.31.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "dc25018e6663b59e2ed38b72fa53bd5e" + }, + { + "dataPath": "params_shard_127.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.31.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "da367dda9968fcec872568ebf148b1cf" + }, + { + "dataPath": "params_shard_128.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.32.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "dccdf6a1211c5bc1e0870925bf7672d7" + }, + { + "dataPath": "params_shard_129.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.32.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "c8f9bfcc187fe23ac8690c1a31d7db33" + }, + { + "dataPath": "params_shard_130.bin", + "format": "raw-shard", + "nbytes": 20697600, + "records": [ + { + "name": "language_model.model.layers.31.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.31.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.31.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.31.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.31.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.31.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.31.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.layers.32.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + }, + { + "name": "language_model.model.layers.32.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 13472256 + } + ], + "md5sum": "3b62f5a9a69a47a6ed3c0194e2af9c57" + }, + { + "dataPath": "params_shard_131.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.32.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.32.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.32.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.32.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.32.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.32.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.32.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.32.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.32.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.32.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "0c352235220135f74bba68497ed2819c" + }, + { + "dataPath": "params_shard_132.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.33.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "37372cebd5b8c4de22c51252940145b2" + }, + { + "dataPath": "params_shard_133.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.33.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "b8c27cca4a321ca5bba4552dd49fa4b4" + }, + { + "dataPath": "params_shard_134.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.32.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.32.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.32.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.32.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.33.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.33.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "ea08f53bcf2f31c1fbda09637513a30c" + }, + { + "dataPath": "params_shard_135.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.33.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.33.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.33.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.33.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.33.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.33.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.33.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.33.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.33.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.33.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "1a11d46a2d1cbec2bcb114946125d13f" + }, + { + "dataPath": "params_shard_136.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.34.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "9b9bc87c4cd8fe22bb9feb21403c350e" + }, + { + "dataPath": "params_shard_137.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.34.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "917705764289f9753602455251c390e6" + }, + { + "dataPath": "params_shard_138.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.33.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.33.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.33.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.33.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.34.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.34.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "63b55b1da1192d7a670139a4bcb53b56" + }, + { + "dataPath": "params_shard_139.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.34.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.34.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.34.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.34.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.34.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.34.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.34.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.34.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.34.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.34.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "484086429a89e15aa819cf47cef35f41" + }, + { + "dataPath": "params_shard_140.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.35.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "b3f34da3e1dbca50612111e2ac838540" + }, + { + "dataPath": "params_shard_141.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.35.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "6a15bc211924d38c9006474d26ef7e25" + }, + { + "dataPath": "params_shard_142.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.34.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.34.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.34.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.34.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.35.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.35.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "c41093d6813155f29f65df50bbee7c58" + }, + { + "dataPath": "params_shard_143.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.35.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.35.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.35.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.35.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.35.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.35.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.35.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.35.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.35.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.35.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "0a0ab5d9bba3292adb3495c8c3fec231" + }, + { + "dataPath": "params_shard_144.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.36.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "1c8f5ad7c052e8766a21834f41927602" + }, + { + "dataPath": "params_shard_145.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.36.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "b6607a3ec7ace5039a557b9afd77e0c3" + }, + { + "dataPath": "params_shard_146.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.35.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.35.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.35.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.35.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.36.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.36.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "7ea60cbfea79ad2c819d086f70caf2c8" + }, + { + "dataPath": "params_shard_147.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.36.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.36.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.36.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.36.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.36.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.36.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.36.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.36.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.36.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.36.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "c9ab3fe31fb72265297c672bc275d61c" + }, + { + "dataPath": "params_shard_148.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.37.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "c202cdf582628cc97b3cd4d37500d489" + }, + { + "dataPath": "params_shard_149.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.36.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.36.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.36.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.36.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.37.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.37.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "b5b4ae74e403d1009de685e4ed875fdf" + }, + { + "dataPath": "params_shard_150.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.37.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.37.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.37.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.37.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.37.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.37.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.37.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "a9e4eab14edbb9b4c5a9a3355c204cc0" + }, + { + "dataPath": "params_shard_151.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.37.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "9e77d3f0dc8248a2c6778702bd34447e" + }, + { + "dataPath": "params_shard_152.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.38.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "8e1dea9f14f3654e73b49cfa625e9e91" + }, + { + "dataPath": "params_shard_153.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.38.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "3d52c36386731d6b5352303a5c38bd94" + }, + { + "dataPath": "params_shard_154.bin", + "format": "raw-shard", + "nbytes": 20697600, + "records": [ + { + "name": "language_model.model.layers.37.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.37.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.37.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.37.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.37.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.37.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.37.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.layers.38.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + }, + { + "name": "language_model.model.layers.38.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 13472256 + } + ], + "md5sum": "79e4e0a73e9554e6f58ec4de68ac438a" + }, + { + "dataPath": "params_shard_155.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.38.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.38.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.38.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.38.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.38.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.38.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.38.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.38.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.38.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.38.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "0bd70d2d1d261f29303f398d93a980d2" + }, + { + "dataPath": "params_shard_156.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.39.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "0dbc4c2c84595d9e0b5ed78e830d17c7" + }, + { + "dataPath": "params_shard_157.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.39.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "4aebc51e61860d4a1b66a51364ee0f1a" + }, + { + "dataPath": "params_shard_158.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.38.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.38.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.38.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.38.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.39.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.39.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "7b7b8843b334b7de779632f48c9313ec" + }, + { + "dataPath": "params_shard_159.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.39.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.39.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.39.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.39.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.39.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.39.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.39.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.39.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.39.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.39.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "e89ca549541dc5af9ef1d266c38f7c1e" + }, + { + "dataPath": "params_shard_160.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.40.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "52a4fae3ee390c9913f400803da6c334" + }, + { + "dataPath": "params_shard_161.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.40.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "fa3628b2c6783dbd32cc5e034b2927d5" + }, + { + "dataPath": "params_shard_162.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.39.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.39.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.39.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.39.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.40.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.40.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "36dd765733c66675361daccae61d90f1" + }, + { + "dataPath": "params_shard_163.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.40.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.40.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.40.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.40.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.40.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.40.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.40.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.40.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.40.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.40.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "380237c83b84a4976a77f2337892be0b" + }, + { + "dataPath": "params_shard_164.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.41.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "e007ef5b9cf3427f4ee7feeacc726b2f" + }, + { + "dataPath": "params_shard_165.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.41.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "36cbf1ec8eae644d4ce26e3d2f0128eb" + }, + { + "dataPath": "params_shard_166.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.40.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.40.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.40.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.40.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.41.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.41.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "bb6e930f62c9c762a726d50b0f00d257" + }, + { + "dataPath": "params_shard_167.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.41.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.41.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.41.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.41.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.41.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.41.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.41.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.41.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.41.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.41.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "b47d55ec0d1aac3009a22210f246579d" + }, + { + "dataPath": "params_shard_168.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.42.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "277100b7eabe49f7e8017add4bb037ba" + }, + { + "dataPath": "params_shard_169.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.42.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "e07889d14bdc6eef4477615bf62eef55" + }, + { + "dataPath": "params_shard_170.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.41.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.41.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.41.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.41.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.42.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.42.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "6a08409795d02e52fadcbbe266b05a5e" + }, + { + "dataPath": "params_shard_171.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.42.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.42.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.42.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.42.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.42.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.42.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.42.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.42.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.42.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.42.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "53758ca174138bf910ef9091755ae35d" + }, + { + "dataPath": "params_shard_172.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.43.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "f168bd7f47649be59c6b8f45168577f6" + }, + { + "dataPath": "params_shard_173.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.42.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.42.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.42.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.42.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.43.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.43.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "1e5c7ee5a314b1580182f0e5a693330d" + }, + { + "dataPath": "params_shard_174.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.43.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.43.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.43.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.43.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.43.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.43.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.43.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "75330c7c1fcade26a58cb6a82f5daedc" + }, + { + "dataPath": "params_shard_175.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.43.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "6fe33ebc79a323e02bc347275a33cf33" + }, + { + "dataPath": "params_shard_176.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.44.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "3f987d7dc780c7e29dba153b8905f3e1" + }, + { + "dataPath": "params_shard_177.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.44.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "5a950678f0cf2e4fdcd8d86f7ce76a6d" + }, + { + "dataPath": "params_shard_178.bin", + "format": "raw-shard", + "nbytes": 20697600, + "records": [ + { + "name": "language_model.model.layers.43.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.43.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.43.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.43.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.43.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.43.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.43.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.layers.44.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + }, + { + "name": "language_model.model.layers.44.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 13472256 + } + ], + "md5sum": "f476bf7ef0e78fbb0438e5bdb6bde5ce" + }, + { + "dataPath": "params_shard_179.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.44.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.44.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.44.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.44.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.44.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.44.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.44.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.44.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.44.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.44.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "a97338943acf61bb5ac9809f0d0daa51" + }, + { + "dataPath": "params_shard_180.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.45.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "97c024902e984d6eb686e12c79bfd972" + }, + { + "dataPath": "params_shard_181.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.45.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "29becf170a77c75e193d457839803861" + }, + { + "dataPath": "params_shard_182.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.44.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.44.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.44.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.44.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.45.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.45.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "2d9e3653b779594a50aa22db5884c4e8" + }, + { + "dataPath": "params_shard_183.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.45.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.45.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.45.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.45.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.45.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.45.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.45.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.45.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.45.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.45.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "16f2db197b788c4f8ad83c7f51d059a1" + }, + { + "dataPath": "params_shard_184.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.46.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "e9d2de7214d3777ed067f0d968d82f1b" + }, + { + "dataPath": "params_shard_185.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.46.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "200b9527718d818453cd41af017c0e54" + }, + { + "dataPath": "params_shard_186.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.45.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.45.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.45.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.45.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.46.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.46.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "0fb20efffc6c024a13c471f1ff696bab" + }, + { + "dataPath": "params_shard_187.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.46.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.46.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.46.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.46.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.46.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.46.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.46.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.46.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.46.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.46.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "a08da9865a4874e4bd657326aa02c0c7" + }, + { + "dataPath": "params_shard_188.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.47.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "42c33c8e2cafda32df24e5ff69bcf5c6" + }, + { + "dataPath": "params_shard_189.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.47.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "5537444329f5fe997db3edd30b9348f8" + }, + { + "dataPath": "params_shard_190.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.46.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.46.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.46.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.46.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.47.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.47.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "05a2e062418df82fd6ff4b923983d2ce" + }, + { + "dataPath": "params_shard_191.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.47.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.47.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.47.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.47.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.47.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.47.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.47.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.47.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.47.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.47.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "90ada46c3bedecb5da028206facf1fa7" + }, + { + "dataPath": "params_shard_192.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.48.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "9493c7bfc38866a8310d6ad8e429858a" + }, + { + "dataPath": "params_shard_193.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.48.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "c96a6689e1a8241a02c5af84ffb60d1c" + }, + { + "dataPath": "params_shard_194.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.47.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.47.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.47.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.47.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.48.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.48.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "cf6f046de42ce6942a2bbffe09de864f" + }, + { + "dataPath": "params_shard_195.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.48.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.48.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.48.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.48.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.48.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.48.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.48.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.48.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.48.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.48.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "d3396069fe0e1052d7a226e393c98d82" + }, + { + "dataPath": "params_shard_196.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.49.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "8394ca94d50c525c61d9e9019005c14b" + }, + { + "dataPath": "params_shard_197.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.48.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.48.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.48.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.48.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.49.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.49.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "5f8a71b6190d212da422042c4abcac6a" + }, + { + "dataPath": "params_shard_198.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.49.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.49.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.49.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.49.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.49.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.49.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.49.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "917c40bc86351bb3471754448723829d" + }, + { + "dataPath": "params_shard_199.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.49.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "3b73435c89e99ac362490a6a84cf98ec" + }, + { + "dataPath": "params_shard_200.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.50.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "84d8ea1992abea2271f995ec0359ffc7" + }, + { + "dataPath": "params_shard_201.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.50.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "6ab7fa59bafd172d3494009828e9709c" + }, + { + "dataPath": "params_shard_202.bin", + "format": "raw-shard", + "nbytes": 20697600, + "records": [ + { + "name": "language_model.model.layers.49.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.49.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.49.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.49.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.49.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.49.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.49.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.layers.50.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + }, + { + "name": "language_model.model.layers.50.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 13472256 + } + ], + "md5sum": "94ff830a728e0428288bcebb9f6d042d" + }, + { + "dataPath": "params_shard_203.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.50.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.50.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.50.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.50.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.50.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.50.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.50.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.50.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.50.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.50.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "22cd9c6dd064382c01e39d01a61a79ec" + }, + { + "dataPath": "params_shard_204.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.51.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "515b2c8d9819af1a6bfc2a2c618992e3" + }, + { + "dataPath": "params_shard_205.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.51.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "dd69a50816dbb2807996f04e6af975b6" + }, + { + "dataPath": "params_shard_206.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.50.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.50.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.50.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.50.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.51.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.51.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "14237e14a5f38e2eff0c9bbade9d2e3d" + }, + { + "dataPath": "params_shard_207.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.51.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.51.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.51.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.51.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.51.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.51.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.51.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.51.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.51.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.51.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "b8eff42f72533cef9d5402663b438e18" + }, + { + "dataPath": "params_shard_208.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.52.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "4b7c979a6fbe8bb5e13ace2113756106" + }, + { + "dataPath": "params_shard_209.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.52.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "1f58b963d03b45618cbc46d5135c7e64" + }, + { + "dataPath": "params_shard_210.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.51.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.51.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.51.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.51.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.52.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.52.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "72bc080f6f5b8e7e90220d87b3368028" + }, + { + "dataPath": "params_shard_211.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.52.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.52.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.52.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.52.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.52.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.52.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.52.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.52.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.52.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.52.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "c958624c014693105d600b44714e6ad9" + }, + { + "dataPath": "params_shard_212.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.53.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "a5d375a712bd30aeab104b580fdacbd8" + }, + { + "dataPath": "params_shard_213.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.53.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "4797c2907d252a10ab4292e1210ee87a" + }, + { + "dataPath": "params_shard_214.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.52.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.52.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.52.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.52.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.53.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.53.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "217151cd511c3a9043599e25de917c23" + }, + { + "dataPath": "params_shard_215.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.53.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.53.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.53.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.53.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.53.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.53.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.53.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.53.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.53.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.53.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "4d3126e69f4364abc89bac380851daed" + }, + { + "dataPath": "params_shard_216.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.54.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "2e040e39bbf70a159b2f2a5f116caca5" + }, + { + "dataPath": "params_shard_217.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.54.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "529297f134e47d0bc1f985451b48dce5" + }, + { + "dataPath": "params_shard_218.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.53.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.53.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.53.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.53.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.54.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.54.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "20886a6f17d799acf2c161581678845a" + }, + { + "dataPath": "params_shard_219.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.54.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.54.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.54.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.54.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.54.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.54.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.54.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.54.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.54.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.54.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "3c915b13f1f406171c3b554b21b982c8" + }, + { + "dataPath": "params_shard_220.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.55.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "6ea41f267a6aad2353591d056935a7fd" + }, + { + "dataPath": "params_shard_221.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.54.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.54.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.54.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.54.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.55.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.55.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "7b885d447ea0827aa7779f6aa6b50e58" + }, + { + "dataPath": "params_shard_222.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.55.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.55.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.55.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.55.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.55.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.55.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.55.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "da8b917b18940c8739ff010a86cd3da6" + }, + { + "dataPath": "params_shard_223.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.55.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "9607bbf32c83fb0361a8755c66f4b4da" + }, + { + "dataPath": "params_shard_224.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.56.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "b9bc8e417b40773da1f4c67445c50a7a" + }, + { + "dataPath": "params_shard_225.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.56.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "818b5a4bd709f24ff21084e4647d9b16" + }, + { + "dataPath": "params_shard_226.bin", + "format": "raw-shard", + "nbytes": 20697600, + "records": [ + { + "name": "language_model.model.layers.55.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.55.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.55.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.55.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.55.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.55.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.55.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.layers.56.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + }, + { + "name": "language_model.model.layers.56.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 13472256 + } + ], + "md5sum": "a323a97863ddcf8d98af9a13ae86b15b" + }, + { + "dataPath": "params_shard_227.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.56.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.56.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.56.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.56.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.56.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.56.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.56.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.56.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.56.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.56.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "c82f7bfa8b6d4a14239fe52805813aef" + }, + { + "dataPath": "params_shard_228.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.57.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "17214040d557db8b0469349f2f3de6e4" + }, + { + "dataPath": "params_shard_229.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.57.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "68d0687253582907e7963f2e7c3840c3" + }, + { + "dataPath": "params_shard_230.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.56.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.56.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.56.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.56.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.57.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.57.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "ac5f37c239bbb91dbecc7ad697baee9e" + }, + { + "dataPath": "params_shard_231.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.57.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.57.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.57.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.57.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.57.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.57.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.57.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.57.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.57.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.57.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "681de413ba8ffa2289480259ac4439aa" + }, + { + "dataPath": "params_shard_232.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.58.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "c051d79b40675a550d60b4fb791de0a9" + }, + { + "dataPath": "params_shard_233.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.58.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "98a63fcb9109311605a671af8332a0ff" + }, + { + "dataPath": "params_shard_234.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.57.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.57.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.57.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.57.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.58.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.58.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "d46c558f78488604d472895a08a86cac" + }, + { + "dataPath": "params_shard_235.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.58.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.58.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.58.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.58.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.58.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.58.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.58.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.58.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.58.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.58.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "087cb6a6a3aa470856b158a3ec04b8e3" + }, + { + "dataPath": "params_shard_236.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.59.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "7d454ba3d98f619645aed9fdc1b657f8" + }, + { + "dataPath": "params_shard_237.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.59.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "d3da1a3ad7d135527fae3b40c75db395" + }, + { + "dataPath": "params_shard_238.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.58.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.58.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.58.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.58.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.59.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.59.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "c2d87053b0202f69c5b6b6ba949907b5" + }, + { + "dataPath": "params_shard_239.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.59.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.59.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.59.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.59.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.59.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.59.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.59.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.59.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.59.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.59.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "301f73a1fbbffee8e4f7597dc45457e3" + }, + { + "dataPath": "params_shard_240.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.60.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "ae4f101b8a24cd864a6828930b7d31c2" + }, + { + "dataPath": "params_shard_241.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.60.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "a89661783e309c3e601ee7754b45523b" + }, + { + "dataPath": "params_shard_242.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.59.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.59.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.59.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.59.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.60.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.60.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "1a0ffe0289ea788390cc7faa71b69fb1" + }, + { + "dataPath": "params_shard_243.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.60.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.60.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.60.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.60.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.60.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.60.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.60.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.60.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.60.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.60.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "b40fce012b4d8273db6d0bac9586a15e" + }, + { + "dataPath": "params_shard_244.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.61.mlp.gate_up_proj.q_weight", + "shape": [ + 672, + 43008 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "8910ee952b3fd6220c96eb81f6cc891e" + }, + { + "dataPath": "params_shard_245.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.60.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.60.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.60.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.60.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.61.mlp.gate_up_proj.q_scale", + "shape": [ + 168, + 43008 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.61.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "1d281711e7069f88160fab5c95db6384" + }, + { + "dataPath": "params_shard_246.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.61.self_attn.k_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.61.self_attn.k_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.61.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.61.self_attn.o_proj.q_scale", + "shape": [ + 128, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.61.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.61.self_attn.q_proj.q_weight", + "shape": [ + 672, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.61.self_attn.q_proj.q_scale", + "shape": [ + 168, + 4096 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "b2a0c1dcce300b5f7964e484f99cc8ee" + }, + { + "dataPath": "params_shard_247.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.61.mlp.down_proj.q_weight", + "shape": [ + 2688, + 5376 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "7c882068b39a58e3e5c64ebc16542989" + }, + { + "dataPath": "params_shard_248.bin", + "format": "raw-shard", + "nbytes": 13472256, + "records": [ + { + "name": "language_model.model.layers.61.self_attn.v_proj.q_weight", + "shape": [ + 672, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.61.self_attn.v_proj.q_scale", + "shape": [ + 168, + 2048 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.61.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.61.mlp.down_proj.q_scale", + "shape": [ + 672, + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.61.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.61.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.61.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.norm.weight", + "shape": [ + 5376 + ], + "dtype": "bfloat16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + } + ], + "md5sum": "b917dbf15641d5a2f0ee00eea3341ffa" + } + ] +} \ No newline at end of file