diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-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": "1e2532906bd2680ea9654eba1a5b8f1f" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 88101888, + "records": [ + { + "name": "language_model.model.embed_tokens.q_scale", + "shape": [ + 262208, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 88101888, + "byteOffset": 0 + } + ], + "md5sum": "681f5d5d19a1541ab9a8e3cc3b8feefc" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "a5efcdf762eda45de8d3a981c368a9d8" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "6107c7413bd5827113fb12809f19347c" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 27912448, + "records": [ + { + "name": "language_model.model.layers.0.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 10752 + }, + { + "name": "language_model.model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 7236096 + }, + { + "name": "language_model.model.layers.0.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21686784 + }, + { + "name": "language_model.model.layers.0.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21697536 + }, + { + "name": "language_model.model.layers.0.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21708288 + }, + { + "name": "language_model.model.layers.0.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21719040 + }, + { + "name": "language_model.model.layers.0.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 21719296 + }, + { + "name": "language_model.model.layers.0.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 27224320 + } + ], + "md5sum": "35eb2856375a9e699caab83b456eca72" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "dfdef0ad1523cf6bf3818034a2915846" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.0.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.0.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 12386560 + }, + { + "name": "language_model.model.layers.0.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 23396608 + }, + { + "name": "language_model.model.layers.0.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 24772864 + }, + { + "name": "language_model.model.layers.0.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 30277888 + } + ], + "md5sum": "5b5515c2f80bd94112ea670d7d920c8e" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 33030656, + "records": [ + { + "name": "language_model.model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.1.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.1.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14450944 + }, + { + "name": "language_model.model.layers.1.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19955968 + }, + { + "name": "language_model.model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20644096 + }, + { + "name": "language_model.model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31654144 + }, + { + "name": "language_model.model.layers.1.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030400 + } + ], + "md5sum": "604fe8212f7e2e13c7a5efff1f16ad22" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "33759af6b0c07ffdbc27f27468d51671" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "40ef4b382f7ec6de7a2c522ab95967b9" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "9f25ca6082db58e7b2aac5d21d7d68e3" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 33083904, + "records": [ + { + "name": "language_model.model.layers.1.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.1.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.1.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.1.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.1.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + }, + { + "name": "language_model.model.layers.1.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25815552 + }, + { + "name": "language_model.model.layers.1.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25826304 + }, + { + "name": "language_model.model.layers.1.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25837056 + }, + { + "name": "language_model.model.layers.2.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25847808 + }, + { + "name": "language_model.model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 25858560 + } + ], + "md5sum": "0cab04383d00146fc5e70dad99599408" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.2.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.2.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.2.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.2.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.2.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.2.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.2.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "5bb67143e4c58fb32a78aa1485c82b49" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "de8b4fee37108d2ea26b70643ffadb44" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "ddc194e868651373efa4101b8c91bacf" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.2.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.2.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.2.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.2.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.3.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "941fe1e8cc57c667b415107a60023984" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.3.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.3.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.3.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.3.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.3.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.3.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.3.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "d89733cebdf20bac7bc24cec97ce727f" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "60631c07ad828c9e63fd8f1c4ca5fae3" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "105422eb527b6e9d99f8748e606a2da8" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.3.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.3.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.3.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.3.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.4.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "5199021d597700659b370135fd42944c" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.4.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.4.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.4.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.4.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.4.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.4.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.4.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "caecaa29888e9ba03ba5d75267c40344" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "6eab7551db897226824753212ffa7dfe" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "fea9f47ccfd7b7c9d6fe54ddf8e6da1b" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.4.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.4.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.4.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.4.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.5.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "96bfb75b7422201d447c3c655e835cee" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.5.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.5.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.5.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.5.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.5.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.5.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.5.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "772de92d4610d1ec69255d6acf934969" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "cc8117206c740bdec8542c1f7b2a96b2" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "de20c7d57458b916bf54b0583ba2aa94" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.5.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.5.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.5.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.5.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.6.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "80f607c4ad65992e2bdc1a80af621791" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.6.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.6.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.6.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.6.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.6.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.6.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.6.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "d401d97a4386f8c50e2e7b7e0dae223e" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "755c009c22d5d4d839bc58215b49bf53" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.6.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.6.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.6.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.6.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.7.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "66276de033fb770fd5a281c9e814d863" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.7.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.7.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.7.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.7.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.7.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "ceda554641e6881c133e277562e6f276" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "ce6f8e039a477c2353c25c65cda5f965" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "98b6578d41cc2dbe66e37ea2df22a459" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 33417472, + "records": [ + { + "name": "language_model.model.layers.7.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.7.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.10.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.10.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 27879936 + }, + { + "name": "language_model.model.layers.10.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 27890688 + }, + { + "name": "language_model.model.layers.10.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 27901440 + }, + { + "name": "language_model.model.layers.10.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 27912192 + }, + { + "name": "language_model.model.layers.10.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 27912448 + } + ], + "md5sum": "e264fb83c07a31e42f5d7ab04d0c0b1f" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "baeb78cf752024cc6c35792e91fa2945" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 31664896, + "records": [ + { + "name": "language_model.model.layers.10.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 688128 + }, + { + "name": "language_model.model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11698176 + }, + { + "name": "language_model.model.layers.10.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 13074432 + }, + { + "name": "language_model.model.layers.10.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 13074688 + }, + { + "name": "language_model.model.layers.10.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 24084736 + }, + { + "name": "language_model.model.layers.10.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 25460992 + }, + { + "name": "language_model.model.layers.10.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 30966016 + }, + { + "name": "language_model.model.layers.11.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 31654144 + } + ], + "md5sum": "106d238167cd953ac1599fb1d087bc1e" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "56f47eb6292949d57e336546351d958b" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 27901696, + "records": [ + { + "name": "language_model.model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 7225344 + }, + { + "name": "language_model.model.layers.11.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21676032 + }, + { + "name": "language_model.model.layers.11.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21686784 + }, + { + "name": "language_model.model.layers.11.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21697536 + }, + { + "name": "language_model.model.layers.11.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21708288 + }, + { + "name": "language_model.model.layers.11.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 21708544 + }, + { + "name": "language_model.model.layers.11.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 27213568 + } + ], + "md5sum": "a3a7eeead86afe29a7ca882cd4b74f80" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "755948b648c5ec87293bd646f666a87c" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 30976768, + "records": [ + { + "name": "language_model.model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.11.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.11.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 12386560 + }, + { + "name": "language_model.model.layers.11.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 23396608 + }, + { + "name": "language_model.model.layers.11.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 24772864 + }, + { + "name": "language_model.model.layers.11.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 30277888 + }, + { + "name": "language_model.model.layers.12.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 30966016 + } + ], + "md5sum": "edc376bb2c4c88d381c5e2eb90795894" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "7c2e59d86886da0dbb62d13bfc0ec4bd" + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 27901696, + "records": [ + { + "name": "language_model.model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 7225344 + }, + { + "name": "language_model.model.layers.12.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21676032 + }, + { + "name": "language_model.model.layers.12.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21686784 + }, + { + "name": "language_model.model.layers.12.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 21697536 + }, + { + "name": "language_model.model.layers.12.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21708288 + }, + { + "name": "language_model.model.layers.12.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 21708544 + }, + { + "name": "language_model.model.layers.12.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 27213568 + } + ], + "md5sum": "ba51267f7896a7e1ac453521676b88b7" + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "c2271355331c00b897ff3647322be54a" + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.12.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.12.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 12386560 + }, + { + "name": "language_model.model.layers.12.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 23396608 + }, + { + "name": "language_model.model.layers.12.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 24772864 + }, + { + "name": "language_model.model.layers.12.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 30277888 + } + ], + "md5sum": "4b250ebd530c5f29b1e358da11510ca8" + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 33030656, + "records": [ + { + "name": "language_model.model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.13.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.13.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14450944 + }, + { + "name": "language_model.model.layers.13.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19955968 + }, + { + "name": "language_model.model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20644096 + }, + { + "name": "language_model.model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31654144 + }, + { + "name": "language_model.model.layers.13.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030400 + } + ], + "md5sum": "6ff415d9dcaa4dc7dffd1a1069162818" + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "dc03dd6efecccd94515679e0a459d53c" + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "6bfd3330ff45ae073475f19f5690ec81" + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "1742ec26e458e57918b702552fe10ad3" + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 33083904, + "records": [ + { + "name": "language_model.model.layers.13.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.13.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.13.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.13.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.7.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + }, + { + "name": "language_model.model.layers.7.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25815552 + }, + { + "name": "language_model.model.layers.7.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25826304 + }, + { + "name": "language_model.model.layers.7.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25837056 + }, + { + "name": "language_model.model.layers.8.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25847808 + }, + { + "name": "language_model.model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 25858560 + } + ], + "md5sum": "1b7fa274534ba5316f605f81d69916a8" + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.8.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.8.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.8.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.8.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.8.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.8.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.8.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "7677a2e631737419cd5771dfad0134c5" + }, + { + "dataPath": "params_shard_51.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "fd9f3eeecf1b402993bda49ec41fff0d" + }, + { + "dataPath": "params_shard_52.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "5b39be58ef6d151b7096e32db210cf79" + }, + { + "dataPath": "params_shard_53.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.8.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.8.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.8.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.8.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.9.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "2158262add4b091c1e2316e028c846f4" + }, + { + "dataPath": "params_shard_54.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.9.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.9.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.9.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.9.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.9.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.9.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.9.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "4687066bf0a406ae8241b39cc66eb475" + }, + { + "dataPath": "params_shard_55.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "716bba80cda2a8d23186f2a6e1d6066f" + }, + { + "dataPath": "params_shard_56.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "8a0e0505f7b21577f5b408b0e6a7b013" + }, + { + "dataPath": "params_shard_57.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "9b159bd37fe39b6e3edee2cde3cb767b" + }, + { + "dataPath": "params_shard_58.bin", + "format": "raw-shard", + "nbytes": 33083904, + "records": [ + { + "name": "language_model.model.layers.9.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.9.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.9.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.9.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.13.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + }, + { + "name": "language_model.model.layers.13.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25815552 + }, + { + "name": "language_model.model.layers.13.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25826304 + }, + { + "name": "language_model.model.layers.13.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25837056 + }, + { + "name": "language_model.model.layers.14.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 25847808 + }, + { + "name": "language_model.model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 25858560 + } + ], + "md5sum": "6290f6dcd5167670770f448c095ead94" + }, + { + "dataPath": "params_shard_59.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.14.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.14.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.14.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.14.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.14.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.14.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.14.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "250e512cd0d13affc173f98bedb0a955" + }, + { + "dataPath": "params_shard_60.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "cedd675b2ef32d3861f57bd8b3a6aa47" + }, + { + "dataPath": "params_shard_61.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "a13cbf6ebe66f5cc143cdded2c9bac17" + }, + { + "dataPath": "params_shard_62.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.14.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.14.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.14.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.14.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.15.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "513994501ff1b1be15400e42f476a080" + }, + { + "dataPath": "params_shard_63.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.15.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.15.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.15.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.15.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.15.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.15.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.15.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "3eb2fdc4114e579202463da16db93826" + }, + { + "dataPath": "params_shard_64.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "fbd8d10749bcb8c1ed489e7aae6572f7" + }, + { + "dataPath": "params_shard_65.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "7f8acf8e6bf47fff1f422a563589eac1" + }, + { + "dataPath": "params_shard_66.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.15.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.15.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.15.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.15.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.16.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "51e63d277277e8b3261a4866e5dd18c8" + }, + { + "dataPath": "params_shard_67.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.16.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.16.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.16.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.16.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.16.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.16.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.16.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "8322e8a0164ad88d23fceb89c4eaecfe" + }, + { + "dataPath": "params_shard_68.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "5080948ed98924a908fa2633c58d62c7" + }, + { + "dataPath": "params_shard_69.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "ab549dc8a492f2201fc30b737980acd7" + }, + { + "dataPath": "params_shard_70.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.16.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.16.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.16.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.16.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.17.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "d7572e083858c264bb67a56b45c4555f" + }, + { + "dataPath": "params_shard_71.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.17.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.17.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.17.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.17.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.17.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.17.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.17.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "eb0ec373fe8aaed4049846f02ff215af" + }, + { + "dataPath": "params_shard_72.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "4321da3c1a567278c5fa14f5718ac86a" + }, + { + "dataPath": "params_shard_73.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "f9b9ce5fef61dd8aa56af03a67e1c564" + }, + { + "dataPath": "params_shard_74.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.17.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.17.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.17.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.17.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.18.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "67d2581cc2e1f3aefdeeb52289850b8b" + }, + { + "dataPath": "params_shard_75.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.18.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.18.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.18.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.18.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.18.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.18.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.18.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "48922197f98a18d1faba68d9f5be86d1" + }, + { + "dataPath": "params_shard_76.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "8c9c674a3495d046b9062455945dbb9b" + }, + { + "dataPath": "params_shard_77.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.18.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.18.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.18.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.18.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.19.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "3467102b70ab871a0f66cbcf6dd77bc9" + }, + { + "dataPath": "params_shard_78.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.19.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.19.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.19.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.19.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.19.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "02e09b84c62f18b6206cad354cf98eb5" + }, + { + "dataPath": "params_shard_79.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "e43582aa7dde04a0874710f385f30241" + }, + { + "dataPath": "params_shard_80.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "c3749922e192b74a1c67b4dc420232c8" + }, + { + "dataPath": "params_shard_81.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "a20db6db40b361d2bee71438ed79cc2e" + }, + { + "dataPath": "params_shard_82.bin", + "format": "raw-shard", + "nbytes": 20697600, + "records": [ + { + "name": "language_model.model.layers.19.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.19.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.19.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.19.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.19.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.19.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.layers.20.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + }, + { + "name": "language_model.model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 13472256 + } + ], + "md5sum": "1b83d0945b688d9ca4032d143304cb6b" + }, + { + "dataPath": "params_shard_83.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.20.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.20.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.20.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.20.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.20.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.20.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.20.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "08a15ac0fd7a136a53aa24e8dffeee5c" + }, + { + "dataPath": "params_shard_84.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "8e9e973cf8a590af87fe6a428e8c7487" + }, + { + "dataPath": "params_shard_85.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "b4ec853319ae5005e67d43079ab33f3a" + }, + { + "dataPath": "params_shard_86.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.20.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.20.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.20.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.20.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.21.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "2ae125bd559770d7794c3309bd736179" + }, + { + "dataPath": "params_shard_87.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.21.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.21.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.21.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.21.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.21.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.21.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.21.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "f92609fe22d7292bd0c1e2fd77cd926f" + }, + { + "dataPath": "params_shard_88.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "ece3f489b2de1804e8fe257ab79cd3e7" + }, + { + "dataPath": "params_shard_89.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "5715781ea068a4d498dc7c92119ef6a7" + }, + { + "dataPath": "params_shard_90.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.21.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.21.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.21.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.21.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.22.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "2c2120829c76bce256b61f64a32299a9" + }, + { + "dataPath": "params_shard_91.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.22.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.22.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.22.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.22.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.22.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.22.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.22.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "c93cfd8644c4408c19c31399ce7a388e" + }, + { + "dataPath": "params_shard_92.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "94c9644e405550a086116da824f4dcd6" + }, + { + "dataPath": "params_shard_93.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "d57a8c6a2ac542282cb80b852f53edfd" + }, + { + "dataPath": "params_shard_94.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.22.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.22.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.22.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.22.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.23.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "0c865d7492e4773476273f4684b0f4b8" + }, + { + "dataPath": "params_shard_95.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.23.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.23.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.23.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.23.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.23.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.23.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.23.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "14e7b3a4db1bcae7ce3cbe756b0a3f2c" + }, + { + "dataPath": "params_shard_96.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "d909c0c85b54a96678c5e9f1fb3a285b" + }, + { + "dataPath": "params_shard_97.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "6c5fb2a38cbda2c40cc414093a34e269" + }, + { + "dataPath": "params_shard_98.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.23.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.23.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.23.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.23.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.24.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "d7f75bb63c4f75d7d81c45ca4b94f4cb" + }, + { + "dataPath": "params_shard_99.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.24.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.24.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.24.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.24.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.24.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.24.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.24.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "f45fc848494343927d5a276305d26ff0" + }, + { + "dataPath": "params_shard_100.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "3899745ffce46bbdc432da1740db6153" + }, + { + "dataPath": "params_shard_101.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.24.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.24.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.24.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.24.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.25.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "e4c232be94a02cbf4810fa87d528314f" + }, + { + "dataPath": "params_shard_102.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.25.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.25.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.25.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.25.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.25.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "53bc4a176af4ed3e25d9905007f1aa2a" + }, + { + "dataPath": "params_shard_103.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "abe24b3470266e5f99f812a7654bf74f" + }, + { + "dataPath": "params_shard_104.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.26.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "53dc25a38d572cb876efe18ce392d116" + }, + { + "dataPath": "params_shard_105.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.26.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "caf6b2f8b924a66bb60e15faa46ab79d" + }, + { + "dataPath": "params_shard_106.bin", + "format": "raw-shard", + "nbytes": 20697600, + "records": [ + { + "name": "language_model.model.layers.25.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.25.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.25.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.25.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.25.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.25.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.layers.26.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + }, + { + "name": "language_model.model.layers.26.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 13472256 + } + ], + "md5sum": "e69850cae1777aee2248eb9cc4929b4d" + }, + { + "dataPath": "params_shard_107.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.26.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.26.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.26.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.26.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.26.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.26.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.26.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.26.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.26.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.26.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "a06939100639236fe6a99f9f4ad44109" + }, + { + "dataPath": "params_shard_108.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.27.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "da9c482599a978a957ee01c1c9df1133" + }, + { + "dataPath": "params_shard_109.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.27.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "431c0e44b6fb6b8ebd1c0bdfddf12dee" + }, + { + "dataPath": "params_shard_110.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.26.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.26.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.26.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.26.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.27.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.27.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "5251c4ede2562829056a2103c5f2607a" + }, + { + "dataPath": "params_shard_111.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.27.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.27.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.27.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.27.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.27.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.27.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.27.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.27.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.27.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.27.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "b5d55ce3930a53f7929ad7f45cf23925" + }, + { + "dataPath": "params_shard_112.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.28.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "e01f5498d5040d9140d3d6feb4796ba4" + }, + { + "dataPath": "params_shard_113.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.28.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "18a09b60bce7f4a0d00947a3a663359f" + }, + { + "dataPath": "params_shard_114.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.27.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.27.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.27.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.27.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.28.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.28.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "fd297b26751d0541aaeb28f985d05ef5" + }, + { + "dataPath": "params_shard_115.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.28.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.28.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.28.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.28.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.28.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.28.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.28.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.28.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.28.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.28.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "3291410614623b940e36aa25f149da47" + }, + { + "dataPath": "params_shard_116.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.29.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "395267de595cfbee068bf0054b5d51c1" + }, + { + "dataPath": "params_shard_117.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.29.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "606a18300112e5c68bb28a90714fb488" + }, + { + "dataPath": "params_shard_118.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.28.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.28.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.28.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.28.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.29.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.29.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "7dda9684fed4a43308dcaf2362261f05" + }, + { + "dataPath": "params_shard_119.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.29.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.29.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.29.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.29.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.29.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.29.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.29.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.29.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.29.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.29.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "8f4e1f4a9fd88a7d55220ac5b252d6ad" + }, + { + "dataPath": "params_shard_120.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.30.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "d1619dcc0426d00e833d00cbad6439be" + }, + { + "dataPath": "params_shard_121.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.30.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "b73e2e31c2495a09563c650752e8abb9" + }, + { + "dataPath": "params_shard_122.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.29.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.29.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.29.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.29.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.30.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.30.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "101568e2bd90942b03514e8d0d46fdb5" + }, + { + "dataPath": "params_shard_123.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.30.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.30.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.30.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.30.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.30.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.30.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.30.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.30.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.30.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.30.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "c5324747473a28365c1d281e47455cd1" + }, + { + "dataPath": "params_shard_124.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.31.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "06b06accde8e20ac2918d430f88d4c3a" + }, + { + "dataPath": "params_shard_125.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.30.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.30.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.30.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.30.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.31.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.31.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "b70a6c6811b94170d80a999b9fc2f29e" + }, + { + "dataPath": "params_shard_126.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.31.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.31.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.31.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.31.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.31.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.31.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.31.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "d134b8ca9e2cfa6029fa078ee49a9816" + }, + { + "dataPath": "params_shard_127.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.31.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "918ea887614e52db96d7bd30e3c9a47f" + }, + { + "dataPath": "params_shard_128.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.32.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "4ab94b467403f34916a8dd11c9b75553" + }, + { + "dataPath": "params_shard_129.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.32.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "483136ec545e430a0ba94ae51294942b" + }, + { + "dataPath": "params_shard_130.bin", + "format": "raw-shard", + "nbytes": 20697600, + "records": [ + { + "name": "language_model.model.layers.31.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.31.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.31.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.31.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.31.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.31.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.31.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.layers.32.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + }, + { + "name": "language_model.model.layers.32.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 13472256 + } + ], + "md5sum": "76dcd253a58fa9d67d984bff1c7e5936" + }, + { + "dataPath": "params_shard_131.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.32.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.32.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.32.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.32.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.32.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.32.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.32.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.32.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.32.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.32.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "ecf25d19c37552773d84546efb05b87a" + }, + { + "dataPath": "params_shard_132.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.33.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "2f52f716e999ffc3847d9ba1adc85257" + }, + { + "dataPath": "params_shard_133.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.33.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "0ab3f45d68347cf6ebb6a0e14215732e" + }, + { + "dataPath": "params_shard_134.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.32.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.32.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.32.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.32.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.33.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.33.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "d046d61adf21297ddd09b4cc39f115b1" + }, + { + "dataPath": "params_shard_135.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.33.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.33.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.33.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.33.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.33.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.33.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.33.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.33.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.33.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.33.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "8d35e9d2e09439f8c7020bb66aa50fe5" + }, + { + "dataPath": "params_shard_136.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.34.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "4c940c6de7eaa29586d76fbc396952ff" + }, + { + "dataPath": "params_shard_137.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.34.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "fce5240d8042c96780c224b94c63a21c" + }, + { + "dataPath": "params_shard_138.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.33.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.33.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.33.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.33.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.34.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.34.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "73e79e1479de0d6cbb39c4de3dfbd3bd" + }, + { + "dataPath": "params_shard_139.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.34.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.34.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.34.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.34.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.34.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.34.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.34.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.34.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.34.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.34.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "1e8ab7d76c015c652083f1e9557474d9" + }, + { + "dataPath": "params_shard_140.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.35.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "00d6e968fc80bea42f5ffbcd16fb61ac" + }, + { + "dataPath": "params_shard_141.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.35.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "889212477540ef8a2b4500da89c6b0e0" + }, + { + "dataPath": "params_shard_142.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.34.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.34.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.34.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.34.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.35.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.35.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "0b489bcb4b620682ed3537fdefd9acdb" + }, + { + "dataPath": "params_shard_143.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.35.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.35.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.35.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.35.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.35.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.35.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.35.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.35.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.35.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.35.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "ae04d0afa5978196f1713c088ac6556c" + }, + { + "dataPath": "params_shard_144.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.36.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "62e698c3acc4fada8fb3517b0de6fea1" + }, + { + "dataPath": "params_shard_145.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.36.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "11dc0473a88f1d9471bc884f4541c5be" + }, + { + "dataPath": "params_shard_146.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.35.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.35.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.35.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.35.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.36.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.36.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "f39b92d2fc6eeec381bd33504b51111e" + }, + { + "dataPath": "params_shard_147.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.36.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.36.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.36.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.36.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.36.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.36.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.36.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.36.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.36.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.36.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "03319a780c3daffe4a1c5f1a1ca28d86" + }, + { + "dataPath": "params_shard_148.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.37.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "9daf2cf1fa09a303912a64971ce43873" + }, + { + "dataPath": "params_shard_149.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.36.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.36.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.36.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.36.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.37.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.37.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "048d3025a1a823f4666b4a2728c6a688" + }, + { + "dataPath": "params_shard_150.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.37.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.37.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.37.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.37.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.37.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.37.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.37.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "3dbc876d9d87d48fc3944deddcf03181" + }, + { + "dataPath": "params_shard_151.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.37.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "bf5e1f0b91f3474d57a3a3414d80597d" + }, + { + "dataPath": "params_shard_152.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.38.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "8d214d9c2c43260405297e35b47e634a" + }, + { + "dataPath": "params_shard_153.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.38.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "3d4ff2bf79b0cf2f2b1143fd83936220" + }, + { + "dataPath": "params_shard_154.bin", + "format": "raw-shard", + "nbytes": 20697600, + "records": [ + { + "name": "language_model.model.layers.37.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.37.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.37.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.37.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.37.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.37.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.37.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.layers.38.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + }, + { + "name": "language_model.model.layers.38.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 13472256 + } + ], + "md5sum": "e2db6867e04cf727d6dbc753e196d96c" + }, + { + "dataPath": "params_shard_155.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.38.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.38.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.38.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.38.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.38.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.38.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.38.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.38.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.38.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.38.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "3b143aa451fb8ff39b0e374ae103cae4" + }, + { + "dataPath": "params_shard_156.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.39.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "46ec5783a87fe177fc6c7bc49a0744d1" + }, + { + "dataPath": "params_shard_157.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.39.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "7f07ec42414e08559d4ecb2e3f9c2ed8" + }, + { + "dataPath": "params_shard_158.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.38.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.38.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.38.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.38.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.39.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.39.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "4fba9751d6248d26f0d1d196b9e1d961" + }, + { + "dataPath": "params_shard_159.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.39.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.39.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.39.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.39.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.39.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.39.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.39.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.39.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.39.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.39.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "05d48a67854de61587052e3b885c0217" + }, + { + "dataPath": "params_shard_160.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.40.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "84b41e9a1e0ac2e8405be70faa6ac814" + }, + { + "dataPath": "params_shard_161.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.40.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "32e928cad86c8883497555d5618d96e7" + }, + { + "dataPath": "params_shard_162.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.39.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.39.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.39.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.39.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.40.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.40.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "b4db99560839636266902cb5727dd8d9" + }, + { + "dataPath": "params_shard_163.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.40.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.40.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.40.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.40.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.40.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.40.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.40.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.40.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.40.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.40.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "cae26ad7d38d662b7f3d8ba2efec4068" + }, + { + "dataPath": "params_shard_164.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.41.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "9423a4a8f576cbafc09de87ee66ad64b" + }, + { + "dataPath": "params_shard_165.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.41.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "144e4dd5590594ed3d5b1b63d02dca4a" + }, + { + "dataPath": "params_shard_166.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.40.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.40.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.40.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.40.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.41.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.41.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "e03479a5b4282d2ef608e5cd350bac70" + }, + { + "dataPath": "params_shard_167.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.41.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.41.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.41.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.41.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.41.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.41.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.41.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.41.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.41.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.41.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "5fd81c56ef0e911b94e565084aef5aaf" + }, + { + "dataPath": "params_shard_168.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.42.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "f4d5f93974c882620db6dddf02051ba1" + }, + { + "dataPath": "params_shard_169.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.42.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "757df6c532af23f3d010074bad5263f1" + }, + { + "dataPath": "params_shard_170.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.41.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.41.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.41.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.41.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.42.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.42.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "6ece945d08961ea0f3839d95ddbcf972" + }, + { + "dataPath": "params_shard_171.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.42.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.42.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.42.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.42.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.42.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.42.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.42.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.42.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.42.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.42.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "da44387ea716f7a2b408de538ff5d49f" + }, + { + "dataPath": "params_shard_172.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.43.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "0584cf041b3064f6005a6681d9634533" + }, + { + "dataPath": "params_shard_173.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.42.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.42.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.42.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.42.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.43.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.43.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "fb231532aa5d0a656050aa67f353045f" + }, + { + "dataPath": "params_shard_174.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.43.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.43.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.43.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.43.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.43.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.43.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.43.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "660585b420412db5a8823fe29ac4a64e" + }, + { + "dataPath": "params_shard_175.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.43.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "1c916847a30fd3002fc7a7a4572f2f6a" + }, + { + "dataPath": "params_shard_176.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.44.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "6e70aaf5d8de5b23dcfad46a667cedbb" + }, + { + "dataPath": "params_shard_177.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.44.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "9311b7565ccd6551dbfb10b384684a1b" + }, + { + "dataPath": "params_shard_178.bin", + "format": "raw-shard", + "nbytes": 20697600, + "records": [ + { + "name": "language_model.model.layers.43.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.43.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.43.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.43.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.43.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.43.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.43.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.layers.44.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + }, + { + "name": "language_model.model.layers.44.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 13472256 + } + ], + "md5sum": "6b20f4581054ad60941d5ef4b6fa655a" + }, + { + "dataPath": "params_shard_179.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.44.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.44.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.44.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.44.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.44.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.44.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.44.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.44.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.44.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.44.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "5527de5c057faf7ee99677c4dce57744" + }, + { + "dataPath": "params_shard_180.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.45.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "6edaf39f3af76916a8a46cb8602ec028" + }, + { + "dataPath": "params_shard_181.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.45.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "726a15a06fb9b5155eb672555e97f563" + }, + { + "dataPath": "params_shard_182.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.44.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.44.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.44.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.44.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.45.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.45.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "cdc0446395109d49326cf66aeaff5dfc" + }, + { + "dataPath": "params_shard_183.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.45.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.45.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.45.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.45.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.45.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.45.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.45.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.45.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.45.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.45.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "680791b7956d03ee9272aabf762c931f" + }, + { + "dataPath": "params_shard_184.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.46.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "fe74c46a2ebe6630639f99d4a31b3a29" + }, + { + "dataPath": "params_shard_185.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.46.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "ac669fd0c47b2390b0a01c46b077de05" + }, + { + "dataPath": "params_shard_186.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.45.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.45.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.45.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.45.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.46.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.46.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "c967eb53688260dba1337e4df267f2f1" + }, + { + "dataPath": "params_shard_187.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.46.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.46.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.46.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.46.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.46.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.46.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.46.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.46.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.46.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.46.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "eeb64054b4d2ff11e1ad3db12b2851e5" + }, + { + "dataPath": "params_shard_188.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.47.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "75058fe8cd9556ddce88353fa93a91ff" + }, + { + "dataPath": "params_shard_189.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.47.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "24f3aff51c783813be4fc5e6f9cbc66d" + }, + { + "dataPath": "params_shard_190.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.46.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.46.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.46.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.46.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.47.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.47.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "ec7d36e651d6492fc962fea672d4da63" + }, + { + "dataPath": "params_shard_191.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.47.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.47.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.47.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.47.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.47.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.47.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.47.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.47.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.47.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.47.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "0dffca4e00ccd9ff2b5a1720ead42f3f" + }, + { + "dataPath": "params_shard_192.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.48.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "20ad8d789137a6a602ee6e5da553d485" + }, + { + "dataPath": "params_shard_193.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.48.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "26bc634aaefd3a7e37219824431ec5cb" + }, + { + "dataPath": "params_shard_194.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.47.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.47.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.47.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.47.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.48.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.48.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "443ace1688377df91b1b5f6d4bcb4609" + }, + { + "dataPath": "params_shard_195.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.48.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.48.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.48.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.48.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.48.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.48.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.48.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.48.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.48.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.48.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "654709bd67075c81a858430f42d0a2af" + }, + { + "dataPath": "params_shard_196.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.49.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "e20a4f7fedc946e6f6dd243891c8d9c5" + }, + { + "dataPath": "params_shard_197.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.48.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.48.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.48.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.48.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.49.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.49.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "12e30a69985ce7f6ae6a58570ced30e2" + }, + { + "dataPath": "params_shard_198.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.49.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.49.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.49.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.49.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.49.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.49.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.49.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "6620dda648c6479529af67059958e86d" + }, + { + "dataPath": "params_shard_199.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.49.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "66c03826a7d0a840e54b17776a0234b9" + }, + { + "dataPath": "params_shard_200.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.50.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "6c577e7a5df1d860933101f73959e94a" + }, + { + "dataPath": "params_shard_201.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.50.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "68561754d0679a96c17269f1df04ff13" + }, + { + "dataPath": "params_shard_202.bin", + "format": "raw-shard", + "nbytes": 20697600, + "records": [ + { + "name": "language_model.model.layers.49.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.49.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.49.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.49.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.49.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.49.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.49.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.layers.50.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + }, + { + "name": "language_model.model.layers.50.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 13472256 + } + ], + "md5sum": "871ae843e79ce2f59b1c13ee32f6f2b4" + }, + { + "dataPath": "params_shard_203.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.50.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.50.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.50.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.50.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.50.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.50.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.50.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.50.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.50.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.50.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "df76942163c4f009464170a3ce6dea46" + }, + { + "dataPath": "params_shard_204.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.51.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "0af7686162d5d4e6355a4780d2795e1c" + }, + { + "dataPath": "params_shard_205.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.51.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "fb7d88b0d119d70924be39bcd7d23964" + }, + { + "dataPath": "params_shard_206.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.50.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.50.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.50.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.50.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.51.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.51.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "fc731d4de00ff7577aae41363de3809a" + }, + { + "dataPath": "params_shard_207.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.51.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.51.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.51.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.51.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.51.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.51.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.51.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.51.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.51.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.51.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "5218af2b77f9031f5cd5b75664947d16" + }, + { + "dataPath": "params_shard_208.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.52.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "fc4369f591722e6f63cb95279a7d4064" + }, + { + "dataPath": "params_shard_209.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.52.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "f66b4d3aba24afa5ba81ab21c7c8567c" + }, + { + "dataPath": "params_shard_210.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.51.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.51.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.51.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.51.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.52.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.52.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "c76ce4be9bb7c1e3fb85ef5e52702d69" + }, + { + "dataPath": "params_shard_211.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.52.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.52.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.52.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.52.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.52.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.52.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.52.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.52.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.52.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.52.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "e8620744fb17e6026843431797b9c123" + }, + { + "dataPath": "params_shard_212.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.53.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "cb717937373b0f538bc23d2d6b52f5f8" + }, + { + "dataPath": "params_shard_213.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.53.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "2bcb6e5344fd4d8c9ba4b6ae1664066e" + }, + { + "dataPath": "params_shard_214.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.52.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.52.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.52.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.52.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.53.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.53.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "7aa343ca0e714374f7dc97171279b1af" + }, + { + "dataPath": "params_shard_215.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.53.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.53.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.53.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.53.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.53.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.53.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.53.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.53.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.53.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.53.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "39f02c6c8480e11885711af4156bf982" + }, + { + "dataPath": "params_shard_216.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.54.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "3995ab97a37b4ec09c913850345d81e2" + }, + { + "dataPath": "params_shard_217.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.54.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "79970acd8fea7917f679109b370a58a6" + }, + { + "dataPath": "params_shard_218.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.53.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.53.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.53.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.53.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.54.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.54.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "41cdd54c57519ed8805e36466a7d25ca" + }, + { + "dataPath": "params_shard_219.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.54.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.54.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.54.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.54.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.54.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.54.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.54.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.54.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.54.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.54.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "546dc140458f949e79aa4a351307df46" + }, + { + "dataPath": "params_shard_220.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.55.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "7268c016c38844a823d0518ae39116c3" + }, + { + "dataPath": "params_shard_221.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.54.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.54.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.54.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.54.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.55.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.55.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "0f830bccf5830d2f414f304e96b47d65" + }, + { + "dataPath": "params_shard_222.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.55.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.55.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.55.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.55.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.55.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.55.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.55.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "9d0ff5dc0c1db631952457f674827eb5" + }, + { + "dataPath": "params_shard_223.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.55.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "ca3ca489cfbce9048d96c389ae2e0c5f" + }, + { + "dataPath": "params_shard_224.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.56.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "1fe52db468a2c36c29f9ba6194e29553" + }, + { + "dataPath": "params_shard_225.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.56.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "1d5605d9729b214b5ff347b7ec7272ba" + }, + { + "dataPath": "params_shard_226.bin", + "format": "raw-shard", + "nbytes": 20697600, + "records": [ + { + "name": "language_model.model.layers.55.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.55.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.55.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.55.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.55.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.55.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.55.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.layers.56.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + }, + { + "name": "language_model.model.layers.56.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 13472256 + } + ], + "md5sum": "59a5359faa55b735f7a1fcc2eaf0f3b3" + }, + { + "dataPath": "params_shard_227.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.56.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.56.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.56.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.56.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.56.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.56.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.56.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.56.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.56.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.56.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "431ad7c7c3fbaa6f7adfd5c377e4bff9" + }, + { + "dataPath": "params_shard_228.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.57.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "76cded8d475d897832b64c4f655ec55f" + }, + { + "dataPath": "params_shard_229.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.57.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "62ca8287839b6df34f9d31e572d08107" + }, + { + "dataPath": "params_shard_230.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.56.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.56.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.56.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.56.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.57.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.57.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "eb64a6aaa0c09be796270bc5752f915b" + }, + { + "dataPath": "params_shard_231.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.57.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.57.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.57.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.57.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.57.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.57.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.57.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.57.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.57.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.57.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "51fef849c1ed5a64aa4e32d92ecec8f5" + }, + { + "dataPath": "params_shard_232.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.58.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "533fc6d01ac1d8661c520ba8238324af" + }, + { + "dataPath": "params_shard_233.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.58.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "aca322e0512450de244966fa344f7806" + }, + { + "dataPath": "params_shard_234.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.57.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.57.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.57.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.57.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.58.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.58.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "4d01e9b2dc17867546451588e8b325a9" + }, + { + "dataPath": "params_shard_235.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.58.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.58.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.58.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.58.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.58.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.58.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.58.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.58.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.58.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.58.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "c883067eb009c308732f49df53d9ae28" + }, + { + "dataPath": "params_shard_236.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.59.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "0f9b23c5c86f80668e52537756bd8f30" + }, + { + "dataPath": "params_shard_237.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.59.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "515fe124866f01c11fb9817abc961fe0" + }, + { + "dataPath": "params_shard_238.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.58.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.58.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.58.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.58.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.59.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.59.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "5f951c7159633711c8b74ac1374dddb0" + }, + { + "dataPath": "params_shard_239.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.59.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.59.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.59.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.59.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.59.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.59.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.59.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.59.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.59.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.59.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "cc44cf43c8f312b4e1dea7f0a4d4c744" + }, + { + "dataPath": "params_shard_240.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.60.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "643c842e1ad650f5b4e9c04eecf84123" + }, + { + "dataPath": "params_shard_241.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.60.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "538dee75b07441e7e7888b20c66c5f07" + }, + { + "dataPath": "params_shard_242.bin", + "format": "raw-shard", + "nbytes": 25815552, + "records": [ + { + "name": "language_model.model.layers.59.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.59.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.59.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.59.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.60.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.60.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 18590208 + } + ], + "md5sum": "b1d8f45516f74bf2b61f7fa89894137b" + }, + { + "dataPath": "params_shard_243.bin", + "format": "raw-shard", + "nbytes": 33062912, + "records": [ + { + "name": "language_model.model.layers.60.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.60.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14450688 + }, + { + "name": "language_model.model.layers.60.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14461440 + }, + { + "name": "language_model.model.layers.60.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 14472192 + }, + { + "name": "language_model.model.layers.60.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14482944 + }, + { + "name": "language_model.model.layers.60.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 14483200 + }, + { + "name": "language_model.model.layers.60.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 19988224 + }, + { + "name": "language_model.model.layers.60.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 20676352 + }, + { + "name": "language_model.model.layers.60.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 31686400 + }, + { + "name": "language_model.model.layers.60.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33062656 + } + ], + "md5sum": "362dd5d4b90ffe990b40ddc675ecf731" + }, + { + "dataPath": "params_shard_244.bin", + "format": "raw-shard", + "nbytes": 115605504, + "records": [ + { + "name": "language_model.model.layers.61.mlp.gate_up_proj.q_weight", + "shape": [ + 43008, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 115605504, + "byteOffset": 0 + } + ], + "md5sum": "8e49e5414edc3140cdbdbd297efc8239" + }, + { + "dataPath": "params_shard_245.bin", + "format": "raw-shard", + "nbytes": 33030400, + "records": [ + { + "name": "language_model.model.layers.60.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.60.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 11010048 + }, + { + "name": "language_model.model.layers.60.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 12386304 + }, + { + "name": "language_model.model.layers.60.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 17891328 + }, + { + "name": "language_model.model.layers.61.mlp.gate_up_proj.q_scale", + "shape": [ + 43008, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14450688, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.61.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 33030144 + } + ], + "md5sum": "7caed423eef7cfe04782c98e25f942f7" + }, + { + "dataPath": "params_shard_246.bin", + "format": "raw-shard", + "nbytes": 30966016, + "records": [ + { + "name": "language_model.model.layers.61.self_attn.k_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.61.self_attn.k_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.61.self_attn.o_proj.q_weight", + "shape": [ + 5376, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.61.self_attn.o_proj.q_scale", + "shape": [ + 5376, + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 17203200 + }, + { + "name": "language_model.model.layers.61.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 18579456 + }, + { + "name": "language_model.model.layers.61.self_attn.q_proj.q_weight", + "shape": [ + 4096, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11010048, + "byteOffset": 18579712 + }, + { + "name": "language_model.model.layers.61.self_attn.q_proj.q_scale", + "shape": [ + 4096, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1376256, + "byteOffset": 29589760 + } + ], + "md5sum": "36714c8e96227dadcb39c49d9967532a" + }, + { + "dataPath": "params_shard_247.bin", + "format": "raw-shard", + "nbytes": 57802752, + "records": [ + { + "name": "language_model.model.layers.61.mlp.down_proj.q_weight", + "shape": [ + 5376, + 2688 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 57802752, + "byteOffset": 0 + } + ], + "md5sum": "667e889f21b6cba383fcd0ebce4da771" + }, + { + "dataPath": "params_shard_248.bin", + "format": "raw-shard", + "nbytes": 13472256, + "records": [ + { + "name": "language_model.model.layers.61.self_attn.v_proj.q_weight", + "shape": [ + 2048, + 672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "language_model.model.layers.61.self_attn.v_proj.q_scale", + "shape": [ + 2048, + 168 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "language_model.model.layers.61.input_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 6193152 + }, + { + "name": "language_model.model.layers.61.mlp.down_proj.q_scale", + "shape": [ + 5376, + 672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7225344, + "byteOffset": 6203904 + }, + { + "name": "language_model.model.layers.61.post_attention_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13429248 + }, + { + "name": "language_model.model.layers.61.post_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13440000 + }, + { + "name": "language_model.model.layers.61.pre_feedforward_layernorm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13450752 + }, + { + "name": "language_model.model.norm.weight", + "shape": [ + 5376 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10752, + "byteOffset": 13461504 + } + ], + "md5sum": "7a2e751dbee863813e736817810bf9a6" + } + ] +} \ No newline at end of file