diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,5625 @@ +{ + "metadata": { + "ParamSize": 486, + "ParamBytes": 1791006720.0, + "BitsPerParam": 4.728151176594026 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "lm_head.q_weight", + "shape": [ + 49152, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "941234c45c7316f3f9c7b311442ab08c" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.embed_tokens.q_weight", + "shape": [ + 49152, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "941234c45c7316f3f9c7b311442ab08c" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.0.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "4f418c75b0d507744dffacd0ec7ff386" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.0.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "7d21e4ccd49b4cacc5a3a964370093db" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 29854720, + "records": [ + { + "name": "lm_head.q_scale", + "shape": [ + 49152, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 0 + }, + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 49152, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 9437184 + }, + { + "name": "model.layers.0.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 18874368 + }, + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 18880512 + }, + { + "name": "model.layers.0.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 18886656 + }, + { + "name": "model.layers.0.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 18911232 + }, + { + "name": "model.layers.0.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 21270528 + }, + { + "name": "model.layers.0.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 21276672 + }, + { + "name": "model.layers.0.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23635968 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23642112 + }, + { + "name": "model.layers.0.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 23648256 + }, + { + "name": "model.layers.0.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 23655424 + }, + { + "name": "model.layers.0.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 29160448 + }, + { + "name": "model.layers.0.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29848576 + } + ], + "md5sum": "2c6ebcd4aa48a6079676fac6359ec3a4" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.1.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "e9c041335243bd4caae1d19cab45659b" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 28963840, + "records": [ + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 4718592 + }, + { + "name": "model.layers.1.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5308416 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5314560 + }, + { + "name": "model.layers.1.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 5320704 + }, + { + "name": "model.layers.1.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 5345280 + }, + { + "name": "model.layers.1.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 24219648 + }, + { + "name": "model.layers.1.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 26578944 + }, + { + "name": "model.layers.1.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 26585088 + }, + { + "name": "model.layers.1.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28944384 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28950528 + }, + { + "name": "model.layers.1.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 28956672 + } + ], + "md5sum": "775a61ffc467fce90933f6a4580c8e50" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.10.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "cd40f50de2c523081cfb518eea342db4" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 32784384, + "records": [ + { + "name": "model.layers.1.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "model.layers.1.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "model.layers.1.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 6193152 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 6199296 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 10917888 + }, + { + "name": "model.layers.10.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11507712 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11513856 + }, + { + "name": "model.layers.10.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 11520000 + }, + { + "name": "model.layers.10.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 11544576 + }, + { + "name": "model.layers.10.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30418944 + }, + { + "name": "model.layers.10.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 32778240 + } + ], + "md5sum": "182576f16ad3eff797b62b77747a0842" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 32797696, + "records": [ + { + "name": "model.layers.10.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.10.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2365440 + }, + { + "name": "model.layers.10.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2371584 + }, + { + "name": "model.layers.10.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 2378752 + }, + { + "name": "model.layers.10.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 7883776 + }, + { + "name": "model.layers.10.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 8571904 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 8578048 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 13296640 + }, + { + "name": "model.layers.11.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13886464 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13892608 + }, + { + "name": "model.layers.11.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 13898752 + }, + { + "name": "model.layers.11.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 13923328 + } + ], + "md5sum": "bbf4bcb956e00bdea89b2a16f983e61c" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 29817856, + "records": [ + { + "name": "model.layers.11.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.11.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.11.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 2365440 + }, + { + "name": "model.layers.11.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 21239808 + }, + { + "name": "model.layers.11.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23605248 + }, + { + "name": "model.layers.11.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 23611392 + }, + { + "name": "model.layers.11.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 23618560 + }, + { + "name": "model.layers.11.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 29123584 + }, + { + "name": "model.layers.11.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29811712 + } + ], + "md5sum": "7dec6ad3ff0300146eeb72aa0f7e00b3" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.12.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "14b37a3bafc5c322d783b0a5635145e6" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 28963840, + "records": [ + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 4718592 + }, + { + "name": "model.layers.12.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5308416 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5314560 + }, + { + "name": "model.layers.12.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 5320704 + }, + { + "name": "model.layers.12.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 5345280 + }, + { + "name": "model.layers.12.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 24219648 + }, + { + "name": "model.layers.12.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 26578944 + }, + { + "name": "model.layers.12.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 26585088 + }, + { + "name": "model.layers.12.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28944384 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28950528 + }, + { + "name": "model.layers.12.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 28956672 + } + ], + "md5sum": "9307a70e49d8aa12e99f40cb1d6bb99a" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.13.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "b09c254b10e75882d2c5c4cf4fd82bfa" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 32784384, + "records": [ + { + "name": "model.layers.12.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "model.layers.12.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "model.layers.12.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 6193152 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 6199296 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 10917888 + }, + { + "name": "model.layers.13.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11507712 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11513856 + }, + { + "name": "model.layers.13.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 11520000 + }, + { + "name": "model.layers.13.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 11544576 + }, + { + "name": "model.layers.13.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30418944 + }, + { + "name": "model.layers.13.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 32778240 + } + ], + "md5sum": "fe3de2f30082594543c2bbf05afc0ac1" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 32797696, + "records": [ + { + "name": "model.layers.13.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.13.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2365440 + }, + { + "name": "model.layers.13.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2371584 + }, + { + "name": "model.layers.13.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 2378752 + }, + { + "name": "model.layers.13.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 7883776 + }, + { + "name": "model.layers.13.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 8571904 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 8578048 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 13296640 + }, + { + "name": "model.layers.14.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13886464 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13892608 + }, + { + "name": "model.layers.14.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 13898752 + }, + { + "name": "model.layers.14.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 13923328 + } + ], + "md5sum": "3703fa39adb82be45e9c562682477c84" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 29817856, + "records": [ + { + "name": "model.layers.14.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.14.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.14.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 2365440 + }, + { + "name": "model.layers.14.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 21239808 + }, + { + "name": "model.layers.14.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23605248 + }, + { + "name": "model.layers.14.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 23611392 + }, + { + "name": "model.layers.14.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 23618560 + }, + { + "name": "model.layers.14.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 29123584 + }, + { + "name": "model.layers.14.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29811712 + } + ], + "md5sum": "4a26789e16eb005f32d188950c684780" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.15.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "e994fc6c64939d4b54a0ac5eb9b51827" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 28963840, + "records": [ + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 4718592 + }, + { + "name": "model.layers.15.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5308416 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5314560 + }, + { + "name": "model.layers.15.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 5320704 + }, + { + "name": "model.layers.15.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 5345280 + }, + { + "name": "model.layers.15.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 24219648 + }, + { + "name": "model.layers.15.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 26578944 + }, + { + "name": "model.layers.15.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 26585088 + }, + { + "name": "model.layers.15.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28944384 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28950528 + }, + { + "name": "model.layers.15.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 28956672 + } + ], + "md5sum": "7f589cdc785458e5dcf9edff6a73a5d4" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.16.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "98e14cd494b11cc9fc8ecc9f3d779910" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 32784384, + "records": [ + { + "name": "model.layers.15.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "model.layers.15.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "model.layers.15.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 6193152 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 6199296 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 10917888 + }, + { + "name": "model.layers.16.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11507712 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11513856 + }, + { + "name": "model.layers.16.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 11520000 + }, + { + "name": "model.layers.16.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 11544576 + }, + { + "name": "model.layers.16.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30418944 + }, + { + "name": "model.layers.16.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 32778240 + } + ], + "md5sum": "f6a519feef268d72705e094903a6b1ef" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 32797696, + "records": [ + { + "name": "model.layers.16.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.16.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2365440 + }, + { + "name": "model.layers.16.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2371584 + }, + { + "name": "model.layers.16.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 2378752 + }, + { + "name": "model.layers.16.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 7883776 + }, + { + "name": "model.layers.16.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 8571904 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 8578048 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 13296640 + }, + { + "name": "model.layers.17.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13886464 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13892608 + }, + { + "name": "model.layers.17.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 13898752 + }, + { + "name": "model.layers.17.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 13923328 + } + ], + "md5sum": "df942b969b51fc2b6567d3567bd85603" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 29817856, + "records": [ + { + "name": "model.layers.17.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.17.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.17.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 2365440 + }, + { + "name": "model.layers.17.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 21239808 + }, + { + "name": "model.layers.17.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23605248 + }, + { + "name": "model.layers.17.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 23611392 + }, + { + "name": "model.layers.17.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 23618560 + }, + { + "name": "model.layers.17.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 29123584 + }, + { + "name": "model.layers.17.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29811712 + } + ], + "md5sum": "7766c6c3dc01655f8a50da37ea6cbafe" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.18.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "570035df6a467f1880a75ba3558825f5" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 28963840, + "records": [ + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 4718592 + }, + { + "name": "model.layers.18.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5308416 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5314560 + }, + { + "name": "model.layers.18.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 5320704 + }, + { + "name": "model.layers.18.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 5345280 + }, + { + "name": "model.layers.18.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 24219648 + }, + { + "name": "model.layers.18.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 26578944 + }, + { + "name": "model.layers.18.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 26585088 + }, + { + "name": "model.layers.18.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28944384 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28950528 + }, + { + "name": "model.layers.18.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 28956672 + } + ], + "md5sum": "845af371e9e3c4b6973b9dd3bacbd438" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.19.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "9c3a39940e8c606b0a87e1639729d5dc" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 32784384, + "records": [ + { + "name": "model.layers.18.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "model.layers.18.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "model.layers.18.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 6193152 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 6199296 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 10917888 + }, + { + "name": "model.layers.19.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11507712 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11513856 + }, + { + "name": "model.layers.19.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 11520000 + }, + { + "name": "model.layers.19.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 11544576 + }, + { + "name": "model.layers.19.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30418944 + }, + { + "name": "model.layers.19.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 32778240 + } + ], + "md5sum": "821ed15bec8ad3961a12ad0978510c41" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 32797696, + "records": [ + { + "name": "model.layers.19.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.19.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2365440 + }, + { + "name": "model.layers.19.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2371584 + }, + { + "name": "model.layers.19.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 2378752 + }, + { + "name": "model.layers.19.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 7883776 + }, + { + "name": "model.layers.19.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 8571904 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 8578048 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 13296640 + }, + { + "name": "model.layers.2.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13886464 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13892608 + }, + { + "name": "model.layers.2.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 13898752 + }, + { + "name": "model.layers.2.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 13923328 + } + ], + "md5sum": "1452e5b9bfb1a24b4a26dea601dfd2d2" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 29817856, + "records": [ + { + "name": "model.layers.2.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.2.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.2.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 2365440 + }, + { + "name": "model.layers.2.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 21239808 + }, + { + "name": "model.layers.2.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23605248 + }, + { + "name": "model.layers.2.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 23611392 + }, + { + "name": "model.layers.2.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 23618560 + }, + { + "name": "model.layers.2.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 29123584 + }, + { + "name": "model.layers.2.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29811712 + } + ], + "md5sum": "11fa45da415758689691aef8119aa983" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.20.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "95d263b9597a33e6875da5289596f568" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 28963840, + "records": [ + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 4718592 + }, + { + "name": "model.layers.20.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5308416 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5314560 + }, + { + "name": "model.layers.20.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 5320704 + }, + { + "name": "model.layers.20.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 5345280 + }, + { + "name": "model.layers.20.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 24219648 + }, + { + "name": "model.layers.20.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 26578944 + }, + { + "name": "model.layers.20.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 26585088 + }, + { + "name": "model.layers.20.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28944384 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28950528 + }, + { + "name": "model.layers.20.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 28956672 + } + ], + "md5sum": "639337f01fa88fb60a53a1588c4ee56a" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.21.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "db5e86b4d93e6b303c0a91b25fd912de" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 32784384, + "records": [ + { + "name": "model.layers.20.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "model.layers.20.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "model.layers.20.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 6193152 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 6199296 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 10917888 + }, + { + "name": "model.layers.21.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11507712 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11513856 + }, + { + "name": "model.layers.21.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 11520000 + }, + { + "name": "model.layers.21.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 11544576 + }, + { + "name": "model.layers.21.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30418944 + }, + { + "name": "model.layers.21.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 32778240 + } + ], + "md5sum": "fa30d6acb6ba6a720662f12547a7dae7" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 32797696, + "records": [ + { + "name": "model.layers.21.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.21.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2365440 + }, + { + "name": "model.layers.21.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2371584 + }, + { + "name": "model.layers.21.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 2378752 + }, + { + "name": "model.layers.21.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 7883776 + }, + { + "name": "model.layers.21.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 8571904 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 8578048 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 13296640 + }, + { + "name": "model.layers.22.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13886464 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13892608 + }, + { + "name": "model.layers.22.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 13898752 + }, + { + "name": "model.layers.22.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 13923328 + } + ], + "md5sum": "2f5aaed34040b9c7e4009d73b2ebb80f" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 29817856, + "records": [ + { + "name": "model.layers.22.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.22.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.22.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 2365440 + }, + { + "name": "model.layers.22.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 21239808 + }, + { + "name": "model.layers.22.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23605248 + }, + { + "name": "model.layers.22.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 23611392 + }, + { + "name": "model.layers.22.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 23618560 + }, + { + "name": "model.layers.22.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 29123584 + }, + { + "name": "model.layers.22.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29811712 + } + ], + "md5sum": "b0216afc78f92baa8133d1cd6faee648" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.23.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "3d70dec29a064e53c1d1dd876cbe1455" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 28963840, + "records": [ + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 4718592 + }, + { + "name": "model.layers.23.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5308416 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5314560 + }, + { + "name": "model.layers.23.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 5320704 + }, + { + "name": "model.layers.23.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 5345280 + }, + { + "name": "model.layers.23.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 24219648 + }, + { + "name": "model.layers.23.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 26578944 + }, + { + "name": "model.layers.23.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 26585088 + }, + { + "name": "model.layers.23.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28944384 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28950528 + }, + { + "name": "model.layers.23.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 28956672 + } + ], + "md5sum": "8b669940e47d82fcd1a7f5c6317f7ea2" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.24.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "b2ee821348633d1556f183ed0fb45fe3" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 32784384, + "records": [ + { + "name": "model.layers.23.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "model.layers.23.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "model.layers.23.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 6193152 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 6199296 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 10917888 + }, + { + "name": "model.layers.24.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11507712 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11513856 + }, + { + "name": "model.layers.24.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 11520000 + }, + { + "name": "model.layers.24.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 11544576 + }, + { + "name": "model.layers.24.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30418944 + }, + { + "name": "model.layers.24.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 32778240 + } + ], + "md5sum": "c6335d7a2a4be7a2bd98bd264f000a7a" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 32797696, + "records": [ + { + "name": "model.layers.24.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.24.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2365440 + }, + { + "name": "model.layers.24.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2371584 + }, + { + "name": "model.layers.24.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 2378752 + }, + { + "name": "model.layers.24.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 7883776 + }, + { + "name": "model.layers.24.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 8571904 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 8578048 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 13296640 + }, + { + "name": "model.layers.25.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13886464 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13892608 + }, + { + "name": "model.layers.25.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 13898752 + }, + { + "name": "model.layers.25.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 13923328 + } + ], + "md5sum": "ce556bbb6b89235b9ef76544786966e4" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 29817856, + "records": [ + { + "name": "model.layers.25.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.25.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.25.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 2365440 + }, + { + "name": "model.layers.25.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 21239808 + }, + { + "name": "model.layers.25.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23605248 + }, + { + "name": "model.layers.25.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 23611392 + }, + { + "name": "model.layers.25.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 23618560 + }, + { + "name": "model.layers.25.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 29123584 + }, + { + "name": "model.layers.25.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29811712 + } + ], + "md5sum": "ddae47f7dfe6928737bf6bf9cc4284c1" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.26.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "0dd27b5f509320b483ea55c7da3fda0f" + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 28963840, + "records": [ + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 4718592 + }, + { + "name": "model.layers.26.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5308416 + }, + { + "name": "model.layers.26.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5314560 + }, + { + "name": "model.layers.26.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 5320704 + }, + { + "name": "model.layers.26.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 5345280 + }, + { + "name": "model.layers.26.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 24219648 + }, + { + "name": "model.layers.26.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 26578944 + }, + { + "name": "model.layers.26.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 26585088 + }, + { + "name": "model.layers.26.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28944384 + }, + { + "name": "model.layers.26.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28950528 + }, + { + "name": "model.layers.26.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 28956672 + } + ], + "md5sum": "a67355fd995db677c01736fe1c33e375" + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.27.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "6befcee12d340cf1797d4961ab99bc13" + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 32784384, + "records": [ + { + "name": "model.layers.26.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "model.layers.26.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "model.layers.26.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 6193152 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 6199296 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 10917888 + }, + { + "name": "model.layers.27.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11507712 + }, + { + "name": "model.layers.27.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11513856 + }, + { + "name": "model.layers.27.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 11520000 + }, + { + "name": "model.layers.27.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 11544576 + }, + { + "name": "model.layers.27.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30418944 + }, + { + "name": "model.layers.27.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 32778240 + } + ], + "md5sum": "a79bbc1c84c2d75f11f589e70b9a94e8" + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 32797696, + "records": [ + { + "name": "model.layers.27.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.27.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.27.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2365440 + }, + { + "name": "model.layers.27.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2371584 + }, + { + "name": "model.layers.27.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 2378752 + }, + { + "name": "model.layers.27.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 7883776 + }, + { + "name": "model.layers.27.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 8571904 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 8578048 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 13296640 + }, + { + "name": "model.layers.28.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13886464 + }, + { + "name": "model.layers.28.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13892608 + }, + { + "name": "model.layers.28.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 13898752 + }, + { + "name": "model.layers.28.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 13923328 + } + ], + "md5sum": "7ceeeccb436f1f7e9fcf9345b4d3e879" + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 29817856, + "records": [ + { + "name": "model.layers.28.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.28.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.28.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 2365440 + }, + { + "name": "model.layers.28.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 21239808 + }, + { + "name": "model.layers.28.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.28.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23605248 + }, + { + "name": "model.layers.28.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 23611392 + }, + { + "name": "model.layers.28.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 23618560 + }, + { + "name": "model.layers.28.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 29123584 + }, + { + "name": "model.layers.28.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29811712 + } + ], + "md5sum": "b36bc962d44b0116c1ed85baffdda68a" + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.29.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "a26ea07b3794292364632dfc945fa731" + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 28963840, + "records": [ + { + "name": "model.layers.28.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 4718592 + }, + { + "name": "model.layers.29.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5308416 + }, + { + "name": "model.layers.29.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5314560 + }, + { + "name": "model.layers.29.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 5320704 + }, + { + "name": "model.layers.29.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 5345280 + }, + { + "name": "model.layers.29.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 24219648 + }, + { + "name": "model.layers.29.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 26578944 + }, + { + "name": "model.layers.29.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 26585088 + }, + { + "name": "model.layers.29.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28944384 + }, + { + "name": "model.layers.29.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28950528 + }, + { + "name": "model.layers.29.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 28956672 + } + ], + "md5sum": "e1ec677cb596aba3c7c7c910ae815bc7" + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.3.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "03b42bca0dc87c8bca36f77e445baa7b" + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 32784384, + "records": [ + { + "name": "model.layers.29.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "model.layers.29.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "model.layers.29.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 6193152 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 6199296 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 10917888 + }, + { + "name": "model.layers.3.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11507712 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11513856 + }, + { + "name": "model.layers.3.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 11520000 + }, + { + "name": "model.layers.3.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 11544576 + }, + { + "name": "model.layers.3.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30418944 + }, + { + "name": "model.layers.3.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 32778240 + } + ], + "md5sum": "31fff012ee4d260649347c438c17afa5" + }, + { + "dataPath": "params_shard_51.bin", + "format": "raw-shard", + "nbytes": 32797696, + "records": [ + { + "name": "model.layers.3.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.3.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2365440 + }, + { + "name": "model.layers.3.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2371584 + }, + { + "name": "model.layers.3.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 2378752 + }, + { + "name": "model.layers.3.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 7883776 + }, + { + "name": "model.layers.3.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 8571904 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 8578048 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 13296640 + }, + { + "name": "model.layers.4.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13886464 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13892608 + }, + { + "name": "model.layers.4.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 13898752 + }, + { + "name": "model.layers.4.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 13923328 + } + ], + "md5sum": "11757018d8329a20b53b4e439e1461e8" + }, + { + "dataPath": "params_shard_52.bin", + "format": "raw-shard", + "nbytes": 29817856, + "records": [ + { + "name": "model.layers.4.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.4.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.4.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 2365440 + }, + { + "name": "model.layers.4.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 21239808 + }, + { + "name": "model.layers.4.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23605248 + }, + { + "name": "model.layers.4.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 23611392 + }, + { + "name": "model.layers.4.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 23618560 + }, + { + "name": "model.layers.4.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 29123584 + }, + { + "name": "model.layers.4.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29811712 + } + ], + "md5sum": "21d83ac9a5722e6109261682a85a2eb2" + }, + { + "dataPath": "params_shard_53.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.5.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "ce3e0c2701927c3be3b4648ca7d5c770" + }, + { + "dataPath": "params_shard_54.bin", + "format": "raw-shard", + "nbytes": 28963840, + "records": [ + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 4718592 + }, + { + "name": "model.layers.5.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5308416 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5314560 + }, + { + "name": "model.layers.5.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 5320704 + }, + { + "name": "model.layers.5.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 5345280 + }, + { + "name": "model.layers.5.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 24219648 + }, + { + "name": "model.layers.5.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 26578944 + }, + { + "name": "model.layers.5.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 26585088 + }, + { + "name": "model.layers.5.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28944384 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28950528 + }, + { + "name": "model.layers.5.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 28956672 + } + ], + "md5sum": "a2789cd1c0d0574f804c357d06d1841a" + }, + { + "dataPath": "params_shard_55.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.6.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "17f13f096f44c21fe9301ca09d8be971" + }, + { + "dataPath": "params_shard_56.bin", + "format": "raw-shard", + "nbytes": 32784384, + "records": [ + { + "name": "model.layers.5.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "model.layers.5.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "model.layers.5.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 6193152 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 6199296 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 10917888 + }, + { + "name": "model.layers.6.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11507712 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11513856 + }, + { + "name": "model.layers.6.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 11520000 + }, + { + "name": "model.layers.6.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 11544576 + }, + { + "name": "model.layers.6.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30418944 + }, + { + "name": "model.layers.6.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 32778240 + } + ], + "md5sum": "20991121a9e1c432ff07169fa3a194b1" + }, + { + "dataPath": "params_shard_57.bin", + "format": "raw-shard", + "nbytes": 32797696, + "records": [ + { + "name": "model.layers.6.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.6.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2365440 + }, + { + "name": "model.layers.6.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2371584 + }, + { + "name": "model.layers.6.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 2378752 + }, + { + "name": "model.layers.6.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 7883776 + }, + { + "name": "model.layers.6.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 8571904 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 8578048 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 13296640 + }, + { + "name": "model.layers.7.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13886464 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13892608 + }, + { + "name": "model.layers.7.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 13898752 + }, + { + "name": "model.layers.7.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 13923328 + } + ], + "md5sum": "b9ebb62dadfdbabb22011d00a832c58a" + }, + { + "dataPath": "params_shard_58.bin", + "format": "raw-shard", + "nbytes": 29817856, + "records": [ + { + "name": "model.layers.7.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.7.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.7.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 2365440 + }, + { + "name": "model.layers.7.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 21239808 + }, + { + "name": "model.layers.7.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23605248 + }, + { + "name": "model.layers.7.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 23611392 + }, + { + "name": "model.layers.7.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 23618560 + }, + { + "name": "model.layers.7.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 29123584 + }, + { + "name": "model.layers.7.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29811712 + } + ], + "md5sum": "a88fe6583c6f5e9084260ace6b833c0d" + }, + { + "dataPath": "params_shard_59.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.8.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "2635a575384ef7ed8959e3c7330b3be3" + }, + { + "dataPath": "params_shard_60.bin", + "format": "raw-shard", + "nbytes": 28963840, + "records": [ + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 4718592 + }, + { + "name": "model.layers.8.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5308416 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 5314560 + }, + { + "name": "model.layers.8.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 5320704 + }, + { + "name": "model.layers.8.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 5345280 + }, + { + "name": "model.layers.8.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 24219648 + }, + { + "name": "model.layers.8.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 26578944 + }, + { + "name": "model.layers.8.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 26585088 + }, + { + "name": "model.layers.8.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28944384 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 28950528 + }, + { + "name": "model.layers.8.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 28956672 + } + ], + "md5sum": "b1b3793d962869844cb252ae2c5a90ef" + }, + { + "dataPath": "params_shard_61.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.9.mlp.c_proj.q_weight", + "shape": [ + 3072, + 1536 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "5462b0bcefc4ff32be889db4ae9fd441" + }, + { + "dataPath": "params_shard_62.bin", + "format": "raw-shard", + "nbytes": 32784384, + "records": [ + { + "name": "model.layers.8.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 0 + }, + { + "name": "model.layers.8.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 5505024 + }, + { + "name": "model.layers.8.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 6193152 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 6199296 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 10917888 + }, + { + "name": "model.layers.9.input_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11507712 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11513856 + }, + { + "name": "model.layers.9.mlp.c_fc.bias", + "shape": [ + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 24576, + "byteOffset": 11520000 + }, + { + "name": "model.layers.9.mlp.c_fc.q_weight", + "shape": [ + 12288, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 11544576 + }, + { + "name": "model.layers.9.mlp.c_fc.q_scale", + "shape": [ + 12288, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30418944 + }, + { + "name": "model.layers.9.mlp.c_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 32778240 + } + ], + "md5sum": "88dfd966c57f60ada94889cf5f6c580c" + }, + { + "dataPath": "params_shard_63.bin", + "format": "raw-shard", + "nbytes": 13898752, + "records": [ + { + "name": "model.layers.9.mlp.c_proj.q_scale", + "shape": [ + 3072, + 384 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.9.post_attention_layernorm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2359296 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2365440 + }, + { + "name": "model.layers.9.self_attn.wqkv_pack.bias", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2371584 + }, + { + "name": "model.layers.9.self_attn.wqkv_pack.q_weight", + "shape": [ + 3584, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5505024, + "byteOffset": 2378752 + }, + { + "name": "model.layers.9.self_attn.wqkv_pack.q_scale", + "shape": [ + 3584, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 688128, + "byteOffset": 7883776 + }, + { + "name": "model.layers.9.self_attn.o_proj.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 8571904 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 8578048 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 13296640 + }, + { + "name": "model.norm.bias", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13886464 + }, + { + "name": "model.norm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13892608 + } + ], + "md5sum": "c1b523298edc41ceb2f9b8df513f294e" + } + ] +} \ No newline at end of file