diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,5383 @@ +{ + "metadata": { + "ParamSize": 413, + "ParamBytes": 8057294848.0, + "BitsPerParam": 2.4521189954213174 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 155582464, + "records": [ + { + "name": "lm_head.q_weight", + "shape": [ + 151936, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 155582464, + "byteOffset": 0 + } + ], + "md5sum": "ec5cd947237ea99b3444b9d18930f7e8" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 25939968, + "records": [ + { + "name": "lm_head.q_scale", + "shape": [ + 151936, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 19447808, + "byteOffset": 0 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19447808 + }, + { + "name": "model.layers.23.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 19451904 + }, + { + "name": "model.layers.23.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 25219072 + } + ], + "md5sum": "dfc3f1ad8248c2058cfee2c97378d664" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 155582464, + "records": [ + { + "name": "model.embed_tokens.q_weight", + "shape": [ + 151936, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 155582464, + "byteOffset": 0 + } + ], + "md5sum": "f7a331f892c09a575c50f202ab3fa39c" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.0.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "909a07bd2d6ad37ecd7ba8b2009d6be5" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 32440320, + "records": [ + { + "name": "model.layers.23.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 0 + }, + { + "name": "model.layers.23.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 11534336 + }, + { + "name": "model.layers.23.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 12976128 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 12980224 + }, + { + "name": "model.norm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 12984320 + }, + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 151936, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 19447808, + "byteOffset": 12988416 + }, + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 32436224 + } + ], + "md5sum": "81a02951ff5b71a72f885cb701bf53b7" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.0.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "28d9c6d794a64fe1a2769d377137a167" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.0.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.0.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.0.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "3a7355e3a22dfafdf8f65ea088938edd" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.1.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "61e2005178c8fd6968a5d954789f98bf" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 28925952, + "records": [ + { + "name": "model.layers.0.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.0.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.0.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.0.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.0.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.0.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.0.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.0.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28921856 + } + ], + "md5sum": "6e231624455c552d8bd80f6587149102" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.1.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "c0ee3a8f51db35c90f3c910298aabc01" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.1.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.1.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.1.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "120bcc025cdddc7763dbe0d7915c3c95" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.2.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "6aa0ebc017ae81bc0637dd05999cacb9" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 28921856, + "records": [ + { + "name": "model.layers.1.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.1.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.1.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.1.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.1.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.1.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.1.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.1.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + } + ], + "md5sum": "5fe335d02e8ba50bd724cf0b8f91417b" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.2.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "e858366441df2fe42e45a0001f2fb795" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.2.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.2.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.2.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "1e21091eb856504cd809cbe1498a77e4" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.10.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "1a117cd2c28def885dfad2eae350cda8" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.10.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "08de0f9a09167917b7dcdcbcd5ba180d" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 21626880, + "records": [ + { + "name": "model.layers.10.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 0 + } + ], + "md5sum": "ea6b20ab0c4f2fec8163b483c1cc15db" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 33488896, + "records": [ + { + "name": "model.layers.2.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 0 + }, + { + "name": "model.layers.2.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 11534336 + }, + { + "name": "model.layers.2.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 12976128 + }, + { + "name": "model.layers.2.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 12988416 + }, + { + "name": "model.layers.2.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 19279872 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 20066304 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 22163456 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 22425600 + }, + { + "name": "model.layers.10.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 22429696 + }, + { + "name": "model.layers.10.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 33243136 + } + ], + "md5sum": "7d83197686204b382399518b457ef27e" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.11.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "aa3b5009fed50c5ad3461676e5455acf" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 28925952, + "records": [ + { + "name": "model.layers.10.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.10.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.10.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.10.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.10.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.10.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.10.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.10.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28921856 + } + ], + "md5sum": "e861b4b91aec4e6d4525521a18ac6755" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.11.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "a39c73ca577c4139eef45add1352583f" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.11.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.11.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.11.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "8621a542e994004e51d08c8ed94cb4c0" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.12.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "2c2a3b6a837748753e14c805557b533b" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 28925952, + "records": [ + { + "name": "model.layers.11.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.11.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.11.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.11.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.11.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.11.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.11.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.11.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28921856 + } + ], + "md5sum": "5e2022e6666a147e2c4ba44cceccdecc" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.12.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "b7b81fb6c05089547f6a05d94b4edc8e" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.12.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.12.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.12.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "553f3edcfdf5bb381b4fbee7614859f5" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.13.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "e5d450f6e6480313dc6ddbe140e0cfba" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 28921856, + "records": [ + { + "name": "model.layers.12.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.12.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.12.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.12.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.12.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.12.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.12.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.12.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + } + ], + "md5sum": "0e7d9271de9b2c32adb39afe6b996f51" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.13.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "bebeb1404a6af8a2bb11b2df8fdab6e5" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 32698368, + "records": [ + { + "name": "model.layers.13.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.13.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.13.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + }, + { + "name": "model.layers.13.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 32686080 + } + ], + "md5sum": "d2862dd23bbd924791c33ce2d5100abc" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 28917760, + "records": [ + { + "name": "model.layers.13.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 0 + }, + { + "name": "model.layers.13.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 6291456 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 7077888 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 9175040 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "model.layers.9.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 9441280 + }, + { + "name": "model.layers.9.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 15208448 + }, + { + "name": "model.layers.9.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 15929344 + }, + { + "name": "model.layers.9.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 27463680 + }, + { + "name": "model.layers.9.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28905472 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28909568 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28913664 + } + ], + "md5sum": "ee44c4aea62642d04d17c87eaf25c2e0" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.14.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "b0fa2508cdbaba6e11208b7912e94b2b" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.14.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "0866c269f4ab08cc4228c626823b6a9d" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 21626880, + "records": [ + { + "name": "model.layers.14.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 0 + } + ], + "md5sum": "b7162eef7455c137746a4cb12b472a72" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 30535680, + "records": [ + { + "name": "model.layers.13.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.13.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.13.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.13.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.13.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19472384 + }, + { + "name": "model.layers.14.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 19476480 + }, + { + "name": "model.layers.14.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 30289920 + } + ], + "md5sum": "7b28de118bd26216a235b77e293f35f8" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.15.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "a41a854dbf77ef5a41c3dfeccc4683d4" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 28925952, + "records": [ + { + "name": "model.layers.14.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.14.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.14.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.14.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.14.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.14.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.14.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.14.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28921856 + } + ], + "md5sum": "eb766d118475b92dc220f63269fef581" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.15.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "16c52bd25b7a60ba4e4aeb64b9985fda" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.15.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.15.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.15.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "08c7d6f60fc609a6d91ab06b5546fc27" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.16.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "531d4623d664ddfce1e955938f33f745" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 28921856, + "records": [ + { + "name": "model.layers.15.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.15.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.15.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.15.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.15.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.15.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.15.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.15.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + } + ], + "md5sum": "f0402c7fb1d34273ba31ae392feb2dd0" + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.16.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "f0ce7b2e1acf5aa073f116f8e0fa0742" + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.16.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.16.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.16.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "16e5089f89f9f3320299c19a6a047380" + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.17.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "b322ca99f040a58246b248bf0c39b31b" + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 28930048, + "records": [ + { + "name": "model.layers.16.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 0 + }, + { + "name": "model.layers.16.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 11534336 + }, + { + "name": "model.layers.16.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 12976128 + }, + { + "name": "model.layers.16.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 12988416 + }, + { + "name": "model.layers.16.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 19279872 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 20066304 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 22163456 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 22425600 + }, + { + "name": "model.layers.16.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 22429696 + }, + { + "name": "model.layers.16.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 28196864 + }, + { + "name": "model.layers.16.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28917760 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28921856 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28925952 + } + ], + "md5sum": "6a94504f4190c052425978c97bb9ea9b" + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.17.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "3cb04bb4684772e10bbc8228ba1e79f9" + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.17.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.17.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.17.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "3387d69983a203c497afaf4a615f9e36" + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.18.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "35e338e06aed9296958f01364d58f11d" + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 28925952, + "records": [ + { + "name": "model.layers.17.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.17.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.17.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.17.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.17.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.17.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.17.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.17.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28921856 + } + ], + "md5sum": "130daa9eddfdb14a80ea4671105458b3" + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.18.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "ca27f88f38757ef37b80fcce6206d64d" + }, + { + "dataPath": "params_shard_51.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.18.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.18.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.18.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "b20538e91682bdd8cc0ad281420441cc" + }, + { + "dataPath": "params_shard_52.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.19.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "7524b89cdf1e66c40b51f7338bfcbe73" + }, + { + "dataPath": "params_shard_53.bin", + "format": "raw-shard", + "nbytes": 28925952, + "records": [ + { + "name": "model.layers.18.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.18.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.18.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.18.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.18.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.18.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.18.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.18.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28921856 + } + ], + "md5sum": "5e9d698369358f2a6420dc704ecbff7b" + }, + { + "dataPath": "params_shard_54.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.19.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "0460278124d203bdfe707564d5a1b943" + }, + { + "dataPath": "params_shard_55.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.19.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.19.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.19.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "7ef11d193adac1bf24549bea7340ebac" + }, + { + "dataPath": "params_shard_56.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.20.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "1e807c15ac309bc2765cbd3b2654d943" + }, + { + "dataPath": "params_shard_57.bin", + "format": "raw-shard", + "nbytes": 28921856, + "records": [ + { + "name": "model.layers.19.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.19.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.19.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.19.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.19.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.19.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.19.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.19.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + } + ], + "md5sum": "1029bea6a493fbad5947e5b376b5faaa" + }, + { + "dataPath": "params_shard_58.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.20.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "4129bf43a701760f9a0a9da3cacf11c1" + }, + { + "dataPath": "params_shard_59.bin", + "format": "raw-shard", + "nbytes": 32698368, + "records": [ + { + "name": "model.layers.20.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.20.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.20.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + }, + { + "name": "model.layers.20.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 32686080 + } + ], + "md5sum": "b20a2b6349e4f490e3705ea709d10973" + }, + { + "dataPath": "params_shard_60.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.3.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "b0240df6ffcfb6acd326311cd3da7b7d" + }, + { + "dataPath": "params_shard_61.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.3.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "d402dbd16d4e6cd8055beaaf2396cc79" + }, + { + "dataPath": "params_shard_62.bin", + "format": "raw-shard", + "nbytes": 21626880, + "records": [ + { + "name": "model.layers.3.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 0 + } + ], + "md5sum": "83115e8bf8c91888547c6ec4de2d446a" + }, + { + "dataPath": "params_shard_63.bin", + "format": "raw-shard", + "nbytes": 33488896, + "records": [ + { + "name": "model.layers.20.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 0 + }, + { + "name": "model.layers.20.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 6291456 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 7077888 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 9175040 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "model.layers.2.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 9441280 + }, + { + "name": "model.layers.2.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 15208448 + }, + { + "name": "model.layers.2.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 15929344 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 15933440 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 15937536 + }, + { + "name": "model.layers.3.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 15941632 + }, + { + "name": "model.layers.3.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 26755072 + }, + { + "name": "model.layers.3.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 27000832 + }, + { + "name": "model.layers.3.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 32768000 + } + ], + "md5sum": "ffaefe394bdc9a03faed21747d33d441" + }, + { + "dataPath": "params_shard_64.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.4.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "c81931b749d68cdd6a3a3a028b8dfa7b" + }, + { + "dataPath": "params_shard_65.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.4.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "66cabe9ff3224218625fd6b87571eb23" + }, + { + "dataPath": "params_shard_66.bin", + "format": "raw-shard", + "nbytes": 21626880, + "records": [ + { + "name": "model.layers.4.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 0 + } + ], + "md5sum": "4d4794fa1e0409fb3c6d75554bd48e69" + }, + { + "dataPath": "params_shard_67.bin", + "format": "raw-shard", + "nbytes": 33497088, + "records": [ + { + "name": "model.layers.3.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 0 + }, + { + "name": "model.layers.3.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 11534336 + }, + { + "name": "model.layers.3.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 12976128 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 12980224 + }, + { + "name": "model.layers.3.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 12984320 + }, + { + "name": "model.layers.3.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 12996608 + }, + { + "name": "model.layers.3.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 19288064 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 20074496 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 22171648 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 22433792 + }, + { + "name": "model.layers.4.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 22437888 + }, + { + "name": "model.layers.4.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 33251328 + } + ], + "md5sum": "ba6530d930366fc3087a7c9d7c62fbf3" + }, + { + "dataPath": "params_shard_68.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.5.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "4eb5ba864c9a4ef3f5242462ea8169f9" + }, + { + "dataPath": "params_shard_69.bin", + "format": "raw-shard", + "nbytes": 28925952, + "records": [ + { + "name": "model.layers.4.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.4.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.4.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.4.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.4.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.4.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.4.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.4.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28921856 + } + ], + "md5sum": "6bf40ee6b1f0c19c7fee2c86994e4e13" + }, + { + "dataPath": "params_shard_70.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.5.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "cd41b05b7087c595dbf5f196f7967bba" + }, + { + "dataPath": "params_shard_71.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.5.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.5.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.5.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "7ad927fca86ec05be1f12c869833203c" + }, + { + "dataPath": "params_shard_72.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.6.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "2323632854c4f167eabc0fb935fb3fab" + }, + { + "dataPath": "params_shard_73.bin", + "format": "raw-shard", + "nbytes": 28921856, + "records": [ + { + "name": "model.layers.5.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.5.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.5.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.5.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.5.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.5.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.5.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.5.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + } + ], + "md5sum": "32ec694d867b0a257dc247d38fc32fe1" + }, + { + "dataPath": "params_shard_74.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.6.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "1267183ad7259ff42512f6b94dc2fa17" + }, + { + "dataPath": "params_shard_75.bin", + "format": "raw-shard", + "nbytes": 32698368, + "records": [ + { + "name": "model.layers.6.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.6.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.6.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + }, + { + "name": "model.layers.6.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 32686080 + } + ], + "md5sum": "81a49e35d2959b3922f8f725cc12e2c9" + }, + { + "dataPath": "params_shard_76.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.21.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "5815ff395ff5a53dd599e2e4996aff9a" + }, + { + "dataPath": "params_shard_77.bin", + "format": "raw-shard", + "nbytes": 28917760, + "records": [ + { + "name": "model.layers.6.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 0 + }, + { + "name": "model.layers.6.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 6291456 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 7077888 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 9175040 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "model.layers.20.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 9441280 + }, + { + "name": "model.layers.20.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 15208448 + }, + { + "name": "model.layers.20.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 15929344 + }, + { + "name": "model.layers.20.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 27463680 + }, + { + "name": "model.layers.20.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28905472 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28909568 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28913664 + } + ], + "md5sum": "0de033e30d333b486bd83963f2400ec0" + }, + { + "dataPath": "params_shard_78.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.21.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "9656e5294ef09f11a825d25b0823a019" + }, + { + "dataPath": "params_shard_79.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.21.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.21.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.21.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "b6df0364e8645da6f857ebe94d591f02" + }, + { + "dataPath": "params_shard_80.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.22.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "3c5f0908cc79a18c6a74c4c0d03ffe29" + }, + { + "dataPath": "params_shard_81.bin", + "format": "raw-shard", + "nbytes": 28925952, + "records": [ + { + "name": "model.layers.21.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.21.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.21.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.21.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.21.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.21.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.21.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.21.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28921856 + } + ], + "md5sum": "abcf0d5ceb6d47b65898598329b0c298" + }, + { + "dataPath": "params_shard_82.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.22.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "c6b801ce635fa79f827b477395af0ece" + }, + { + "dataPath": "params_shard_83.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.22.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.22.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.22.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "2fdfd327a00129aa5d4fb3568f857961" + }, + { + "dataPath": "params_shard_84.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.23.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "c84f9ca2c3c181b69adcbc487ffd6e5c" + }, + { + "dataPath": "params_shard_85.bin", + "format": "raw-shard", + "nbytes": 28921856, + "records": [ + { + "name": "model.layers.22.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.22.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.22.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.22.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.22.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.22.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.22.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.22.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + } + ], + "md5sum": "eab932b3772b6a19354263ada14f1cad" + }, + { + "dataPath": "params_shard_86.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.23.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "521a56a89027061a85e89e9caa5afc85" + }, + { + "dataPath": "params_shard_87.bin", + "format": "raw-shard", + "nbytes": 32698368, + "records": [ + { + "name": "model.layers.23.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.23.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.23.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + }, + { + "name": "model.layers.23.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 32686080 + } + ], + "md5sum": "9823509eaa35dc361761ce2bf1eac264" + }, + { + "dataPath": "params_shard_88.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.7.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "231eb578d23fa7e0b2d54457720c6acd" + }, + { + "dataPath": "params_shard_89.bin", + "format": "raw-shard", + "nbytes": 28917760, + "records": [ + { + "name": "model.layers.23.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 0 + }, + { + "name": "model.layers.23.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 6291456 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 7077888 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 9175040 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "model.layers.6.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 9441280 + }, + { + "name": "model.layers.6.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 15208448 + }, + { + "name": "model.layers.6.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 15929344 + }, + { + "name": "model.layers.6.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 27463680 + }, + { + "name": "model.layers.6.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28905472 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28909568 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28913664 + } + ], + "md5sum": "dd7292de182c5d7e080ac4889134d568" + }, + { + "dataPath": "params_shard_90.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.7.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "f9ca51abd3477193006a5fec2981accb" + }, + { + "dataPath": "params_shard_91.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.7.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.7.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.7.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "ac8abf39dc26aa1ed5b772de74ba6ae6" + }, + { + "dataPath": "params_shard_92.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.8.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "490a209090a51280a6ac11d8b8d8f616" + }, + { + "dataPath": "params_shard_93.bin", + "format": "raw-shard", + "nbytes": 28925952, + "records": [ + { + "name": "model.layers.7.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.7.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.7.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.7.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.7.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.7.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.7.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.7.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28921856 + } + ], + "md5sum": "bd18bd55567440696c1b3f567b611e1d" + }, + { + "dataPath": "params_shard_94.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.8.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "e4e74dfe3b589e1cfade2710108b873e" + }, + { + "dataPath": "params_shard_95.bin", + "format": "raw-shard", + "nbytes": 32686080, + "records": [ + { + "name": "model.layers.8.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.8.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.8.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + } + ], + "md5sum": "fedf9ec13539b0814251c337bf3742c0" + }, + { + "dataPath": "params_shard_96.bin", + "format": "raw-shard", + "nbytes": 86507520, + "records": [ + { + "name": "model.layers.9.mlp.moe_down_proj.q_weight", + "shape": [ + 60, + 2048, + 176 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 86507520, + "byteOffset": 0 + } + ], + "md5sum": "cd6d850f617662e66767fdfdb9f67d55" + }, + { + "dataPath": "params_shard_97.bin", + "format": "raw-shard", + "nbytes": 28921856, + "records": [ + { + "name": "model.layers.8.mlp.shared_expert.down_proj.q_weight", + "shape": [ + 2048, + 704 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 5767168, + "byteOffset": 0 + }, + { + "name": "model.layers.8.mlp.shared_expert.down_proj.q_scale", + "shape": [ + 2048, + 176 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 720896, + "byteOffset": 5767168 + }, + { + "name": "model.layers.8.mlp.shared_expert.gate_up_proj.q_weight", + "shape": [ + 11264, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11534336, + "byteOffset": 6488064 + }, + { + "name": "model.layers.8.mlp.shared_expert.gate_up_proj.q_scale", + "shape": [ + 11264, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1441792, + "byteOffset": 18022400 + }, + { + "name": "model.layers.8.mlp.shared_expert_gate.weight", + "shape": [ + 1, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19464192 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19468288 + }, + { + "name": "model.layers.8.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 19472384 + }, + { + "name": "model.layers.8.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19484672 + }, + { + "name": "model.layers.8.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25776128 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26562560 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 28659712 + } + ], + "md5sum": "d20978dc6b6dc02f49e38408889783bc" + }, + { + "dataPath": "params_shard_98.bin", + "format": "raw-shard", + "nbytes": 173015040, + "records": [ + { + "name": "model.layers.9.mlp.moe_gate_up_proj.q_weight", + "shape": [ + 60, + 2816, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 173015040, + "byteOffset": 0 + } + ], + "md5sum": "53ee925fadc7cc31a64323f9b55e22cb" + }, + { + "dataPath": "params_shard_99.bin", + "format": "raw-shard", + "nbytes": 32698368, + "records": [ + { + "name": "model.layers.9.mlp.moe_down_proj.q_scale", + "shape": [ + 60, + 2048, + 44 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10813440, + "byteOffset": 0 + }, + { + "name": "model.layers.9.mlp.moe_gate_up_proj.q_scale", + "shape": [ + 60, + 2816, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 21626880, + "byteOffset": 10813440 + }, + { + "name": "model.layers.9.mlp.gate.weight", + "shape": [ + 60, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 245760, + "byteOffset": 32440320 + }, + { + "name": "model.layers.9.self_attn.c_attn.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 12288, + "byteOffset": 32686080 + } + ], + "md5sum": "2228a11979f7d8a9bf2f68220fc5f0e3" + }, + { + "dataPath": "params_shard_100.bin", + "format": "raw-shard", + "nbytes": 9437184, + "records": [ + { + "name": "model.layers.9.self_attn.c_attn.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 0 + }, + { + "name": "model.layers.9.self_attn.c_attn.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 6291456 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 7077888 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 9175040 + } + ], + "md5sum": "c29355dd96de7462d45ebae8e938bb30" + } + ] +} \ No newline at end of file