diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,3865 @@ +{ + "metadata": { + "ParamSize": 339, + "ParamBytes": 968001536.0, + "BitsPerParam": 4.500828151065705 + }, + "records": [ + { + "dataPath": "params_shard_0.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": "529aed61a274c19024f5c9431c844191" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 26529792, + "records": [ + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 151936, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 19447808, + "byteOffset": 0 + }, + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19447808 + }, + { + "name": "model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 19451904 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 25743360 + } + ], + "md5sum": "c79a793977aca8d9c4a5963f19f913d6" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.0.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.0.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.0.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.0.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "1f742601c6f09d63ea3207fcda916a0d" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.1.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.1.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.1.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.1.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "6f4308450ca8b88071cd5934914ab71c" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.10.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.10.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.10.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.10.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "eadb67efb3d89fc2c52f1f83338ff46d" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.11.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.11.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.11.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.11.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "000a246dbaf83ee7272d50da109bd079" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.12.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.12.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.12.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.12.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "95e6d547e1673a76ac9da7ea00627df6" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.13.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.13.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.13.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.13.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "53e58cff84f1349826f59bf73770a9ed" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.14.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.14.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.14.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.14.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "78aba5a651817a6229418e831ace1a90" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.15.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.15.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.15.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.15.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "094a18a575689a4ae975617c250c83ce" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.16.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.16.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.16.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.16.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "2f5761d9229885cd60df7327e0038c80" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.17.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.17.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.17.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.17.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "c849fe40af8cf340fd9bba5acb64da7b" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.18.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.18.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.18.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.18.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "e8a4aafdfaea2c18c9f32b75fd9ee8fb" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.19.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.19.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.19.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.19.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "a9045909b758fc7c1c29461a5446123c" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.2.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.2.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.2.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.2.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "f2135cc93813b69cfdbeb620a52e99a7" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.20.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.20.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.20.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.20.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "9a6674373fa4eba36442cc2407b948e9" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.21.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.21.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.21.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.21.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "d303035e9088e5b0470f70f01255d8a0" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.22.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.22.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.22.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.22.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "03bcba2a4d5a09c86ecad326500bf964" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.23.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.23.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.23.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.23.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "1cfc51aebc592dc555aed97ded2beba2" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.24.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.24.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.24.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.24.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "ac51b7329f10e6437d49788010f54c5f" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.25.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.25.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.25.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.25.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.26.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.26.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.26.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "249e208f9b439ba1bdde1f83936504c4" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.26.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.26.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.26.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.26.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.26.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.26.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.26.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.27.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.27.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.27.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "26e834dfbace25bb0eda835bc65e0550" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.27.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.27.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.27.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.27.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.27.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.27.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.27.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "991a6842d7143e73ffd7fb2eabecf5cf" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.3.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.3.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.3.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.3.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "37d3e6d5b699337a9152316dc58aa11c" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.4.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.4.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.4.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.4.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "bac77e98427b0f87217661220714b21d" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.5.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.5.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.5.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.5.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "9be73d4adb85227c137dd47b3073a266" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.6.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.6.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.6.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.6.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "948fd4fe4c3dd843a23e07060ce1ac88" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.7.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.7.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.7.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.7.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "4fad97f69eae311a1ddcfe719b42ccb0" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 28320256, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.8.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.8.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.8.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.8.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + }, + { + "name": "model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 768, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 21242368 + }, + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 192, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 27533824 + } + ], + "md5sum": "4ce88284bd84adca767f4f045a67126f" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 21242368, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 64, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14155776 + }, + { + "name": "model.layers.9.self_attn.k_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 14159872 + }, + { + "name": "model.layers.9.self_attn.c_attn.q_weight", + "shape": [ + 256, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4194304, + "byteOffset": 14160128 + }, + { + "name": "model.layers.9.self_attn.c_attn.q_scale", + "shape": [ + 64, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 524288, + "byteOffset": 18354432 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 18878720 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 20975872 + }, + { + "name": "model.layers.9.self_attn.q_norm.weight", + "shape": [ + 128 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 256, + "byteOffset": 21238016 + }, + { + "name": "model.norm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 21238272 + } + ], + "md5sum": "093cc91abd84c360fc4cfb85275a4a45" + } + ] +} \ No newline at end of file