{ "metadata": { "ParamSize": 284, "ParamBytes": 1504787200.0, "BitsPerParam": 13.78264604288513 }, "records": [ { "dataPath": "params_shard_0.bin", "format": "raw-shard", "nbytes": 508559360, "records": [ { "name": "model.embed_tokens.weight", "shape": [ 248320, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 508559360, "byteOffset": 0 } ], "md5sum": "246222e939a26ca09cceab523cf06eed" }, { "dataPath": "params_shard_1.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.14.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 0 }, { "name": "model.layers.12.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 12582912 } ], "md5sum": "d65600f3ef3d5f0632c5adfbe3e3e646" }, { "dataPath": "params_shard_2.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.18.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 0 }, { "name": "model.layers.13.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 12582912 } ], "md5sum": "694f31e3be1cf18d09fec4207f888861" }, { "dataPath": "params_shard_3.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.2.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 0 }, { "name": "model.layers.4.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 12582912 } ], "md5sum": "78e849c95f1a3c302d4553536e506830" }, { "dataPath": "params_shard_4.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.20.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 0 }, { "name": "model.layers.1.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 12582912 } ], "md5sum": "874992a3527c015e0ab2b3ad85cbcba9" }, { "dataPath": "params_shard_5.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.10.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 0 }, { "name": "model.layers.16.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 12582912 } ], "md5sum": "53720a12c082b964e8c0d4d23e9e93d3" }, { "dataPath": "params_shard_6.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.17.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 0 }, { "name": "model.layers.5.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 12582912 } ], "md5sum": "047ab4bf061154e09b2c2109c9afd7ea" }, { "dataPath": "params_shard_7.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.6.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 0 }, { "name": "model.layers.8.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 12582912 } ], "md5sum": "eaa3365ebbd33f993f1ad2bd2e5c81f9" }, { "dataPath": "params_shard_8.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.9.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 0 }, { "name": "model.layers.21.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 12582912 } ], "md5sum": "44a0a848d9a5f8810dac8e7571f1ee64" }, { "dataPath": "params_shard_9.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.22.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 0 }, { "name": "model.layers.0.linear_attn.in_proj_qkv.weight", "shape": [ 6144, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 12582912, "byteOffset": 12582912 } ], "md5sum": "6ac28602730004220f5831b681f8d9a4" }, { "dataPath": "params_shard_10.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.23.self_attn.c_attn.weight", "shape": [ 5120, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 0 }, { "name": "model.layers.15.self_attn.c_attn.weight", "shape": [ 5120, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 10485760 }, { "name": "model.layers.7.self_attn.c_attn.weight", "shape": [ 5120, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 20971520 } ], "md5sum": "0b878eb3eeaac694c5b56c3d6e3e6fb8" }, { "dataPath": "params_shard_11.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.11.self_attn.c_attn.weight", "shape": [ 5120, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 0 }, { "name": "model.layers.19.self_attn.c_attn.weight", "shape": [ 5120, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 10485760 }, { "name": "model.layers.3.self_attn.c_attn.weight", "shape": [ 5120, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 10485760, "byteOffset": 20971520 } ], "md5sum": "3907392971b5faaef72a37d809b14c8f" }, { "dataPath": "params_shard_12.bin", "format": "raw-shard", "nbytes": 29360128, "records": [ { "name": "model.layers.22.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 0 }, { "name": "model.layers.22.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 7340032 }, { "name": "model.layers.23.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 22020096 } ], "md5sum": "f9c23faa977b3080971568308e045ba5" }, { "dataPath": "params_shard_13.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.23.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.3.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "8ed0b3f1ac350c4e3cb40ce2ed791f31" }, { "dataPath": "params_shard_14.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.3.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.14.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "d22f51093394b92f020c24f8eefb44ea" }, { "dataPath": "params_shard_15.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.14.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.15.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "90158fb203c885aa380264827d1d468b" }, { "dataPath": "params_shard_16.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.15.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.6.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "8cb387f90d5c2760836e3de976aaa456" }, { "dataPath": "params_shard_17.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.6.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.7.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "7f1a28c2777c77bf4b819e11d9c20272" }, { "dataPath": "params_shard_18.bin", "format": "raw-shard", "nbytes": 29360128, "records": [ { "name": "model.layers.7.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.10.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 }, { "name": "model.layers.11.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 22020096 } ], "md5sum": "3c2c851f5cb6634d58efec82921a565d" }, { "dataPath": "params_shard_19.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.11.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.9.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "4b8dcd911b901bd152853aa1535d5ccf" }, { "dataPath": "params_shard_20.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.9.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.17.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "7c37e75b9bf6e799f445e546e5f0bbda" }, { "dataPath": "params_shard_21.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.17.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.18.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "c69d8520a5b5f613e8f46cd75710bd19" }, { "dataPath": "params_shard_22.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.18.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.12.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "743ed45e987a55f6d5dcc3be58d7ef71" }, { "dataPath": "params_shard_23.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.12.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.13.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "e0a0ce0c79310e5a23473a365669fd68" }, { "dataPath": "params_shard_24.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.13.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.19.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "beb0fbe9f91b1651bb7426104187ddc8" }, { "dataPath": "params_shard_25.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.19.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.4.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "e5f9883b952c2b5bcb4501d44ef19cc1" }, { "dataPath": "params_shard_26.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.4.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.2.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "e2f88ee0053bc4303c0c6a55699d0c5f" }, { "dataPath": "params_shard_27.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.2.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.20.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "b4e43f5c3480739b97c32e8861e9bb3d" }, { "dataPath": "params_shard_28.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.20.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.1.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "15d0713eb60d2e50605e9b24c9045c04" }, { "dataPath": "params_shard_29.bin", "format": "raw-shard", "nbytes": 29360128, "records": [ { "name": "model.layers.1.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.10.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 14680064 } ], "md5sum": "2795f8b57c3c5a3ec47a187cc936d770" }, { "dataPath": "params_shard_30.bin", "format": "raw-shard", "nbytes": 29360128, "records": [ { "name": "model.layers.16.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 0 }, { "name": "model.layers.16.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 7340032 }, { "name": "model.layers.5.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 22020096 } ], "md5sum": "088777494cbce51754d1a26601d387e8" }, { "dataPath": "params_shard_31.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.5.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.8.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "0b9daa2a429221bdefeea56d83ddc034" }, { "dataPath": "params_shard_32.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.8.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.21.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "e2abcd31371838c0524b72af28c37143" }, { "dataPath": "params_shard_33.bin", "format": "raw-shard", "nbytes": 22020096, "records": [ { "name": "model.layers.21.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.0.mlp.down_proj.weight", "shape": [ 1024, 3584 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 7340032, "byteOffset": 14680064 } ], "md5sum": "330c5d76a9d04c57c1c8bfd6a2209585" }, { "dataPath": "params_shard_34.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.0.mlp.gate_up_proj.weight", "shape": [ 7168, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 14680064, "byteOffset": 0 }, { "name": "model.layers.23.self_attn.o_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 14680064 }, { "name": "model.layers.3.self_attn.o_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 18874368 }, { "name": "model.layers.14.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 23068672 }, { "name": "model.layers.14.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 27262976 } ], "md5sum": "f5b484ad7c6dce97e10ce5d2bf93323f" }, { "dataPath": "params_shard_35.bin", "format": "raw-shard", "nbytes": 29360128, "records": [ { "name": "model.layers.15.self_attn.o_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 0 }, { "name": "model.layers.6.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 4194304 }, { "name": "model.layers.7.self_attn.o_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 8388608 }, { "name": "model.layers.11.self_attn.o_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 12582912 }, { "name": "model.layers.12.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 16777216 }, { "name": "model.layers.12.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 20971520 }, { "name": "model.layers.17.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 25165824 } ], "md5sum": "76d7394b1eae3d423c4046e1b0c434ec" }, { "dataPath": "params_shard_36.bin", "format": "raw-shard", "nbytes": 29360128, "records": [ { "name": "model.layers.18.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 0 }, { "name": "model.layers.18.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 4194304 }, { "name": "model.layers.13.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 8388608 }, { "name": "model.layers.13.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 12582912 }, { "name": "model.layers.19.self_attn.o_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 16777216 }, { "name": "model.layers.2.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 20971520 }, { "name": "model.layers.2.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 25165824 } ], "md5sum": "a6b4b22bac16ddd08fbccb130ec079ba" }, { "dataPath": "params_shard_37.bin", "format": "raw-shard", "nbytes": 29360128, "records": [ { "name": "model.layers.4.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 0 }, { "name": "model.layers.4.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 4194304 }, { "name": "model.layers.20.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 8388608 }, { "name": "model.layers.20.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 12582912 }, { "name": "model.layers.1.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 16777216 }, { "name": "model.layers.1.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 20971520 }, { "name": "model.layers.10.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 25165824 } ], "md5sum": "4f2f10f9542a8caa7598862e18db9227" }, { "dataPath": "params_shard_38.bin", "format": "raw-shard", "nbytes": 29360128, "records": [ { "name": "model.layers.10.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 0 }, { "name": "model.layers.16.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 4194304 }, { "name": "model.layers.16.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 8388608 }, { "name": "model.layers.17.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 12582912 }, { "name": "model.layers.5.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 16777216 }, { "name": "model.layers.5.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 20971520 }, { "name": "model.layers.6.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 25165824 } ], "md5sum": "6c732c7982e09afebc47d39112f1b8a5" }, { "dataPath": "params_shard_39.bin", "format": "raw-shard", "nbytes": 29360128, "records": [ { "name": "model.layers.8.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 0 }, { "name": "model.layers.8.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 4194304 }, { "name": "model.layers.9.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 8388608 }, { "name": "model.layers.9.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 12582912 }, { "name": "model.layers.21.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 16777216 }, { "name": "model.layers.21.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 20971520 }, { "name": "model.layers.22.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 25165824 } ], "md5sum": "630ddcbb8f19df30dbe922c670c112db" }, { "dataPath": "params_shard_40.bin", "format": "raw-shard", "nbytes": 14759552, "records": [ { "name": "model.layers.22.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 0 }, { "name": "model.layers.0.linear_attn.in_proj_z.weight", "shape": [ 2048, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 4194304 }, { "name": "model.layers.0.linear_attn.out_proj.weight", "shape": [ 1024, 2048 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4194304, "byteOffset": 8388608 }, { "name": "model.layers.8.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 12582912 }, { "name": "model.layers.12.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 12632064 }, { "name": "model.layers.18.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 12681216 }, { "name": "model.layers.13.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 12730368 }, { "name": "model.layers.14.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 12779520 }, { "name": "model.layers.2.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 12828672 }, { "name": "model.layers.4.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 12877824 }, { "name": "model.layers.5.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 12926976 }, { "name": "model.layers.20.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 12976128 }, { "name": "model.layers.21.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 13025280 }, { "name": "model.layers.10.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 13074432 }, { "name": "model.layers.16.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 13123584 }, { "name": "model.layers.17.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 13172736 }, { "name": "model.layers.6.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 13221888 }, { "name": "model.layers.9.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 13271040 }, { "name": "model.layers.22.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 13320192 }, { "name": "model.layers.0.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 13369344 }, { "name": "model.layers.1.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 13418496 }, { "name": "model.layers.14.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13467648 }, { "name": "model.layers.14.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13500416 }, { "name": "model.layers.12.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13533184 }, { "name": "model.layers.12.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13565952 }, { "name": "model.layers.18.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13598720 }, { "name": "model.layers.18.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13631488 }, { "name": "model.layers.13.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13664256 }, { "name": "model.layers.13.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13697024 }, { "name": "model.layers.2.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13729792 }, { "name": "model.layers.2.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13762560 }, { "name": "model.layers.4.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13795328 }, { "name": "model.layers.4.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13828096 }, { "name": "model.layers.20.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13860864 }, { "name": "model.layers.20.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13893632 }, { "name": "model.layers.1.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13926400 }, { "name": "model.layers.1.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13959168 }, { "name": "model.layers.10.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 13991936 }, { "name": "model.layers.10.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14024704 }, { "name": "model.layers.16.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14057472 }, { "name": "model.layers.16.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14090240 }, { "name": "model.layers.17.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14123008 }, { "name": "model.layers.17.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14155776 }, { "name": "model.layers.5.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14188544 }, { "name": "model.layers.5.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14221312 }, { "name": "model.layers.6.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14254080 }, { "name": "model.layers.6.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14286848 }, { "name": "model.layers.8.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14319616 }, { "name": "model.layers.8.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14352384 }, { "name": "model.layers.9.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14385152 }, { "name": "model.layers.9.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14417920 }, { "name": "model.layers.21.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14450688 }, { "name": "model.layers.21.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14483456 }, { "name": "model.layers.22.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14516224 }, { "name": "model.layers.22.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14548992 }, { "name": "model.layers.0.linear_attn.in_proj_b.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14581760 }, { "name": "model.layers.0.linear_attn.in_proj_a.weight", "shape": [ 16, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 32768, "byteOffset": 14614528 }, { "name": "model.layers.22.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14647296 }, { "name": "model.layers.23.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14649344 }, { "name": "model.layers.23.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14651392 }, { "name": "model.layers.3.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14653440 }, { "name": "model.layers.14.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14655488 }, { "name": "model.layers.14.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14657536 }, { "name": "model.layers.15.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14659584 }, { "name": "model.layers.15.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14661632 }, { "name": "model.layers.6.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14663680 }, { "name": "model.layers.7.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14665728 }, { "name": "model.layers.7.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14667776 }, { "name": "model.layers.8.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14669824 }, { "name": "model.layers.11.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14671872 }, { "name": "model.layers.11.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14673920 }, { "name": "model.layers.12.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14675968 }, { "name": "model.layers.17.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14678016 }, { "name": "model.layers.18.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14680064 }, { "name": "model.layers.18.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14682112 }, { "name": "model.layers.19.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14684160 }, { "name": "model.layers.12.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "model.layers.13.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14688256 }, { "name": "model.layers.13.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14690304 }, { "name": "model.layers.19.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14692352 }, { "name": "model.layers.2.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14694400 }, { "name": "model.layers.2.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14696448 }, { "name": "model.layers.3.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14698496 }, { "name": "model.layers.4.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14700544 }, { "name": "model.layers.4.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14702592 }, { "name": "model.layers.20.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14704640 }, { "name": "model.layers.20.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14706688 }, { "name": "model.layers.1.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14708736 }, { "name": "model.layers.10.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14710784 }, { "name": "model.layers.10.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14712832 }, { "name": "model.layers.16.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14714880 }, { "name": "model.layers.16.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14716928 }, { "name": "model.layers.17.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14718976 }, { "name": "model.layers.5.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14721024 }, { "name": "model.layers.5.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14723072 }, { "name": "model.layers.6.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14725120 }, { "name": "model.layers.8.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14727168 }, { "name": "model.layers.9.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14729216 }, { "name": "model.layers.9.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14731264 }, { "name": "model.layers.21.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14733312 }, { "name": "model.layers.21.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14735360 }, { "name": "model.layers.22.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14737408 }, { "name": "model.layers.0.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14739456 }, { "name": "model.layers.0.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14741504 }, { "name": "model.layers.1.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14743552 }, { "name": "model.norm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14745600 }, { "name": "model.layers.23.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 14747648 }, { "name": "model.layers.23.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 14748160 }, { "name": "model.layers.3.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 14748672 }, { "name": "model.layers.14.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14749184 }, { "name": "model.layers.15.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 14749440 }, { "name": "model.layers.15.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 14749952 }, { "name": "model.layers.7.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 14750464 }, { "name": "model.layers.7.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 14750976 }, { "name": "model.layers.11.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 14751488 }, { "name": "model.layers.11.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 14752000 }, { "name": "model.layers.12.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14752512 }, { "name": "model.layers.18.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14752768 }, { "name": "model.layers.13.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14753024 }, { "name": "model.layers.19.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 14753280 }, { "name": "model.layers.19.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 14753792 }, { "name": "model.layers.2.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14754304 }, { "name": "model.layers.3.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 14754560 }, { "name": "model.layers.4.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14755072 }, { "name": "model.layers.20.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14755328 }, { "name": "model.layers.1.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14755584 }, { "name": "model.layers.10.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14755840 }, { "name": "model.layers.16.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14756096 }, { "name": "model.layers.17.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14756352 }, { "name": "model.layers.5.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14756608 }, { "name": "model.layers.6.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14756864 }, { "name": "model.layers.8.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14757120 }, { "name": "model.layers.9.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14757376 }, { "name": "model.layers.21.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14757632 }, { "name": "model.layers.22.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14757888 }, { "name": "model.layers.0.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 14758144 }, { "name": "model.layers.16.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758400 }, { "name": "model.layers.8.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758432 }, { "name": "model.layers.12.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758464 }, { "name": "model.layers.18.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758496 }, { "name": "model.layers.13.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758528 }, { "name": "model.layers.14.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758560 }, { "name": "model.layers.2.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758592 }, { "name": "model.layers.4.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758624 }, { "name": "model.layers.5.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758656 }, { "name": "model.layers.20.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758688 }, { "name": "model.layers.21.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758720 }, { "name": "model.layers.10.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758752 }, { "name": "model.layers.17.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758784 }, { "name": "model.layers.6.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758816 }, { "name": "model.layers.9.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758848 }, { "name": "model.layers.22.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758880 }, { "name": "model.layers.0.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758912 }, { "name": "model.layers.1.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758944 }, { "name": "model.layers.8.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14758976 }, { "name": "model.layers.12.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759008 }, { "name": "model.layers.18.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759040 }, { "name": "model.layers.13.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759072 }, { "name": "model.layers.14.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759104 }, { "name": "model.layers.2.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759136 }, { "name": "model.layers.4.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759168 }, { "name": "model.layers.20.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759200 }, { "name": "model.layers.21.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759232 }, { "name": "model.layers.10.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759264 }, { "name": "model.layers.16.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759296 }, { "name": "model.layers.17.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759328 }, { "name": "model.layers.5.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759360 }, { "name": "model.layers.6.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759392 }, { "name": "model.layers.9.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759424 }, { "name": "model.layers.22.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759456 }, { "name": "model.layers.0.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759488 }, { "name": "model.layers.1.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 14759520 } ], "md5sum": "f715e0ed4e5e2b031d8e714c13166128" } ] }