{ "metadata": { "ParamSize": 194, "ParamBytes": 1447284480.0, "BitsPerParam": 32.0 }, "records": [ { "dataPath": "params_shard_0.bin", "format": "raw-shard", "nbytes": 94371840, "records": [ { "name": "model.embed_tokens.weight", "shape": [ 49152, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 94371840, "byteOffset": 0 } ], "md5sum": "5be787bc4697eb222b2ceca7c8265c81" }, { "dataPath": "params_shard_1.bin", "format": "raw-shard", "nbytes": 24581760, "records": [ { "name": "model.layers.0.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 0 }, { "name": "model.layers.0.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 1920 }, { "name": "model.layers.0.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 4917120 }, { "name": "model.layers.0.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 14747520 }, { "name": "model.layers.0.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 14749440 }, { "name": "model.layers.0.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 17821440 }, { "name": "model.layers.1.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 19664640 }, { "name": "model.layers.1.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 19666560 } ], "md5sum": "b208ccc7647c050e1ba9e94e506f3fc2" }, { "dataPath": "params_shard_2.bin", "format": "raw-shard", "nbytes": 32568960, "records": [ { "name": "model.layers.1.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "model.layers.1.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 9830400 }, { "name": "model.layers.1.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 9832320 }, { "name": "model.layers.1.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 12904320 }, { "name": "model.layers.10.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 14747520 }, { "name": "model.layers.10.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 14749440 }, { "name": "model.layers.10.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 19664640 }, { "name": "model.layers.10.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 29495040 }, { "name": "model.layers.10.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 29496960 } ], "md5sum": "a72d46f969a31c4b480eb55998c49f16" }, { "dataPath": "params_shard_3.bin", "format": "raw-shard", "nbytes": 26424960, "records": [ { "name": "model.layers.10.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 0 }, { "name": "model.layers.11.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 1843200 }, { "name": "model.layers.11.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 1845120 }, { "name": "model.layers.11.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 6760320 }, { "name": "model.layers.11.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 16590720 }, { "name": "model.layers.11.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 16592640 }, { "name": "model.layers.11.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 19664640 }, { "name": "model.layers.12.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 21507840 }, { "name": "model.layers.12.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 21509760 } ], "md5sum": "1726583df239f5946e8052d6d5c4cb92" }, { "dataPath": "params_shard_4.bin", "format": "raw-shard", "nbytes": 32568960, "records": [ { "name": "model.layers.12.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "model.layers.12.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 9830400 }, { "name": "model.layers.12.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 9832320 }, { "name": "model.layers.12.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 12904320 }, { "name": "model.layers.13.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 14747520 }, { "name": "model.layers.13.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 14749440 }, { "name": "model.layers.13.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 19664640 }, { "name": "model.layers.13.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 29495040 }, { "name": "model.layers.13.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 29496960 } ], "md5sum": "84ee06b0e5ab0fe7cae9e1dcda544148" }, { "dataPath": "params_shard_5.bin", "format": "raw-shard", "nbytes": 26424960, "records": [ { "name": "model.layers.13.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 0 }, { "name": "model.layers.14.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 1843200 }, { "name": "model.layers.14.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 1845120 }, { "name": "model.layers.14.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 6760320 }, { "name": "model.layers.14.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 16590720 }, { "name": "model.layers.14.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 16592640 }, { "name": "model.layers.14.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 19664640 }, { "name": "model.layers.15.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 21507840 }, { "name": "model.layers.15.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 21509760 } ], "md5sum": "d5bd068a6960e93562566d5fcc4de701" }, { "dataPath": "params_shard_6.bin", "format": "raw-shard", "nbytes": 32568960, "records": [ { "name": "model.layers.15.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "model.layers.15.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 9830400 }, { "name": "model.layers.15.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 9832320 }, { "name": "model.layers.15.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 12904320 }, { "name": "model.layers.16.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 14747520 }, { "name": "model.layers.16.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 14749440 }, { "name": "model.layers.16.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 19664640 }, { "name": "model.layers.16.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 29495040 }, { "name": "model.layers.16.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 29496960 } ], "md5sum": "d02134a396f92c9df577445bcdcc9c2c" }, { "dataPath": "params_shard_7.bin", "format": "raw-shard", "nbytes": 26424960, "records": [ { "name": "model.layers.16.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 0 }, { "name": "model.layers.17.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 1843200 }, { "name": "model.layers.17.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 1845120 }, { "name": "model.layers.17.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 6760320 }, { "name": "model.layers.17.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 16590720 }, { "name": "model.layers.17.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 16592640 }, { "name": "model.layers.17.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 19664640 }, { "name": "model.layers.18.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 21507840 }, { "name": "model.layers.18.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 21509760 } ], "md5sum": "c97ab7ff7c9a01bcadf9f9d723b29f00" }, { "dataPath": "params_shard_8.bin", "format": "raw-shard", "nbytes": 32568960, "records": [ { "name": "model.layers.18.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "model.layers.18.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 9830400 }, { "name": "model.layers.18.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 9832320 }, { "name": "model.layers.18.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 12904320 }, { "name": "model.layers.19.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 14747520 }, { "name": "model.layers.19.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 14749440 }, { "name": "model.layers.19.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 19664640 }, { "name": "model.layers.19.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 29495040 }, { "name": "model.layers.19.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 29496960 } ], "md5sum": "af2777b1da499317262889bc8e5b5db6" }, { "dataPath": "params_shard_9.bin", "format": "raw-shard", "nbytes": 26424960, "records": [ { "name": "model.layers.19.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 0 }, { "name": "model.layers.2.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 1843200 }, { "name": "model.layers.2.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 1845120 }, { "name": "model.layers.2.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 6760320 }, { "name": "model.layers.2.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 16590720 }, { "name": "model.layers.2.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 16592640 }, { "name": "model.layers.2.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 19664640 }, { "name": "model.layers.20.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 21507840 }, { "name": "model.layers.20.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 21509760 } ], "md5sum": "068b4a9c86ca8c5cb8fb27ef49590304" }, { "dataPath": "params_shard_10.bin", "format": "raw-shard", "nbytes": 32568960, "records": [ { "name": "model.layers.20.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "model.layers.20.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 9830400 }, { "name": "model.layers.20.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 9832320 }, { "name": "model.layers.20.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 12904320 }, { "name": "model.layers.21.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 14747520 }, { "name": "model.layers.21.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 14749440 }, { "name": "model.layers.21.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 19664640 }, { "name": "model.layers.21.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 29495040 }, { "name": "model.layers.21.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 29496960 } ], "md5sum": "e9aed1e309179647df515c92c0a5a8cc" }, { "dataPath": "params_shard_11.bin", "format": "raw-shard", "nbytes": 26424960, "records": [ { "name": "model.layers.21.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 0 }, { "name": "model.layers.22.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 1843200 }, { "name": "model.layers.22.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 1845120 }, { "name": "model.layers.22.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 6760320 }, { "name": "model.layers.22.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 16590720 }, { "name": "model.layers.22.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 16592640 }, { "name": "model.layers.22.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 19664640 }, { "name": "model.layers.23.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 21507840 }, { "name": "model.layers.23.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 21509760 } ], "md5sum": "3a9030a123b73a545fc60d0cd6dd2041" }, { "dataPath": "params_shard_12.bin", "format": "raw-shard", "nbytes": 32568960, "records": [ { "name": "model.layers.23.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "model.layers.23.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 9830400 }, { "name": "model.layers.23.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 9832320 }, { "name": "model.layers.23.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 12904320 }, { "name": "model.layers.24.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 14747520 }, { "name": "model.layers.24.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 14749440 }, { "name": "model.layers.24.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 19664640 }, { "name": "model.layers.24.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 29495040 }, { "name": "model.layers.24.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 29496960 } ], "md5sum": "d862c5790442d551cc80bc0e6f0c3d21" }, { "dataPath": "params_shard_13.bin", "format": "raw-shard", "nbytes": 26424960, "records": [ { "name": "model.layers.24.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 0 }, { "name": "model.layers.25.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 1843200 }, { "name": "model.layers.25.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 1845120 }, { "name": "model.layers.25.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 6760320 }, { "name": "model.layers.25.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 16590720 }, { "name": "model.layers.25.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 16592640 }, { "name": "model.layers.25.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 19664640 }, { "name": "model.layers.26.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 21507840 }, { "name": "model.layers.26.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 21509760 } ], "md5sum": "bcf75b9e95c4d57da8fe7a356c49d71d" }, { "dataPath": "params_shard_14.bin", "format": "raw-shard", "nbytes": 32568960, "records": [ { "name": "model.layers.26.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "model.layers.26.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 9830400 }, { "name": "model.layers.26.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 9832320 }, { "name": "model.layers.26.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 12904320 }, { "name": "model.layers.27.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 14747520 }, { "name": "model.layers.27.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 14749440 }, { "name": "model.layers.27.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 19664640 }, { "name": "model.layers.27.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 29495040 }, { "name": "model.layers.27.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 29496960 } ], "md5sum": "eb8eacc2c6664c516d3cc22d7ea82e05" }, { "dataPath": "params_shard_15.bin", "format": "raw-shard", "nbytes": 26424960, "records": [ { "name": "model.layers.27.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 0 }, { "name": "model.layers.28.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 1843200 }, { "name": "model.layers.28.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 1845120 }, { "name": "model.layers.28.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 6760320 }, { "name": "model.layers.28.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 16590720 }, { "name": "model.layers.28.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 16592640 }, { "name": "model.layers.28.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 19664640 }, { "name": "model.layers.29.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 21507840 }, { "name": "model.layers.29.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 21509760 } ], "md5sum": "ee0875d55d1b9eb9d12e01f951b37507" }, { "dataPath": "params_shard_16.bin", "format": "raw-shard", "nbytes": 32568960, "records": [ { "name": "model.layers.29.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "model.layers.29.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 9830400 }, { "name": "model.layers.29.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 9832320 }, { "name": "model.layers.29.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 12904320 }, { "name": "model.layers.3.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 14747520 }, { "name": "model.layers.3.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 14749440 }, { "name": "model.layers.3.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 19664640 }, { "name": "model.layers.3.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 29495040 }, { "name": "model.layers.3.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 29496960 } ], "md5sum": "2aa59b3731da255d332f150b0400849e" }, { "dataPath": "params_shard_17.bin", "format": "raw-shard", "nbytes": 26424960, "records": [ { "name": "model.layers.3.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 0 }, { "name": "model.layers.30.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 1843200 }, { "name": "model.layers.30.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 1845120 }, { "name": "model.layers.30.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 6760320 }, { "name": "model.layers.30.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 16590720 }, { "name": "model.layers.30.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 16592640 }, { "name": "model.layers.30.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 19664640 }, { "name": "model.layers.31.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 21507840 }, { "name": "model.layers.31.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 21509760 } ], "md5sum": "0138c2d51bdc01afb096d701812495c7" }, { "dataPath": "params_shard_18.bin", "format": "raw-shard", "nbytes": 32568960, "records": [ { "name": "model.layers.31.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "model.layers.31.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 9830400 }, { "name": "model.layers.31.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 9832320 }, { "name": "model.layers.31.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 12904320 }, { "name": "model.layers.4.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 14747520 }, { "name": "model.layers.4.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 14749440 }, { "name": "model.layers.4.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 19664640 }, { "name": "model.layers.4.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 29495040 }, { "name": "model.layers.4.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 29496960 } ], "md5sum": "036511b44e6deffc647f5e9951f05832" }, { "dataPath": "params_shard_19.bin", "format": "raw-shard", "nbytes": 26424960, "records": [ { "name": "model.layers.4.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 0 }, { "name": "model.layers.5.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 1843200 }, { "name": "model.layers.5.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 1845120 }, { "name": "model.layers.5.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 6760320 }, { "name": "model.layers.5.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 16590720 }, { "name": "model.layers.5.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 16592640 }, { "name": "model.layers.5.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 19664640 }, { "name": "model.layers.6.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 21507840 }, { "name": "model.layers.6.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 21509760 } ], "md5sum": "1823faad68f2bc822f8471354077d0b7" }, { "dataPath": "params_shard_20.bin", "format": "raw-shard", "nbytes": 32568960, "records": [ { "name": "model.layers.6.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "model.layers.6.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 9830400 }, { "name": "model.layers.6.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 9832320 }, { "name": "model.layers.6.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 12904320 }, { "name": "model.layers.7.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 14747520 }, { "name": "model.layers.7.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 14749440 }, { "name": "model.layers.7.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 19664640 }, { "name": "model.layers.7.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 29495040 }, { "name": "model.layers.7.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 29496960 } ], "md5sum": "96454401789448760687dfc871799149" }, { "dataPath": "params_shard_21.bin", "format": "raw-shard", "nbytes": 26424960, "records": [ { "name": "model.layers.7.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 0 }, { "name": "model.layers.8.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 1843200 }, { "name": "model.layers.8.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 1845120 }, { "name": "model.layers.8.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 6760320 }, { "name": "model.layers.8.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 16590720 }, { "name": "model.layers.8.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 16592640 }, { "name": "model.layers.8.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 19664640 }, { "name": "model.layers.9.input_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 21507840 }, { "name": "model.layers.9.mlp.down_proj.weight", "shape": [ 960, 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4915200, "byteOffset": 21509760 } ], "md5sum": "2d98f3fb30060da28884ea1fe6c1397f" }, { "dataPath": "params_shard_22.bin", "format": "raw-shard", "nbytes": 14749440, "records": [ { "name": "model.layers.9.mlp.gate_up_proj.weight", "shape": [ 5120, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "model.layers.9.post_attention_layernorm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 9830400 }, { "name": "model.layers.9.self_attn.qkv_proj.weight", "shape": [ 1600, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 3072000, "byteOffset": 9832320 }, { "name": "model.layers.9.self_attn.o_proj.weight", "shape": [ 960, 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1843200, "byteOffset": 12904320 }, { "name": "model.norm.weight", "shape": [ 960 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1920, "byteOffset": 14747520 } ], "md5sum": "ab93c420d97ef880287aaf94ea8975f7" } ] }