{ "metadata": { "ParamSize": 210, "ParamBytes": 10457367552.0, "BitsPerParam": 32.0 }, "records": [ { "dataPath": "params_shard_0.bin", "format": "raw-shard", "nbytes": 1179648000, "records": [ { "name": "model.embed_tokens.weight", "shape": [ 256000, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 1179648000, "byteOffset": 0 } ], "md5sum": "a3af7a56ca8b021a828ae0c541922426" }, { "dataPath": "params_shard_1.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.0.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "491c998ac863be9af01e763845324c08" }, { "dataPath": "params_shard_2.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.0.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "39073d97a3ec272a731ee2331653e6d9" }, { "dataPath": "params_shard_3.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.1.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "54b6184aeb869659128f2d1b63a1f71e" }, { "dataPath": "params_shard_4.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.1.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "a24f9b50a18ef172b4240290fa86035f" }, { "dataPath": "params_shard_5.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.layers.1.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "81be37c6285659097108dab7e1167610" }, { "dataPath": "params_shard_6.bin", "format": "raw-shard", "nbytes": 28348416, "records": [ { "name": "model.layers.0.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 0 }, { "name": "model.layers.0.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 4608 }, { "name": "model.layers.0.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9216 }, { "name": "model.layers.0.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 13824 }, { "name": "model.layers.0.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 18432 }, { "name": "model.layers.0.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 18892800 }, { "name": "model.layers.1.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 28329984 }, { "name": "model.layers.1.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 28334592 }, { "name": "model.layers.1.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 28339200 }, { "name": "model.layers.1.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 28343808 } ], "md5sum": "3aca32982f29aff34ee83f2ca4fffc36" }, { "dataPath": "params_shard_7.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.10.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "f0c82749a36607fcca1aebe819918988" }, { "dataPath": "params_shard_8.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.10.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "d98c81b6ff6c3e625a6d9178f03869b9" }, { "dataPath": "params_shard_9.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.1.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.10.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.10.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.10.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.10.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.10.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "9f15986eb6cf2feb330374aa886f574f" }, { "dataPath": "params_shard_10.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.11.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "ba3e941319f5bd84567e7cdff01d562c" }, { "dataPath": "params_shard_11.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.11.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "133664ccf5f0a780c3d527c576f83c82" }, { "dataPath": "params_shard_12.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.10.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.11.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.11.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.11.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.11.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.11.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "c80ffba3d542f5244d7b47ee0686fd15" }, { "dataPath": "params_shard_13.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.12.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "82da0d1860cde0cb675a3ed2a433854c" }, { "dataPath": "params_shard_14.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.12.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "4dbae0a83b35dda0cde8e209e8abc160" }, { "dataPath": "params_shard_15.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.11.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.12.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.12.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.12.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.12.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.12.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "40887d58fb093c33e70f668cf4bf6cee" }, { "dataPath": "params_shard_16.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.13.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "3c99c24df6ecc72184bc666fa37713ed" }, { "dataPath": "params_shard_17.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.13.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "a03408f01d7845d5dd0fdc96981b9795" }, { "dataPath": "params_shard_18.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.12.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.13.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.13.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.13.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.13.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.13.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "4ca815fde850b6d0caec167e60d28116" }, { "dataPath": "params_shard_19.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.14.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "2e41a8aef9bf37abca7872f0bd02441c" }, { "dataPath": "params_shard_20.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.14.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "721736733522c0c2d6928724718b8d80" }, { "dataPath": "params_shard_21.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.13.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.14.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.14.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.14.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.14.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.14.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "f94ebe81d257e783b233265f2ed74c31" }, { "dataPath": "params_shard_22.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.15.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "90b89a80cfc1602369ee44c5d85f1e03" }, { "dataPath": "params_shard_23.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.15.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "f7359bf5843f9300d7b10c1ebf5f1156" }, { "dataPath": "params_shard_24.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.14.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.15.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.15.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.15.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.15.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.15.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "a5d4978c2cb0d3b70036f9fe185d8fd2" }, { "dataPath": "params_shard_25.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.16.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "7e12841b8aabd4d6feecb1269dffaf79" }, { "dataPath": "params_shard_26.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.16.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "676a40b3e8c491f714185ed2e968b50a" }, { "dataPath": "params_shard_27.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.15.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.16.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.16.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.16.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.16.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.16.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "26c69195a7d9bc01816ca7e652779582" }, { "dataPath": "params_shard_28.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.17.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "9e7e3ec3f7a55af727010709324c15db" }, { "dataPath": "params_shard_29.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.17.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "54ccc2b6574312de7af096c8c61adaf9" }, { "dataPath": "params_shard_30.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.16.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.17.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.17.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.17.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.17.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.17.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "2c1de3496b09581ed287384f3cbbe1b6" }, { "dataPath": "params_shard_31.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.18.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "b88955f08bd540b803f1ccdd03306698" }, { "dataPath": "params_shard_32.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.18.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "f3e3014e24bcff20e80b61fa2185999f" }, { "dataPath": "params_shard_33.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.17.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.18.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.18.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.18.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.18.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.18.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "e893e73d14fea8bc08faf7382a6ff46d" }, { "dataPath": "params_shard_34.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.19.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "e3a374df504f2f5f4d47394d57d7022b" }, { "dataPath": "params_shard_35.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.19.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "696543e0b88e30609403bbbcf50bf083" }, { "dataPath": "params_shard_36.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.18.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.19.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.19.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.19.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.19.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.19.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "a1594005ea5dfa20f1e70fa6b63da8bd" }, { "dataPath": "params_shard_37.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.2.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "ae73e2b9420ec8b7686ef1f2a77ad27f" }, { "dataPath": "params_shard_38.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.2.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "c342564cc32289df1d3870f7cc00f5b6" }, { "dataPath": "params_shard_39.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.19.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.2.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.2.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.2.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.2.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.2.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "a9ccf3fbd38aa220652a80a8a29e52ed" }, { "dataPath": "params_shard_40.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.20.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "8cd76694c761abda3b06b03a36bc07a2" }, { "dataPath": "params_shard_41.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.20.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "8c20e62ac573b44910efdf44c9ac91b9" }, { "dataPath": "params_shard_42.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.2.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.20.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.20.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.20.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.20.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.20.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "09ebddcaaca0ab94453db47df26d3b37" }, { "dataPath": "params_shard_43.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.21.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "8313cf6d66b756c4590b67fcb1670bd5" }, { "dataPath": "params_shard_44.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.21.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "590a2da4d50cb705ffccac5190c81c3a" }, { "dataPath": "params_shard_45.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.20.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.21.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.21.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.21.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.21.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.21.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "bbe34bca37764c8c1de9ca37b6f4d6fd" }, { "dataPath": "params_shard_46.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.22.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "433725444ea45ece6b438f1f4eb1230b" }, { "dataPath": "params_shard_47.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.22.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "307ab0846f737090154b6909d89185d9" }, { "dataPath": "params_shard_48.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.21.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.22.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.22.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.22.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.22.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.22.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "1442e2af6d973d018d1c0cc4cd9036e3" }, { "dataPath": "params_shard_49.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.23.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "80e399a33f1ea124a251d8c33b21a99a" }, { "dataPath": "params_shard_50.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.23.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "a7123b5486f1f11071a442c9f5a674aa" }, { "dataPath": "params_shard_51.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.22.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.23.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.23.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.23.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.23.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.23.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "b350d36ef97afd138743456172cded1a" }, { "dataPath": "params_shard_52.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.24.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "3f91c632a3afcc3f848e066fc9fc277d" }, { "dataPath": "params_shard_53.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.23.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.24.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9437184 } ], "md5sum": "91c17d484c0d09522301873050231f44" }, { "dataPath": "params_shard_54.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.3.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "4c29b4a171556d71545eea0b7d1b1ab3" }, { "dataPath": "params_shard_55.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.3.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "fd7945e3f5b8dff9e065ca090a83f374" }, { "dataPath": "params_shard_56.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.24.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.3.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.3.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.3.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.3.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.3.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "1318190259728d3fa1e9f128e4269a3b" }, { "dataPath": "params_shard_57.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.4.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "cf22c371cd3e4f9db93b8867fe14bf7c" }, { "dataPath": "params_shard_58.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.4.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "6a725295146c0ba5677e124e2fe0da34" }, { "dataPath": "params_shard_59.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.3.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.4.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.4.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.4.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.4.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.4.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "72c2ba05a0e0942ccd11b67a21f45cfd" }, { "dataPath": "params_shard_60.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.5.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "6ec841acc8e17ebfa315bf3985b69956" }, { "dataPath": "params_shard_61.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.5.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "2c860adc2f1ad9e44549cbc43a2e9975" }, { "dataPath": "params_shard_62.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.4.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.5.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.5.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.5.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.5.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.5.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "7a5da0311b590481f07a2c6ea22d8dea" }, { "dataPath": "params_shard_63.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.6.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "08797b1aa61dbf64a4e4df7a8be2a8b4" }, { "dataPath": "params_shard_64.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.6.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "3b527f48312ff17dabb69cc66248a62b" }, { "dataPath": "params_shard_65.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.5.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.6.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.6.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.6.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.6.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.6.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "a1e272ec364e278cb17c52a595a3ab9a" }, { "dataPath": "params_shard_66.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.7.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "1aefe1406be82a8a0460a21c1290f891" }, { "dataPath": "params_shard_67.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.7.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "e041ef02cae52d0ed220fecb8ebf90c0" }, { "dataPath": "params_shard_68.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.6.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.7.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.7.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.7.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.7.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.7.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "038478141467fd0252ee1f30c0521018" }, { "dataPath": "params_shard_69.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.8.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "2b4beed107c3497441069808f3935d4c" }, { "dataPath": "params_shard_70.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.8.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "4a3537ab420b63b869d40733dfd14a00" }, { "dataPath": "params_shard_71.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.7.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.8.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.8.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.8.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.8.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.8.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "26f29cb74eaeb8390baeace23314f1f2" }, { "dataPath": "params_shard_72.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.9.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "9668ccce68b22ccd35de7c18001e1ddf" }, { "dataPath": "params_shard_73.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.9.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "a7ca07fc82c6f615e735761d387319e4" }, { "dataPath": "params_shard_74.bin", "format": "raw-shard", "nbytes": 28329984, "records": [ { "name": "model.layers.8.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.9.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.9.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.9.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.9.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.9.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9455616 } ], "md5sum": "2176f447c0bfe088191390dacd403c07" }, { "dataPath": "params_shard_75.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.24.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "ddc4ab12a1794b40d7a7d2f4f5744d45" }, { "dataPath": "params_shard_76.bin", "format": "raw-shard", "nbytes": 42467328, "records": [ { "name": "model.layers.25.mlp.down_proj.weight", "shape": [ 2304, 9216 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 42467328, "byteOffset": 0 } ], "md5sum": "6cb8e7647db51acf325430ce0247514d" }, { "dataPath": "params_shard_77.bin", "format": "raw-shard", "nbytes": 84934656, "records": [ { "name": "model.layers.25.mlp.gate_up_proj.weight", "shape": [ 18432, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 84934656, "byteOffset": 0 } ], "md5sum": "4bb0a856c263aaa8d4224c2dd80f1ba9" }, { "dataPath": "params_shard_78.bin", "format": "raw-shard", "nbytes": 28348416, "records": [ { "name": "model.layers.9.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.layers.24.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 }, { "name": "model.layers.24.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9441792 }, { "name": "model.layers.24.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9446400 }, { "name": "model.layers.24.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9451008 }, { "name": "model.layers.25.input_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9455616 }, { "name": "model.layers.25.post_attention_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9460224 }, { "name": "model.layers.25.post_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9464832 }, { "name": "model.layers.25.pre_feedforward_layernorm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9469440 }, { "name": "model.layers.25.self_attn.qkv_proj.weight", "shape": [ 4096, 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 9474048 } ], "md5sum": "1653aca45ae3049e61516e3eac1dcf5d" }, { "dataPath": "params_shard_79.bin", "format": "raw-shard", "nbytes": 9441792, "records": [ { "name": "model.layers.25.self_attn.o_proj.weight", "shape": [ 2304, 2048 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 9437184, "byteOffset": 0 }, { "name": "model.norm.weight", "shape": [ 2304 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 9437184 } ], "md5sum": "8cb41bb43fc166d2d23be3e590948a66" } ] }