diff --git "a/ndarray-cache-b16.json" "b/ndarray-cache-b16.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache-b16.json" @@ -0,0 +1,14153 @@ +{ + "metadata": { + "ParamSize": 803, + "ParamBytes": 72656372992.0, + "BitsPerParam": 7.974819159215532 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 2101346304, + "records": [ + { + "name": "lm_head.weight", + "shape": [ + 128256, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2101346304, + "byteOffset": 0 + } + ], + "md5sum": "b450ad6564dd69ce196a7a66a8c66182" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 2101346304, + "records": [ + { + "name": "model.embed_tokens.weight", + "shape": [ + 128256, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2101346304, + "byteOffset": 0 + } + ], + "md5sum": "309be16492736a951c706ce5cd988ca1" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "797cfd37229f030c162a31ca1cb8b0dc" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "bef7c44ed2d080ed4358eb51385c27fd" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.0.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "1ef2429046a53c9d399a2917496c3644" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "2ab2d0a67816ef1836379327fcc519a2" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "fec9a3ec6955cd3d94d77adb87677686" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.1.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "b62c6f93216e9a37181a7bb237864e2c" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "6f090a4991ed582ef9a71e2c0d1e566d" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "940457edfe2a861c434c6c50ac2f66a9" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "598de1fab03ef0a9ebadb73a6104dde9" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "c0df5d3e67a6ffe2c7988c09b088d320" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.2.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "0fd8bcca57bee283ecdac25b12b29731" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "920b40904a5ced8e25afe586a06a2c89" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "7fd84684a5700f544e270a325aaa84a7" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "e23177b40fc2425da1dd50fe828be6a9" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.3.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "ce17ffc4528b50e274af94a2ff6f8538" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "f64ef48625d1e104a0fb985c69c0182d" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.4.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "7d8b7850356018aed9df334a9da05acd" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "45d220179063b33c173792305eca6ae3" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "7a6432de03b238e673b1b0cbac40973c" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "11f9b80d8d322938d2a22aba30baf424" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.10.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "5f8b664b7aa863e9ef3e620e272a4032" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "f74d8c58abc34a02b2917f3f97224872" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "ffa0ef958981d7a121d2319271819fac" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "af0a352a7c4402da245352062e91d1cd" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.11.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "ce08b2c98b6cb7a110aa093a34421fb3" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "4741d3ad7694df52870a859633ea620c" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "7de99ba3f772c551c22fc79978029341" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.12.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "1b0c6aaa943ea8d8cf7ac4b2effea528" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "0bd1381bf025a121130b03bdc8f7bed0" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "b53ab505054502b207db81198f211666" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "9b0db11af63270291ec27e8505e13d6f" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "5c7d1f1136b14788da47bf629ecc6e2e" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.13.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "543bb746335ccf7491619ebe2e7f1c3e" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "5c43136ed1e71843e031889dcfbad104" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "94a5b8ac8e3828aabffefbc4a0745757" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "c1ae32f04e71c50b56bbf54598e46248" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.14.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "0b6380399d035b09e712f7ca0dabf74a" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "0bab33f53a5d3dc3053fd723ce8f59dc" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "7c7e8073c1b12462c373b2fb4511ffe1" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.15.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "3cf44f7f71d81e22c4fa93cd27bca168" + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "cd1d0121161be1105ec6056138e6de9f" + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "d55bba6b4623d75aff034fbe29977c93" + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "1e5b67d012cf5bec8f21263b94132823" + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "16c579770bdc7203fb1c15a30be5e747" + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.16.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "30631b8f70ebd2eab457db5eb3a1d61d" + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "d69087ad3d2b43fd142e9afc5e17086d" + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "7d533331c7e2dd0c7feb32d9d1f718c1" + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "68ba3e28dadc0f281e40d8f915925c7f" + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.17.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "5ca627fa46d02b7099549f4da40d7a9f" + }, + { + "dataPath": "params_shard_51.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "35d2aa7e583d7e8ed31b103d092d68d2" + }, + { + "dataPath": "params_shard_52.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.18.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "85083020920b23a76c10ed04f380a807" + }, + { + "dataPath": "params_shard_53.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "5bdff2f392f0804b5059ee97798fb9e6" + }, + { + "dataPath": "params_shard_54.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "fe8758ae9bd2a93b207ff3ee723a1111" + }, + { + "dataPath": "params_shard_55.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "20d29e0d44ab4b4a94229300b8537b87" + }, + { + "dataPath": "params_shard_56.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "ce9da9d0ae00a3e730b1edf5e913b850" + }, + { + "dataPath": "params_shard_57.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "da93cb0b7f4cf9fa895ada3cfd706131" + }, + { + "dataPath": "params_shard_58.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.19.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "19049bd95baa50d73b9040e241ff8e2c" + }, + { + "dataPath": "params_shard_59.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "23ab931b4261417d39ce0da2fada27b9" + }, + { + "dataPath": "params_shard_60.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "a7cb50d94b5543893486ab09ce3a4607" + }, + { + "dataPath": "params_shard_61.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "2584050d9066d45faca22797e136f030" + }, + { + "dataPath": "params_shard_62.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.20.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "9916dccb3ba835fbc6da5596fbfa3835" + }, + { + "dataPath": "params_shard_63.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "e39bd62dd91ecc9b4d71802246a8f06c" + }, + { + "dataPath": "params_shard_64.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.21.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "a083a5ae6b04b1f3f0429bcf8701339a" + }, + { + "dataPath": "params_shard_65.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "3478326a1ff16259fe9ac3ae9bf83f7e" + }, + { + "dataPath": "params_shard_66.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "7a160e43b5455522a5b383d28c3f8068" + }, + { + "dataPath": "params_shard_67.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "e7e3053f6ebd3c133ca96fd7e38c6f91" + }, + { + "dataPath": "params_shard_68.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "442a7bf7223d1aaa97c37a9103671b84" + }, + { + "dataPath": "params_shard_69.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "f19ce5ae0a7990123d41f2b8a47e3cf8" + }, + { + "dataPath": "params_shard_70.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.22.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "9f4625563a90399deb69b5e7f7b19ed1" + }, + { + "dataPath": "params_shard_71.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "c2ea24c720201b4329f137caad792ce9" + }, + { + "dataPath": "params_shard_72.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "03ad1e19d8380394696118780176e574" + }, + { + "dataPath": "params_shard_73.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "8d2785de78aaa291da576d3ffb15ef5b" + }, + { + "dataPath": "params_shard_74.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.23.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "073c0efd3a39470d2cfee04381891409" + }, + { + "dataPath": "params_shard_75.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "a5319c5baa2ef0e5039c53fb6adb1d09" + }, + { + "dataPath": "params_shard_76.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "0bf4630d716d5cece7bcfa606a9eba79" + }, + { + "dataPath": "params_shard_77.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "29847eff9e993ce7cfe1e42da955cd1e" + }, + { + "dataPath": "params_shard_78.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.24.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "a4a60d4c10309c9d76b2d82bd346e96c" + }, + { + "dataPath": "params_shard_79.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "2b74d6fed1536da694081b7108452a40" + }, + { + "dataPath": "params_shard_80.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "8f63e4bc22bf5980e85ee44eaa296dda" + }, + { + "dataPath": "params_shard_81.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "372d993a6d094d6d2054a48106c96f24" + }, + { + "dataPath": "params_shard_82.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.25.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "a7d86e2f9f19ec285e72a15f0e554ef0" + }, + { + "dataPath": "params_shard_83.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "911bec90f66ada4405e9998de55d3901" + }, + { + "dataPath": "params_shard_84.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.26.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "8b633cefa0b7c3794a0757ca719804f3" + }, + { + "dataPath": "params_shard_85.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.26.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "378b1ca4d5e846546cb130f7ff0a2420" + }, + { + "dataPath": "params_shard_86.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.26.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "967a75a9fddd48bbddd54e40ec1ffb3c" + }, + { + "dataPath": "params_shard_87.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.26.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "c4ce3288df351f184495461644a00b59" + }, + { + "dataPath": "params_shard_88.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.27.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "3bd19456fa54b12519c367312baa7fba" + }, + { + "dataPath": "params_shard_89.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.27.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "4bb9e3b8d19568589d4663cc05ebc1a7" + }, + { + "dataPath": "params_shard_90.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.27.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "fe2206acb00a46c7868bf7cac6c27093" + }, + { + "dataPath": "params_shard_91.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.27.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "68b26a27560ead857d460b4159517fc1" + }, + { + "dataPath": "params_shard_92.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.28.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "56900a10ff7e722af7a70218dfcd4499" + }, + { + "dataPath": "params_shard_93.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.28.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "3846c8c940a3ae41e3057f98c2635381" + }, + { + "dataPath": "params_shard_94.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.28.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "c966fdd6d5786ec9f8a3b99b726f379b" + }, + { + "dataPath": "params_shard_95.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.28.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "16e603175c392bf32f4978e7e598a4aa" + }, + { + "dataPath": "params_shard_96.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.29.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "6bf6a369663a91dc6f6a6d8a254cbdb0" + }, + { + "dataPath": "params_shard_97.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.29.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "ac70ac93669ff848a0ff7018380fd72b" + }, + { + "dataPath": "params_shard_98.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.29.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "1c1f6e0c25ed4aeea1db2e907a87fb51" + }, + { + "dataPath": "params_shard_99.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.29.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "860731352abd5f81f505152a059739cd" + }, + { + "dataPath": "params_shard_100.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.30.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "ed755d14b493a75f7bf6f28d29c30077" + }, + { + "dataPath": "params_shard_101.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.30.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "78da8af090aa97327d413a820327b175" + }, + { + "dataPath": "params_shard_102.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.30.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "a06a9cc40d997487fbdfc6999a9bc22d" + }, + { + "dataPath": "params_shard_103.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.30.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "4e8a871d7d35f44fe8effe57667cd950" + }, + { + "dataPath": "params_shard_104.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.31.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "b036939008d8c9b52c4142b65f4b86af" + }, + { + "dataPath": "params_shard_105.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.31.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "e6699abb579e38549de1f2924e74f9eb" + }, + { + "dataPath": "params_shard_106.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.31.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "9e552efcae3b2464a21d08b1bf36d340" + }, + { + "dataPath": "params_shard_107.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.31.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "8a1c0529478df4f3584e652a68fb3f61" + }, + { + "dataPath": "params_shard_108.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.32.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "b428e6b8309838d12cfe50b64ec2d6c9" + }, + { + "dataPath": "params_shard_109.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.32.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "faabf864cf5a7549b3a59dd41783ebe3" + }, + { + "dataPath": "params_shard_110.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.32.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "86379f769dff50378df13638f55c6e30" + }, + { + "dataPath": "params_shard_111.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.32.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "9a8a05339b752995403f0abdf26f9e91" + }, + { + "dataPath": "params_shard_112.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.33.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "5b32c666bfa36e161921eb76bce039d5" + }, + { + "dataPath": "params_shard_113.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.33.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "2bd9a88fa11bf8f1a2f8e906736cd0f3" + }, + { + "dataPath": "params_shard_114.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.33.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "3d8ea10ed5c3141baedacbe071b36ae7" + }, + { + "dataPath": "params_shard_115.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.33.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "935404740012cae8cd2b7e25eb12dbf2" + }, + { + "dataPath": "params_shard_116.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.34.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "f8021b7ef1516cdb24ccf6e02df223e8" + }, + { + "dataPath": "params_shard_117.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.34.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "4c813ac501f8ae9c2deffb8bf46f18e0" + }, + { + "dataPath": "params_shard_118.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.34.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "89acaa356bf9e4243860fe4cb84c0c47" + }, + { + "dataPath": "params_shard_119.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.34.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "63910765915d469ac20be486310d0c92" + }, + { + "dataPath": "params_shard_120.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.35.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "4568fbc2c56ba9e1638eb3e1b651eefd" + }, + { + "dataPath": "params_shard_121.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.35.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "4cb741a6e225d8b2bbd653c5bc075078" + }, + { + "dataPath": "params_shard_122.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.35.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "1044682dbfe15115f78ad4f354d71905" + }, + { + "dataPath": "params_shard_123.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.35.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "7f39ce6ca289ac6ac6adb6a153cd66a9" + }, + { + "dataPath": "params_shard_124.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.36.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "a1ce7dfea3a1a6c9b19388ddc2a07545" + }, + { + "dataPath": "params_shard_125.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.36.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "eff5c99d9481a365806c23c40f9bb598" + }, + { + "dataPath": "params_shard_126.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.36.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "fb5228e36bbbc7030fed74844ee4ff2c" + }, + { + "dataPath": "params_shard_127.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.36.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "b0500301e26f4352a3562091dd80485b" + }, + { + "dataPath": "params_shard_128.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.37.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "c510648f77883c1974e91a4849c1bc58" + }, + { + "dataPath": "params_shard_129.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.37.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "320e2de7c8eb870a473f8243e8be1ca8" + }, + { + "dataPath": "params_shard_130.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.37.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "198b9fcc168b813c513d8ae5b559a5b1" + }, + { + "dataPath": "params_shard_131.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.37.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "058416cf1440072de887057911ad1340" + }, + { + "dataPath": "params_shard_132.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.38.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "54e41ed6c4fda43508d43fd4c08c44ad" + }, + { + "dataPath": "params_shard_133.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.38.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "612c05bd0626b92454f06e9a19f8bd7c" + }, + { + "dataPath": "params_shard_134.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.38.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "b05ea46fd83c8c61e39edc62c9df0874" + }, + { + "dataPath": "params_shard_135.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.38.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "b1c374bea953bfdb3b107ae15c3c6453" + }, + { + "dataPath": "params_shard_136.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.39.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "f4c0d59bf7ac1e45d7f0492a74f8221b" + }, + { + "dataPath": "params_shard_137.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.39.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "1b8e06371ab577b847e03ef7e495eb47" + }, + { + "dataPath": "params_shard_138.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.39.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "f83b078d95b192e74841f2d85ffebc77" + }, + { + "dataPath": "params_shard_139.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.39.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "ce5203ba040e20f74f392574f6ec2c27" + }, + { + "dataPath": "params_shard_140.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.40.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "2b5c724c1787dce3bc5b40b657a9f0c3" + }, + { + "dataPath": "params_shard_141.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.40.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "0acd132c00f93037e84cacaa0c901db2" + }, + { + "dataPath": "params_shard_142.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.40.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "5d083551fd1154a4eaec730b4f669ae7" + }, + { + "dataPath": "params_shard_143.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "3de10e670a8d92268d55a95eb4b39cec" + }, + { + "dataPath": "params_shard_144.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "9f1f917e4ccb64d6d314aa0e717a6de9" + }, + { + "dataPath": "params_shard_145.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "51bc54e46c05f26fe97d5d64aec88115" + }, + { + "dataPath": "params_shard_146.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "a312eb11689ee4a6af9e8a6205323948" + }, + { + "dataPath": "params_shard_147.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.5.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "2be714f964f3d01032ab85611bab16d8" + }, + { + "dataPath": "params_shard_148.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "db5470409b80507b402f77b30d981522" + }, + { + "dataPath": "params_shard_149.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "4d209c240ffcdd32a4c85971f955a3e0" + }, + { + "dataPath": "params_shard_150.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "39319df4503463297554203260385952" + }, + { + "dataPath": "params_shard_151.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.6.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "9e1ba5f4410d11707bfc2350256fbc2c" + }, + { + "dataPath": "params_shard_152.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "eb3fde29c768decfa59683b10bc34cca" + }, + { + "dataPath": "params_shard_153.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.7.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "68179c2bf36f76dc5a5bf9dd3ee99802" + }, + { + "dataPath": "params_shard_154.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.40.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "ced073a5194fa57d07cc4a2995c494ef" + }, + { + "dataPath": "params_shard_155.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.41.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "5a9e87858f202b5568e8fef9a5f032b2" + }, + { + "dataPath": "params_shard_156.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.41.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "688c514c3e735e7d4efc7ee2df3d5a8c" + }, + { + "dataPath": "params_shard_157.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.41.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "ab9d00ae6acd35d586802ac94912005b" + }, + { + "dataPath": "params_shard_158.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.41.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "05bd4d6d964f0abb383c8f061db2ef90" + }, + { + "dataPath": "params_shard_159.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.42.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "ce63beb60f1dfd8371c696bf3d5b8b3d" + }, + { + "dataPath": "params_shard_160.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.42.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "7b3c24455e8e344e545234353bbb73cd" + }, + { + "dataPath": "params_shard_161.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.42.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "98db56c72f28bd692bafc43550f84fc1" + }, + { + "dataPath": "params_shard_162.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.42.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "bea937c98812d62afdb1a82444d8d1e8" + }, + { + "dataPath": "params_shard_163.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.43.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "6afaa589d657620509282d2d6659fd92" + }, + { + "dataPath": "params_shard_164.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.43.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "4595e8d02a7b79367fa898152434b894" + }, + { + "dataPath": "params_shard_165.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.43.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "5bbfb5340bbfc5d6ac08b02fbac35521" + }, + { + "dataPath": "params_shard_166.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.43.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "fffc8202c774831dbba55b14960744b3" + }, + { + "dataPath": "params_shard_167.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.44.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "ec36939eb51dd07b5eaf4603338299e5" + }, + { + "dataPath": "params_shard_168.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.44.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "73524b4f19dfcf6b6f2140428a725563" + }, + { + "dataPath": "params_shard_169.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.44.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "9cbb0b020fe1301effcb230a81d2093e" + }, + { + "dataPath": "params_shard_170.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.44.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "8564ba637ef6eb63671646b13f4ecbcf" + }, + { + "dataPath": "params_shard_171.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.45.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "f90f1bf74663bec68a48a5047d13be2c" + }, + { + "dataPath": "params_shard_172.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.45.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "0cf5103dddb071c5fb81b10c525f41d1" + }, + { + "dataPath": "params_shard_173.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.45.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "701d5ed4d7a22cc994759eaa552e3a6f" + }, + { + "dataPath": "params_shard_174.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.45.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "df28c85499d563c541d4d5d4825de22e" + }, + { + "dataPath": "params_shard_175.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.46.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "880de919a371642c18a2b7571336dc74" + }, + { + "dataPath": "params_shard_176.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.46.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "336059f899161f73cf889965523e54cf" + }, + { + "dataPath": "params_shard_177.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.46.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "ba901a961e071a8fcceeeda99236e138" + }, + { + "dataPath": "params_shard_178.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.46.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "322ee9b5e373adad5e453cd1b25a36eb" + }, + { + "dataPath": "params_shard_179.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.47.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "0eb7d4efd04b1a51ae01238f4f71607c" + }, + { + "dataPath": "params_shard_180.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.47.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "9a1d9d3e3d82bbe642b4103f001fc4bf" + }, + { + "dataPath": "params_shard_181.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.47.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "f70ab1d7ca16b65650c15c0d8a51db39" + }, + { + "dataPath": "params_shard_182.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.47.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "97dd4670c72a55f39e79cb9bd02c192b" + }, + { + "dataPath": "params_shard_183.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.48.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "c71936745fac05f7e32b697453ab3321" + }, + { + "dataPath": "params_shard_184.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.48.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "3a4d353d3a8776386dc7ed78058b399b" + }, + { + "dataPath": "params_shard_185.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.48.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "5aeaaab66ecce4ce775004a4dbd5824a" + }, + { + "dataPath": "params_shard_186.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.48.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "409b178b713de43f236a4e15d59a8498" + }, + { + "dataPath": "params_shard_187.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.49.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "124bc4a69a58d0adfdca10cd08489900" + }, + { + "dataPath": "params_shard_188.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.49.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "5c8e2158a2e72e6e5e5c5b60332dd3da" + }, + { + "dataPath": "params_shard_189.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.49.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "1c23a392b40679fb955febca06c506ce" + }, + { + "dataPath": "params_shard_190.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.49.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "66f5485d89f67710688aa42619366dd8" + }, + { + "dataPath": "params_shard_191.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.50.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "849a69d3105c6297ae888ca24b5ef758" + }, + { + "dataPath": "params_shard_192.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.50.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "2053ef3ce843d79bf73e82c1795e115f" + }, + { + "dataPath": "params_shard_193.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.50.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "28083fd9026e262d993fe9a86b165682" + }, + { + "dataPath": "params_shard_194.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.50.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "24262643adea36c24108b34544bb27c3" + }, + { + "dataPath": "params_shard_195.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.51.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "289f30e51ad56daab06792e6b7b11d39" + }, + { + "dataPath": "params_shard_196.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.51.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "4ab0f4516a273d0aaafec0ee29229681" + }, + { + "dataPath": "params_shard_197.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.51.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "71401bfbe3721db7870f8722edeb1751" + }, + { + "dataPath": "params_shard_198.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.51.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "d61719aa66adc2ca042a9debc2474851" + }, + { + "dataPath": "params_shard_199.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.52.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "3e6cf21f546fe81b66d152c65ce25cee" + }, + { + "dataPath": "params_shard_200.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.52.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "3149070358e4f695de1587e0b40b26cd" + }, + { + "dataPath": "params_shard_201.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.52.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "7550f2936b9fadfe5b5dcae971d1e9e2" + }, + { + "dataPath": "params_shard_202.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.52.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "c9dad67f8312f209af66faf384e337c5" + }, + { + "dataPath": "params_shard_203.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.53.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "65a85b6688ca9c45e6b5b78fa49fc8b4" + }, + { + "dataPath": "params_shard_204.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.53.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "e85655b1a4b62640054169601d79dc13" + }, + { + "dataPath": "params_shard_205.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.53.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "56cc3a0f9128b2184b290110ed195ad3" + }, + { + "dataPath": "params_shard_206.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.53.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "315a1f00462d3096ca318c604f1908c9" + }, + { + "dataPath": "params_shard_207.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.54.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "665c6a6e56cb342b27d5c467c0a54218" + }, + { + "dataPath": "params_shard_208.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.54.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "bc3d4f5451fb629377d2e3a1eb6c68a9" + }, + { + "dataPath": "params_shard_209.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.54.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "4c6af06bc8b2dfb14470a7a0f06f7466" + }, + { + "dataPath": "params_shard_210.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.54.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "d69aead6b77d7efd5ad054a9ceda9980" + }, + { + "dataPath": "params_shard_211.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.55.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "efc1038f95a7c91637b6a495d56501c2" + }, + { + "dataPath": "params_shard_212.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.55.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "d0bfa6a0d0e11ed613a3929ae03f2c4c" + }, + { + "dataPath": "params_shard_213.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.55.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "27080a2884f1b466110800e99e090cff" + }, + { + "dataPath": "params_shard_214.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.55.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "dbeede2c27447137e9f3ad0a93fd2cb2" + }, + { + "dataPath": "params_shard_215.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.56.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "02740f2c04da12951504adff6c0b7fca" + }, + { + "dataPath": "params_shard_216.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.56.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "26e57944ac7b2d3f9b5f87ee7106cd1a" + }, + { + "dataPath": "params_shard_217.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.56.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "3d1b0525a0b46e993b8c8f519c616b32" + }, + { + "dataPath": "params_shard_218.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.56.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "3359361aa99ef0f466f2fe00b7491744" + }, + { + "dataPath": "params_shard_219.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.57.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "367ad87ce6e00da1efa31a0560c493b3" + }, + { + "dataPath": "params_shard_220.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.57.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "099544a5fcea86c8ed7201d93e251561" + }, + { + "dataPath": "params_shard_221.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.57.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "82c66372dde3e01598be4922e702a4b5" + }, + { + "dataPath": "params_shard_222.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.57.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "893fa4313736a16208e34e88dd7f9676" + }, + { + "dataPath": "params_shard_223.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.58.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "1fe83c81c8e51fab0c0a2de35edfce2d" + }, + { + "dataPath": "params_shard_224.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.58.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "7d477bf098cd0ebd9c10061c86429a60" + }, + { + "dataPath": "params_shard_225.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.58.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "d1bcb4069def2c871aba3723e894e894" + }, + { + "dataPath": "params_shard_226.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.58.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "2bdb3f83aa4684de8ff3c32e769c721d" + }, + { + "dataPath": "params_shard_227.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.59.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "241c2fb0eb8e2dfa485fb8cf9306c992" + }, + { + "dataPath": "params_shard_228.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.59.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "ba272ba71b65054c23b5be18204a4c01" + }, + { + "dataPath": "params_shard_229.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.59.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "d9b1467a79a1610ad87943e55df47268" + }, + { + "dataPath": "params_shard_230.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.59.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "0640462d484442ad40c10cb5761e57d2" + }, + { + "dataPath": "params_shard_231.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.60.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "1c34540f85020970d2d3dd8e9d00f6da" + }, + { + "dataPath": "params_shard_232.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.60.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "200855e757c63ad5ff5173d676a12c37" + }, + { + "dataPath": "params_shard_233.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.60.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "ee911906a686fa8f76aed4ef02ead1a4" + }, + { + "dataPath": "params_shard_234.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.60.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "07cc314f09f934c3143ee09fef1bd698" + }, + { + "dataPath": "params_shard_235.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.61.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "d3f4deff995747a359a515bd422f490a" + }, + { + "dataPath": "params_shard_236.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.61.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "1282748b15f861be4ee9b1067af5ab1b" + }, + { + "dataPath": "params_shard_237.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.61.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "d624b9ae120437c1a0a2610394f55fa2" + }, + { + "dataPath": "params_shard_238.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.61.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "ee1d7c992b179dcb06a017899e89410f" + }, + { + "dataPath": "params_shard_239.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.62.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "8fa52fc95d4c6e0768a6e977e84daba0" + }, + { + "dataPath": "params_shard_240.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.62.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "fdb4e5e56356bfd208ec1803d2a95041" + }, + { + "dataPath": "params_shard_241.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.62.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "767cd360012e74c64e1e8c69f0f9c9b8" + }, + { + "dataPath": "params_shard_242.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.62.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "e0b18e4485fb443137bf6bdea536e32f" + }, + { + "dataPath": "params_shard_243.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.63.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "22831efe05d52f802fec2b2b6a763894" + }, + { + "dataPath": "params_shard_244.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.63.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "29e9ecdee83fc0d9146c85860e79be15" + }, + { + "dataPath": "params_shard_245.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.63.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "03a28d88e967b1c5b24da5535161212d" + }, + { + "dataPath": "params_shard_246.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.63.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "4c4e4a648fa391a984ccb13da30db21a" + }, + { + "dataPath": "params_shard_247.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.64.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "6ab53c91e25e2a6b037a19398b16cadc" + }, + { + "dataPath": "params_shard_248.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.64.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "daa553383212ba2648755d7ca97c4d3f" + }, + { + "dataPath": "params_shard_249.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.64.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "5409cf7efb4afd58aaaba04d774359f5" + }, + { + "dataPath": "params_shard_250.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.64.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "9b525878fd653748c2cb874c5ff9f38d" + }, + { + "dataPath": "params_shard_251.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.65.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "ab5ea1b8b29cac9c05cde25c13c492a9" + }, + { + "dataPath": "params_shard_252.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.65.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "2c43d9bb82aad7b7ceacc3d8146e4e31" + }, + { + "dataPath": "params_shard_253.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.65.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "8127b864a905eccf0b26bd138e2a61d5" + }, + { + "dataPath": "params_shard_254.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.65.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "e55e23016bbb73a4f8f729c4f468a718" + }, + { + "dataPath": "params_shard_255.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.66.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "8527d263da170977035324286b1fa62c" + }, + { + "dataPath": "params_shard_256.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.66.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "a50f9422dd4e822543bfaebf15705fa0" + }, + { + "dataPath": "params_shard_257.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.66.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "621b380c2f32b0237b810e70b65d6dff" + }, + { + "dataPath": "params_shard_258.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.66.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "d361f3ebdfec0883d91d1a2968e34f3f" + }, + { + "dataPath": "params_shard_259.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.67.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "8435c418221c476904fe163e504da299" + }, + { + "dataPath": "params_shard_260.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.67.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "ef98eb3490660e41e98d284ab102fa23" + }, + { + "dataPath": "params_shard_261.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.67.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "a3f17268626d1a2c8a92f529c99ec68e" + }, + { + "dataPath": "params_shard_262.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.67.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "9446fbe06a354c96dc26e4d8694603f5" + }, + { + "dataPath": "params_shard_263.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.68.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "f160899c73a2b7eb19234311bc5a0cab" + }, + { + "dataPath": "params_shard_264.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.68.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "8c12af27b49f7eece96d12982995f0bd" + }, + { + "dataPath": "params_shard_265.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.68.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "7c702a9e8dd72734ba07a0166d363e51" + }, + { + "dataPath": "params_shard_266.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.68.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "c9e726a1d0b27561b762735568d8c1e1" + }, + { + "dataPath": "params_shard_267.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.69.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "50f9ef198c28e9eb6d0c272610938e16" + }, + { + "dataPath": "params_shard_268.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.69.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "9f6bd10bc82295ea9766f302b2b1c4b8" + }, + { + "dataPath": "params_shard_269.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.69.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "f2dcf8116e9e69e24d0e509a934351f8" + }, + { + "dataPath": "params_shard_270.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.69.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "deaed9960ddfa6ed4b140b773e62445a" + }, + { + "dataPath": "params_shard_271.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.70.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "94c0f3051e47f0d76c18cbc6ae0434ee" + }, + { + "dataPath": "params_shard_272.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.70.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "43a627c7ac175a04219e913ffacbd5b3" + }, + { + "dataPath": "params_shard_273.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.70.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "15a7a4a4d52d2ea5784c973e5151c48e" + }, + { + "dataPath": "params_shard_274.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.70.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "2147fdc9c5929e40efa304358b673e24" + }, + { + "dataPath": "params_shard_275.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.71.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "a9a6459b5e520c55f239fc94d058382d" + }, + { + "dataPath": "params_shard_276.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.71.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "6567fee89338570d687615f0c8452dbe" + }, + { + "dataPath": "params_shard_277.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.71.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "5e323cfb822efe63ac8822524581fa69" + }, + { + "dataPath": "params_shard_278.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "96c2fa3b1b3efdfa70df57f0b0676acf" + }, + { + "dataPath": "params_shard_279.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "8e777f93fbb9fcca78d94296d9206654" + }, + { + "dataPath": "params_shard_280.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "b51903bf228bbae2fa097ebbb7083f59" + }, + { + "dataPath": "params_shard_281.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "d405efb5211fa003422cc46bcefe540c" + }, + { + "dataPath": "params_shard_282.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "36ba0e3289795904659ba6aad54678c9" + }, + { + "dataPath": "params_shard_283.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.8.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "b83682cc2651fcc57de113bb88a48d5f" + }, + { + "dataPath": "params_shard_284.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "44eb062ed650153bae9febce3f4993f2" + }, + { + "dataPath": "params_shard_285.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "a1ccb8bb2f58ab6ba4db2b1dd0a58871" + }, + { + "dataPath": "params_shard_286.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "127e2582e21da84059ebc2c909bb8d1d" + }, + { + "dataPath": "params_shard_287.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.9.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "e7bbedb4258c1745340f2a86f98db514" + }, + { + "dataPath": "params_shard_288.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "eca3173dd5fda2e87a524a20b6fc414d" + }, + { + "dataPath": "params_shard_289.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.71.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "6205f103eee621b3d09d706b262de16a" + }, + { + "dataPath": "params_shard_290.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.72.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "bcb878d0513ae913c0095a2a285ccc31" + }, + { + "dataPath": "params_shard_291.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.72.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "5f2827e0688fcbff8012d2947af7d690" + }, + { + "dataPath": "params_shard_292.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.72.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "21e6a4c9af805fa55c2aab3fda61310f" + }, + { + "dataPath": "params_shard_293.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.72.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "e79d263a5560df151d9ae5bf2bb8fbbb" + }, + { + "dataPath": "params_shard_294.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.73.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "bb53d44f5b932081354614705de77822" + }, + { + "dataPath": "params_shard_295.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.73.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "220e0d9c014072514b076eed771443d9" + }, + { + "dataPath": "params_shard_296.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.73.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "bbeef1d9424bcd46f5ceaa499dc492f2" + }, + { + "dataPath": "params_shard_297.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.73.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "7a692adb9c99e68ee4f05686b50ea1c7" + }, + { + "dataPath": "params_shard_298.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.74.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "c2d2110a27d43bf441fa18569d2019e8" + }, + { + "dataPath": "params_shard_299.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.74.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "172a5857f83843f277470fc5338cbf25" + }, + { + "dataPath": "params_shard_300.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.74.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "505cc786c17f094efeb6528c8a468bb3" + }, + { + "dataPath": "params_shard_301.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.74.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "d8d762a07a9cc608d08d9a5be00779d6" + }, + { + "dataPath": "params_shard_302.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.75.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "c62915c887ef5ebd065866e65e640578" + }, + { + "dataPath": "params_shard_303.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.75.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "3834e98888e3b8d3115cdd916fe225b0" + }, + { + "dataPath": "params_shard_304.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.75.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "a2536778e7d8162770e0a49bed2b5ea9" + }, + { + "dataPath": "params_shard_305.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.75.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "9895c390535a00ee319cd6cbcf78202b" + }, + { + "dataPath": "params_shard_306.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.76.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "e04d05daa4a1d7dff9c72505ec5e5033" + }, + { + "dataPath": "params_shard_307.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.76.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "413bc17ca708b1c200c0c2f9957fca09" + }, + { + "dataPath": "params_shard_308.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.76.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "86160550862efa5059f1b7eeff3380f0" + }, + { + "dataPath": "params_shard_309.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.76.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "c2d4633ef6cb4fa0a129d4cd00c0156f" + }, + { + "dataPath": "params_shard_310.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.77.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "e0f0859077802767614438ace4923b82" + }, + { + "dataPath": "params_shard_311.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.77.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "870e199308d70e4a927e264fa4a45aed" + }, + { + "dataPath": "params_shard_312.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.77.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "69811870795b8d03e7030bf96b8c3e73" + }, + { + "dataPath": "params_shard_313.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.77.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "9a11f655d8421fa7bf58ef7129ee4f2f" + }, + { + "dataPath": "params_shard_314.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.78.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "9665b6140d2eebf4486f8dbdd8cc21e1" + }, + { + "dataPath": "params_shard_315.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.78.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "dbe92c56650a012ba709e44a6aed1b54" + }, + { + "dataPath": "params_shard_316.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.78.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "4b867ad770b07e0b9a66f0ac719cc900" + }, + { + "dataPath": "params_shard_317.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.78.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "ccd45c12cd6c586db88c021d3913ee3e" + }, + { + "dataPath": "params_shard_318.bin", + "format": "raw-shard", + "nbytes": 234881024, + "records": [ + { + "name": "model.layers.79.mlp.down_proj.q_weight", + "shape": [ + 8192, + 28672 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 234881024, + "byteOffset": 0 + } + ], + "md5sum": "6b335412c60f845071bfad95a2865f4d" + }, + { + "dataPath": "params_shard_319.bin", + "format": "raw-shard", + "nbytes": 469762048, + "records": [ + { + "name": "model.layers.79.mlp.gate_up_proj.q_weight", + "shape": [ + 57344, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 469762048, + "byteOffset": 0 + } + ], + "md5sum": "9eeffae3c6160a0bb9c16d82740505d4" + }, + { + "dataPath": "params_shard_320.bin", + "format": "raw-shard", + "nbytes": 83886080, + "records": [ + { + "name": "model.layers.79.self_attn.qkv_proj.q_weight", + "shape": [ + 10240, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 83886080, + "byteOffset": 0 + } + ], + "md5sum": "69a6463f2da2b1fb99a5b6d88031ed7f" + }, + { + "dataPath": "params_shard_321.bin", + "format": "raw-shard", + "nbytes": 67108864, + "records": [ + { + "name": "model.layers.79.self_attn.o_proj.q_weight", + "shape": [ + 8192, + 8192 + ], + "dtype": "e4m3_float8", + "format": "f32-to-bf16", + "nbytes": 67108864, + "byteOffset": 0 + } + ], + "md5sum": "6f0968bf476cb3778e8ff35a0f4ae217" + }, + { + "dataPath": "params_shard_322.bin", + "format": "raw-shard", + "nbytes": 2638464, + "records": [ + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 0 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 16384 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 16386 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 16388 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 32772 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 32774 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 32776 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 32778 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 32780 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 32782 + }, + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 49166 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 49168 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 65552 + }, + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 81936 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 81938 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 81940 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 98324 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 98326 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 98328 + }, + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 114712 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 114714 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 114716 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 131100 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 131102 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 131104 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 131106 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 131108 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 147492 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 147494 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 147496 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 163880 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 163882 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 163884 + }, + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 180268 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 180270 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 180272 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 196656 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 196658 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 196660 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 196662 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 196664 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 196666 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 213050 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 213052 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 229436 + }, + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 245820 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 245822 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 245824 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 262208 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 262210 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 262212 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 278596 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 278598 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 278600 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 294984 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 294986 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 294988 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 294990 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 294992 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 294994 + }, + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 311378 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 311380 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 327764 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 344148 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 344150 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 344152 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 360536 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 360538 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 360540 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 376924 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 376926 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 376928 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 393312 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 393314 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 393316 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 393318 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 393320 + }, + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 409704 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 409706 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 409708 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 426092 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 442476 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 442478 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 442480 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 458864 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 458866 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 458868 + }, + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 475252 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 475254 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 475256 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 491640 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 491642 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 491644 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 491646 + }, + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 508030 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 508032 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 508034 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 524418 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 524420 + }, + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 540804 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 540806 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 540808 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 557192 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 557194 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 557196 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 573580 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 573582 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 573584 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 589968 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 589970 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 589972 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 606356 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 606358 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 606360 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 622744 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 622746 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 622748 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 639132 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 639134 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 639136 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 655520 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 655522 + }, + { + "name": "model.layers.26.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 655524 + }, + { + "name": "model.layers.26.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 655526 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 655528 + }, + { + "name": "model.layers.26.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 655530 + }, + { + "name": "model.layers.26.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 671914 + }, + { + "name": "model.layers.26.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 671916 + }, + { + "name": "model.layers.27.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 688300 + }, + { + "name": "model.layers.27.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 704684 + }, + { + "name": "model.layers.27.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 704686 + }, + { + "name": "model.layers.27.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 704688 + }, + { + "name": "model.layers.27.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 721072 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 721074 + }, + { + "name": "model.layers.28.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 721076 + }, + { + "name": "model.layers.28.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 737460 + }, + { + "name": "model.layers.28.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 737462 + }, + { + "name": "model.layers.28.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 737464 + }, + { + "name": "model.layers.28.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 753848 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 753850 + }, + { + "name": "model.layers.29.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 753852 + }, + { + "name": "model.layers.29.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 753854 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 753856 + }, + { + "name": "model.layers.29.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 753858 + }, + { + "name": "model.layers.29.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 770242 + }, + { + "name": "model.layers.29.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 770244 + }, + { + "name": "model.layers.30.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 786628 + }, + { + "name": "model.layers.30.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 803012 + }, + { + "name": "model.layers.30.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 803014 + }, + { + "name": "model.layers.30.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 803016 + }, + { + "name": "model.layers.30.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 819400 + }, + { + "name": "model.layers.30.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 819402 + }, + { + "name": "model.layers.31.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 819404 + }, + { + "name": "model.layers.31.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 835788 + }, + { + "name": "model.layers.31.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 835790 + }, + { + "name": "model.layers.31.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 835792 + }, + { + "name": "model.layers.31.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 852176 + }, + { + "name": "model.layers.31.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 852178 + }, + { + "name": "model.layers.32.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 852180 + }, + { + "name": "model.layers.32.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 852182 + }, + { + "name": "model.layers.32.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 852184 + }, + { + "name": "model.layers.32.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 868568 + }, + { + "name": "model.layers.32.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 868570 + }, + { + "name": "model.layers.32.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 868572 + }, + { + "name": "model.layers.33.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 884956 + }, + { + "name": "model.layers.33.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 901340 + }, + { + "name": "model.layers.33.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 901342 + }, + { + "name": "model.layers.33.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 901344 + }, + { + "name": "model.layers.33.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 917728 + }, + { + "name": "model.layers.33.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 917730 + }, + { + "name": "model.layers.34.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 917732 + }, + { + "name": "model.layers.34.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 934116 + }, + { + "name": "model.layers.34.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 934118 + }, + { + "name": "model.layers.34.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 934120 + }, + { + "name": "model.layers.34.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 950504 + }, + { + "name": "model.layers.34.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 950506 + }, + { + "name": "model.layers.35.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 950508 + }, + { + "name": "model.layers.35.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 950510 + }, + { + "name": "model.layers.35.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 966894 + }, + { + "name": "model.layers.35.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 966896 + }, + { + "name": "model.layers.35.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 966898 + }, + { + "name": "model.layers.35.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 983282 + }, + { + "name": "model.layers.36.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 983284 + }, + { + "name": "model.layers.36.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 999668 + }, + { + "name": "model.layers.36.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 999670 + }, + { + "name": "model.layers.36.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 999672 + }, + { + "name": "model.layers.36.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1016056 + }, + { + "name": "model.layers.36.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1016058 + }, + { + "name": "model.layers.37.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1016060 + }, + { + "name": "model.layers.37.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1032444 + }, + { + "name": "model.layers.37.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1032446 + }, + { + "name": "model.layers.37.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1032448 + }, + { + "name": "model.layers.37.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1048832 + }, + { + "name": "model.layers.37.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1048834 + }, + { + "name": "model.layers.38.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1048836 + }, + { + "name": "model.layers.38.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1065220 + }, + { + "name": "model.layers.38.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1065222 + }, + { + "name": "model.layers.38.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1065224 + }, + { + "name": "model.layers.38.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1081608 + }, + { + "name": "model.layers.38.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1081610 + }, + { + "name": "model.layers.39.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1081612 + }, + { + "name": "model.layers.39.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1097996 + }, + { + "name": "model.layers.39.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1097998 + }, + { + "name": "model.layers.39.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1098000 + }, + { + "name": "model.layers.39.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1114384 + }, + { + "name": "model.layers.39.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1114386 + }, + { + "name": "model.layers.40.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1114388 + }, + { + "name": "model.layers.40.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1114390 + }, + { + "name": "model.layers.40.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1114392 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1114394 + }, + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1130778 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1130780 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1130782 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1147166 + }, + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1163550 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1163552 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1163554 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1179938 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1179940 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1179942 + }, + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1196326 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1196328 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1196330 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1212714 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1212716 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1212718 + }, + { + "name": "model.layers.40.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1212720 + }, + { + "name": "model.layers.40.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1229104 + }, + { + "name": "model.layers.40.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1229106 + }, + { + "name": "model.layers.41.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1245490 + }, + { + "name": "model.layers.41.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1261874 + }, + { + "name": "model.layers.41.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1261876 + }, + { + "name": "model.layers.41.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1261878 + }, + { + "name": "model.layers.41.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1278262 + }, + { + "name": "model.layers.41.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1278264 + }, + { + "name": "model.layers.42.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1278266 + }, + { + "name": "model.layers.42.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1294650 + }, + { + "name": "model.layers.42.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1294652 + }, + { + "name": "model.layers.42.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1294654 + }, + { + "name": "model.layers.42.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1311038 + }, + { + "name": "model.layers.42.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1311040 + }, + { + "name": "model.layers.43.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1311042 + }, + { + "name": "model.layers.43.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1311044 + }, + { + "name": "model.layers.43.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1311046 + }, + { + "name": "model.layers.43.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1311048 + }, + { + "name": "model.layers.43.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1327432 + }, + { + "name": "model.layers.43.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1327434 + }, + { + "name": "model.layers.44.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1343818 + }, + { + "name": "model.layers.44.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1360202 + }, + { + "name": "model.layers.44.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1360204 + }, + { + "name": "model.layers.44.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1360206 + }, + { + "name": "model.layers.44.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1376590 + }, + { + "name": "model.layers.44.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1376592 + }, + { + "name": "model.layers.45.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1376594 + }, + { + "name": "model.layers.45.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1392978 + }, + { + "name": "model.layers.45.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1392980 + }, + { + "name": "model.layers.45.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1392982 + }, + { + "name": "model.layers.45.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1409366 + }, + { + "name": "model.layers.45.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1409368 + }, + { + "name": "model.layers.46.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1409370 + }, + { + "name": "model.layers.46.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1409372 + }, + { + "name": "model.layers.46.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1409374 + }, + { + "name": "model.layers.46.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1425758 + }, + { + "name": "model.layers.46.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1425760 + }, + { + "name": "model.layers.46.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1425762 + }, + { + "name": "model.layers.47.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1442146 + }, + { + "name": "model.layers.47.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1458530 + }, + { + "name": "model.layers.47.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1458532 + }, + { + "name": "model.layers.47.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1458534 + }, + { + "name": "model.layers.47.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1474918 + }, + { + "name": "model.layers.47.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1474920 + }, + { + "name": "model.layers.48.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1474922 + }, + { + "name": "model.layers.48.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1491306 + }, + { + "name": "model.layers.48.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1491308 + }, + { + "name": "model.layers.48.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1491310 + }, + { + "name": "model.layers.48.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1507694 + }, + { + "name": "model.layers.48.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1507696 + }, + { + "name": "model.layers.49.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1507698 + }, + { + "name": "model.layers.49.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1507700 + }, + { + "name": "model.layers.49.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1524084 + }, + { + "name": "model.layers.49.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1524086 + }, + { + "name": "model.layers.49.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1524088 + }, + { + "name": "model.layers.49.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1540472 + }, + { + "name": "model.layers.50.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1540474 + }, + { + "name": "model.layers.50.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1556858 + }, + { + "name": "model.layers.50.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1556860 + }, + { + "name": "model.layers.50.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1556862 + }, + { + "name": "model.layers.50.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1573246 + }, + { + "name": "model.layers.50.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1573248 + }, + { + "name": "model.layers.51.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1573250 + }, + { + "name": "model.layers.51.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1589634 + }, + { + "name": "model.layers.51.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1589636 + }, + { + "name": "model.layers.51.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1589638 + }, + { + "name": "model.layers.51.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1606022 + }, + { + "name": "model.layers.51.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1606024 + }, + { + "name": "model.layers.52.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1606026 + }, + { + "name": "model.layers.52.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1622410 + }, + { + "name": "model.layers.52.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1622412 + }, + { + "name": "model.layers.52.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1622414 + }, + { + "name": "model.layers.52.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1638798 + }, + { + "name": "model.layers.52.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1638800 + }, + { + "name": "model.layers.53.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1638802 + }, + { + "name": "model.layers.53.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1655186 + }, + { + "name": "model.layers.53.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1655188 + }, + { + "name": "model.layers.53.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1655190 + }, + { + "name": "model.layers.53.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1671574 + }, + { + "name": "model.layers.53.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1671576 + }, + { + "name": "model.layers.54.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1671578 + }, + { + "name": "model.layers.54.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1671580 + }, + { + "name": "model.layers.54.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1671582 + }, + { + "name": "model.layers.54.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1671584 + }, + { + "name": "model.layers.54.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1687968 + }, + { + "name": "model.layers.54.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1687970 + }, + { + "name": "model.layers.55.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1704354 + }, + { + "name": "model.layers.55.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1720738 + }, + { + "name": "model.layers.55.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1720740 + }, + { + "name": "model.layers.55.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1720742 + }, + { + "name": "model.layers.55.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1737126 + }, + { + "name": "model.layers.55.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1737128 + }, + { + "name": "model.layers.56.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1737130 + }, + { + "name": "model.layers.56.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1753514 + }, + { + "name": "model.layers.56.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1753516 + }, + { + "name": "model.layers.56.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1753518 + }, + { + "name": "model.layers.56.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1769902 + }, + { + "name": "model.layers.56.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1769904 + }, + { + "name": "model.layers.57.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1769906 + }, + { + "name": "model.layers.57.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1769908 + }, + { + "name": "model.layers.57.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1769910 + }, + { + "name": "model.layers.57.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1769912 + }, + { + "name": "model.layers.57.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1786296 + }, + { + "name": "model.layers.57.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1786298 + }, + { + "name": "model.layers.58.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1802682 + }, + { + "name": "model.layers.58.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1819066 + }, + { + "name": "model.layers.58.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1819068 + }, + { + "name": "model.layers.58.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1819070 + }, + { + "name": "model.layers.58.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1835454 + }, + { + "name": "model.layers.58.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1835456 + }, + { + "name": "model.layers.59.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1835458 + }, + { + "name": "model.layers.59.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1851842 + }, + { + "name": "model.layers.59.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1851844 + }, + { + "name": "model.layers.59.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1851846 + }, + { + "name": "model.layers.59.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1868230 + }, + { + "name": "model.layers.59.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1868232 + }, + { + "name": "model.layers.60.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1868234 + }, + { + "name": "model.layers.60.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1868236 + }, + { + "name": "model.layers.60.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1868238 + }, + { + "name": "model.layers.60.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1884622 + }, + { + "name": "model.layers.60.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1884624 + }, + { + "name": "model.layers.60.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1884626 + }, + { + "name": "model.layers.61.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1901010 + }, + { + "name": "model.layers.61.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1917394 + }, + { + "name": "model.layers.61.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1917396 + }, + { + "name": "model.layers.61.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1917398 + }, + { + "name": "model.layers.61.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1933782 + }, + { + "name": "model.layers.61.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1933784 + }, + { + "name": "model.layers.62.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1933786 + }, + { + "name": "model.layers.62.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1950170 + }, + { + "name": "model.layers.62.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1950172 + }, + { + "name": "model.layers.62.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1950174 + }, + { + "name": "model.layers.62.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1966558 + }, + { + "name": "model.layers.62.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1966560 + }, + { + "name": "model.layers.63.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1966562 + }, + { + "name": "model.layers.63.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1966564 + }, + { + "name": "model.layers.63.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1982948 + }, + { + "name": "model.layers.63.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1982950 + }, + { + "name": "model.layers.63.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1982952 + }, + { + "name": "model.layers.63.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 1999336 + }, + { + "name": "model.layers.64.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 1999338 + }, + { + "name": "model.layers.64.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2015722 + }, + { + "name": "model.layers.64.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2015724 + }, + { + "name": "model.layers.64.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2015726 + }, + { + "name": "model.layers.64.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2032110 + }, + { + "name": "model.layers.64.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2032112 + }, + { + "name": "model.layers.65.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2032114 + }, + { + "name": "model.layers.65.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2048498 + }, + { + "name": "model.layers.65.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2048500 + }, + { + "name": "model.layers.65.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2048502 + }, + { + "name": "model.layers.65.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2064886 + }, + { + "name": "model.layers.65.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2064888 + }, + { + "name": "model.layers.66.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2064890 + }, + { + "name": "model.layers.66.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2081274 + }, + { + "name": "model.layers.66.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2081276 + }, + { + "name": "model.layers.66.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2081278 + }, + { + "name": "model.layers.66.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2097662 + }, + { + "name": "model.layers.66.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2097664 + }, + { + "name": "model.layers.67.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2097666 + }, + { + "name": "model.layers.67.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2114050 + }, + { + "name": "model.layers.67.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2114052 + }, + { + "name": "model.layers.67.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2114054 + }, + { + "name": "model.layers.67.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2130438 + }, + { + "name": "model.layers.67.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2130440 + }, + { + "name": "model.layers.68.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2130442 + }, + { + "name": "model.layers.68.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2130444 + }, + { + "name": "model.layers.68.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2130446 + }, + { + "name": "model.layers.68.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2130448 + }, + { + "name": "model.layers.68.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2146832 + }, + { + "name": "model.layers.68.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2146834 + }, + { + "name": "model.layers.69.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2163218 + }, + { + "name": "model.layers.69.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2179602 + }, + { + "name": "model.layers.69.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2179604 + }, + { + "name": "model.layers.69.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2179606 + }, + { + "name": "model.layers.69.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2195990 + }, + { + "name": "model.layers.69.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2195992 + }, + { + "name": "model.layers.70.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2195994 + }, + { + "name": "model.layers.70.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2212378 + }, + { + "name": "model.layers.70.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2212380 + }, + { + "name": "model.layers.70.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2212382 + }, + { + "name": "model.layers.70.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2228766 + }, + { + "name": "model.layers.70.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2228768 + }, + { + "name": "model.layers.71.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2228770 + }, + { + "name": "model.layers.71.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2228772 + }, + { + "name": "model.layers.71.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2228774 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2228776 + }, + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2245160 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2245162 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2245164 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2261548 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2261550 + }, + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2277934 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2277936 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2277938 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2294322 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2294324 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2294326 + }, + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2310710 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2310712 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2310714 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2327098 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2327100 + }, + { + "name": "model.layers.71.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2327102 + }, + { + "name": "model.layers.71.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2343486 + }, + { + "name": "model.layers.71.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2343488 + }, + { + "name": "model.layers.72.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2359872 + }, + { + "name": "model.layers.72.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2376256 + }, + { + "name": "model.layers.72.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2376258 + }, + { + "name": "model.layers.72.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2376260 + }, + { + "name": "model.layers.72.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2392644 + }, + { + "name": "model.layers.72.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2392646 + }, + { + "name": "model.layers.73.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2392648 + }, + { + "name": "model.layers.73.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2409032 + }, + { + "name": "model.layers.73.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2409034 + }, + { + "name": "model.layers.73.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2409036 + }, + { + "name": "model.layers.73.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2425420 + }, + { + "name": "model.layers.73.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2425422 + }, + { + "name": "model.layers.74.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2425424 + }, + { + "name": "model.layers.74.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2425426 + }, + { + "name": "model.layers.74.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2425428 + }, + { + "name": "model.layers.74.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2441812 + }, + { + "name": "model.layers.74.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2441814 + }, + { + "name": "model.layers.74.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2441816 + }, + { + "name": "model.layers.75.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2458200 + }, + { + "name": "model.layers.75.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2474584 + }, + { + "name": "model.layers.75.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2474586 + }, + { + "name": "model.layers.75.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2474588 + }, + { + "name": "model.layers.75.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2490972 + }, + { + "name": "model.layers.75.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2490974 + }, + { + "name": "model.layers.76.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2490976 + }, + { + "name": "model.layers.76.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2507360 + }, + { + "name": "model.layers.76.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2507362 + }, + { + "name": "model.layers.76.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2507364 + }, + { + "name": "model.layers.76.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2523748 + }, + { + "name": "model.layers.76.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2523750 + }, + { + "name": "model.layers.77.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2523752 + }, + { + "name": "model.layers.77.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2523754 + }, + { + "name": "model.layers.77.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2540138 + }, + { + "name": "model.layers.77.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2540140 + }, + { + "name": "model.layers.77.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2540142 + }, + { + "name": "model.layers.77.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2556526 + }, + { + "name": "model.layers.78.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2556528 + }, + { + "name": "model.layers.78.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2572912 + }, + { + "name": "model.layers.78.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2572914 + }, + { + "name": "model.layers.78.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2572916 + }, + { + "name": "model.layers.78.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2589300 + }, + { + "name": "model.layers.78.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2589302 + }, + { + "name": "model.layers.79.input_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2589304 + }, + { + "name": "model.layers.79.mlp.down_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2605688 + }, + { + "name": "model.layers.79.mlp.gate_up_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2605690 + }, + { + "name": "model.layers.79.post_attention_layernorm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2605692 + }, + { + "name": "model.layers.79.self_attn.qkv_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2622076 + }, + { + "name": "model.layers.79.self_attn.o_proj.q_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2622078 + }, + { + "name": "model.norm.weight", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 16384, + "byteOffset": 2622080 + } + ], + "md5sum": "bc20ac32db07b3672c74b423b8bf7df4" + }, + { + "dataPath": "params_shard_323.bin", + "format": "raw-shard", + "nbytes": 640, + "records": [ + { + "name": "model.layers.0.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 0 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 2 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 4 + }, + { + "name": "model.layers.0.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 6 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 8 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 10 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 12 + }, + { + "name": "model.layers.1.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 14 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 16 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 18 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 20 + }, + { + "name": "model.layers.2.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 22 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 24 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 26 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 28 + }, + { + "name": "model.layers.3.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 30 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 32 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 34 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 36 + }, + { + "name": "model.layers.4.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 38 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 40 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 42 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 44 + }, + { + "name": "model.layers.5.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 46 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 48 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 50 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 52 + }, + { + "name": "model.layers.6.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 54 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 56 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 58 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 60 + }, + { + "name": "model.layers.7.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 62 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 64 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 66 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 68 + }, + { + "name": "model.layers.8.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 70 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 72 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 74 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 76 + }, + { + "name": "model.layers.9.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 78 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 80 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 82 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 84 + }, + { + "name": "model.layers.10.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 86 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 88 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 90 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 92 + }, + { + "name": "model.layers.11.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 94 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 96 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 98 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 100 + }, + { + "name": "model.layers.12.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 102 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 104 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 106 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 108 + }, + { + "name": "model.layers.13.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 110 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 112 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 114 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 116 + }, + { + "name": "model.layers.14.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 118 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 120 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 122 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 124 + }, + { + "name": "model.layers.15.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 126 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 128 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 130 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 132 + }, + { + "name": "model.layers.16.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 134 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 136 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 138 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 140 + }, + { + "name": "model.layers.17.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 142 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 144 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 146 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 148 + }, + { + "name": "model.layers.18.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 150 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 152 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 154 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 156 + }, + { + "name": "model.layers.19.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 158 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 160 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 162 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 164 + }, + { + "name": "model.layers.20.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 166 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 168 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 170 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 172 + }, + { + "name": "model.layers.21.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 174 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 176 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 178 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 180 + }, + { + "name": "model.layers.22.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 182 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 184 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 186 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 188 + }, + { + "name": "model.layers.23.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 190 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 192 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 194 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 196 + }, + { + "name": "model.layers.24.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 198 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 200 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 202 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 204 + }, + { + "name": "model.layers.25.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 206 + }, + { + "name": "model.layers.26.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 208 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 210 + }, + { + "name": "model.layers.26.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 212 + }, + { + "name": "model.layers.26.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 214 + }, + { + "name": "model.layers.27.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 216 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 218 + }, + { + "name": "model.layers.27.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 220 + }, + { + "name": "model.layers.27.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 222 + }, + { + "name": "model.layers.28.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 224 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 226 + }, + { + "name": "model.layers.28.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 228 + }, + { + "name": "model.layers.28.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 230 + }, + { + "name": "model.layers.29.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 232 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 234 + }, + { + "name": "model.layers.29.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 236 + }, + { + "name": "model.layers.29.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 238 + }, + { + "name": "model.layers.30.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 240 + }, + { + "name": "model.layers.30.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 242 + }, + { + "name": "model.layers.30.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 244 + }, + { + "name": "model.layers.30.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 246 + }, + { + "name": "model.layers.31.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 248 + }, + { + "name": "model.layers.31.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 250 + }, + { + "name": "model.layers.31.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 252 + }, + { + "name": "model.layers.31.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 254 + }, + { + "name": "model.layers.32.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 256 + }, + { + "name": "model.layers.32.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 258 + }, + { + "name": "model.layers.32.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 260 + }, + { + "name": "model.layers.32.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 262 + }, + { + "name": "model.layers.33.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 264 + }, + { + "name": "model.layers.33.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 266 + }, + { + "name": "model.layers.33.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 268 + }, + { + "name": "model.layers.33.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 270 + }, + { + "name": "model.layers.34.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 272 + }, + { + "name": "model.layers.34.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 274 + }, + { + "name": "model.layers.34.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 276 + }, + { + "name": "model.layers.34.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 278 + }, + { + "name": "model.layers.35.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 280 + }, + { + "name": "model.layers.35.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 282 + }, + { + "name": "model.layers.35.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 284 + }, + { + "name": "model.layers.35.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 286 + }, + { + "name": "model.layers.36.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 288 + }, + { + "name": "model.layers.36.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 290 + }, + { + "name": "model.layers.36.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 292 + }, + { + "name": "model.layers.36.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 294 + }, + { + "name": "model.layers.37.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 296 + }, + { + "name": "model.layers.37.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 298 + }, + { + "name": "model.layers.37.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 300 + }, + { + "name": "model.layers.37.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 302 + }, + { + "name": "model.layers.38.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 304 + }, + { + "name": "model.layers.38.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 306 + }, + { + "name": "model.layers.38.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 308 + }, + { + "name": "model.layers.38.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 310 + }, + { + "name": "model.layers.39.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 312 + }, + { + "name": "model.layers.39.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 314 + }, + { + "name": "model.layers.39.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 316 + }, + { + "name": "model.layers.39.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 318 + }, + { + "name": "model.layers.40.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 320 + }, + { + "name": "model.layers.40.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 322 + }, + { + "name": "model.layers.40.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 324 + }, + { + "name": "model.layers.40.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 326 + }, + { + "name": "model.layers.41.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 328 + }, + { + "name": "model.layers.41.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 330 + }, + { + "name": "model.layers.41.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 332 + }, + { + "name": "model.layers.41.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 334 + }, + { + "name": "model.layers.42.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 336 + }, + { + "name": "model.layers.42.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 338 + }, + { + "name": "model.layers.42.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 340 + }, + { + "name": "model.layers.42.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 342 + }, + { + "name": "model.layers.43.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 344 + }, + { + "name": "model.layers.43.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 346 + }, + { + "name": "model.layers.43.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 348 + }, + { + "name": "model.layers.43.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 350 + }, + { + "name": "model.layers.44.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 352 + }, + { + "name": "model.layers.44.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 354 + }, + { + "name": "model.layers.44.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 356 + }, + { + "name": "model.layers.44.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 358 + }, + { + "name": "model.layers.45.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 360 + }, + { + "name": "model.layers.45.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 362 + }, + { + "name": "model.layers.45.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 364 + }, + { + "name": "model.layers.45.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 366 + }, + { + "name": "model.layers.46.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 368 + }, + { + "name": "model.layers.46.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 370 + }, + { + "name": "model.layers.46.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 372 + }, + { + "name": "model.layers.46.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 374 + }, + { + "name": "model.layers.47.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 376 + }, + { + "name": "model.layers.47.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 378 + }, + { + "name": "model.layers.47.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 380 + }, + { + "name": "model.layers.47.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 382 + }, + { + "name": "model.layers.48.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 384 + }, + { + "name": "model.layers.48.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 386 + }, + { + "name": "model.layers.48.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 388 + }, + { + "name": "model.layers.48.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 390 + }, + { + "name": "model.layers.49.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 392 + }, + { + "name": "model.layers.49.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 394 + }, + { + "name": "model.layers.49.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 396 + }, + { + "name": "model.layers.49.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 398 + }, + { + "name": "model.layers.50.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 400 + }, + { + "name": "model.layers.50.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 402 + }, + { + "name": "model.layers.50.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 404 + }, + { + "name": "model.layers.50.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 406 + }, + { + "name": "model.layers.51.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 408 + }, + { + "name": "model.layers.51.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 410 + }, + { + "name": "model.layers.51.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 412 + }, + { + "name": "model.layers.51.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 414 + }, + { + "name": "model.layers.52.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 416 + }, + { + "name": "model.layers.52.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 418 + }, + { + "name": "model.layers.52.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 420 + }, + { + "name": "model.layers.52.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 422 + }, + { + "name": "model.layers.53.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 424 + }, + { + "name": "model.layers.53.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 426 + }, + { + "name": "model.layers.53.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 428 + }, + { + "name": "model.layers.53.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 430 + }, + { + "name": "model.layers.54.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 432 + }, + { + "name": "model.layers.54.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 434 + }, + { + "name": "model.layers.54.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 436 + }, + { + "name": "model.layers.54.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 438 + }, + { + "name": "model.layers.55.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 440 + }, + { + "name": "model.layers.55.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 442 + }, + { + "name": "model.layers.55.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 444 + }, + { + "name": "model.layers.55.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 446 + }, + { + "name": "model.layers.56.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 448 + }, + { + "name": "model.layers.56.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 450 + }, + { + "name": "model.layers.56.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 452 + }, + { + "name": "model.layers.56.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 454 + }, + { + "name": "model.layers.57.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 456 + }, + { + "name": "model.layers.57.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 458 + }, + { + "name": "model.layers.57.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 460 + }, + { + "name": "model.layers.57.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 462 + }, + { + "name": "model.layers.58.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 464 + }, + { + "name": "model.layers.58.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 466 + }, + { + "name": "model.layers.58.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 468 + }, + { + "name": "model.layers.58.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 470 + }, + { + "name": "model.layers.59.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 472 + }, + { + "name": "model.layers.59.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 474 + }, + { + "name": "model.layers.59.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 476 + }, + { + "name": "model.layers.59.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 478 + }, + { + "name": "model.layers.60.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 480 + }, + { + "name": "model.layers.60.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 482 + }, + { + "name": "model.layers.60.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 484 + }, + { + "name": "model.layers.60.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 486 + }, + { + "name": "model.layers.61.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 488 + }, + { + "name": "model.layers.61.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 490 + }, + { + "name": "model.layers.61.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 492 + }, + { + "name": "model.layers.61.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 494 + }, + { + "name": "model.layers.62.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 496 + }, + { + "name": "model.layers.62.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 498 + }, + { + "name": "model.layers.62.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 500 + }, + { + "name": "model.layers.62.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 502 + }, + { + "name": "model.layers.63.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 504 + }, + { + "name": "model.layers.63.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 506 + }, + { + "name": "model.layers.63.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 508 + }, + { + "name": "model.layers.63.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 510 + }, + { + "name": "model.layers.64.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 512 + }, + { + "name": "model.layers.64.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 514 + }, + { + "name": "model.layers.64.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 516 + }, + { + "name": "model.layers.64.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 518 + }, + { + "name": "model.layers.65.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 520 + }, + { + "name": "model.layers.65.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 522 + }, + { + "name": "model.layers.65.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 524 + }, + { + "name": "model.layers.65.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 526 + }, + { + "name": "model.layers.66.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 528 + }, + { + "name": "model.layers.66.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 530 + }, + { + "name": "model.layers.66.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 532 + }, + { + "name": "model.layers.66.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 534 + }, + { + "name": "model.layers.67.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 536 + }, + { + "name": "model.layers.67.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 538 + }, + { + "name": "model.layers.67.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 540 + }, + { + "name": "model.layers.67.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 542 + }, + { + "name": "model.layers.68.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 544 + }, + { + "name": "model.layers.68.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 546 + }, + { + "name": "model.layers.68.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 548 + }, + { + "name": "model.layers.68.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 550 + }, + { + "name": "model.layers.69.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 552 + }, + { + "name": "model.layers.69.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 554 + }, + { + "name": "model.layers.69.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 556 + }, + { + "name": "model.layers.69.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 558 + }, + { + "name": "model.layers.70.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 560 + }, + { + "name": "model.layers.70.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 562 + }, + { + "name": "model.layers.70.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 564 + }, + { + "name": "model.layers.70.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 566 + }, + { + "name": "model.layers.71.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 568 + }, + { + "name": "model.layers.71.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 570 + }, + { + "name": "model.layers.71.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 572 + }, + { + "name": "model.layers.71.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 574 + }, + { + "name": "model.layers.72.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 576 + }, + { + "name": "model.layers.72.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 578 + }, + { + "name": "model.layers.72.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 580 + }, + { + "name": "model.layers.72.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 582 + }, + { + "name": "model.layers.73.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 584 + }, + { + "name": "model.layers.73.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 586 + }, + { + "name": "model.layers.73.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 588 + }, + { + "name": "model.layers.73.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 590 + }, + { + "name": "model.layers.74.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 592 + }, + { + "name": "model.layers.74.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 594 + }, + { + "name": "model.layers.74.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 596 + }, + { + "name": "model.layers.74.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 598 + }, + { + "name": "model.layers.75.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 600 + }, + { + "name": "model.layers.75.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 602 + }, + { + "name": "model.layers.75.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 604 + }, + { + "name": "model.layers.75.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 606 + }, + { + "name": "model.layers.76.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 608 + }, + { + "name": "model.layers.76.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 610 + }, + { + "name": "model.layers.76.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 612 + }, + { + "name": "model.layers.76.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 614 + }, + { + "name": "model.layers.77.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 616 + }, + { + "name": "model.layers.77.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 618 + }, + { + "name": "model.layers.77.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 620 + }, + { + "name": "model.layers.77.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 622 + }, + { + "name": "model.layers.78.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 624 + }, + { + "name": "model.layers.78.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 626 + }, + { + "name": "model.layers.78.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 628 + }, + { + "name": "model.layers.78.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 630 + }, + { + "name": "model.layers.79.self_attn.qkv_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 632 + }, + { + "name": "model.layers.79.self_attn.o_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 634 + }, + { + "name": "model.layers.79.mlp.gate_up_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 636 + }, + { + "name": "model.layers.79.mlp.down_proj.q_calibration_scale", + "shape": [ + 1 + ], + "dtype": "bfloat16", + "format": "raw", + "nbytes": 2, + "byteOffset": 638 + } + ], + "md5sum": "34e32dfc0814e3592da944303c6bc891" + } + ] +} \ No newline at end of file