{ "metadata": { "ParamSize": 581, "ParamBytes": 5599709184.0, "BitsPerParam": 4.640753043187236 }, "records": [ { "dataPath": "params_shard_0.bin", "format": "raw-shard", "nbytes": 508559360, "records": [ { "name": "lm_head.q_weight", "shape": [ 248320, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 508559360, "byteOffset": 0 } ], "md5sum": "473d2e097ad5fe2818a479b6ea565b70" }, { "dataPath": "params_shard_1.bin", "format": "raw-shard", "nbytes": 63569920, "records": [ { "name": "lm_head.q_scale", "shape": [ 248320, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 63569920, "byteOffset": 0 } ], "md5sum": "399ac9a1d275f9ad217f288685e033d8" }, { "dataPath": "params_shard_2.bin", "format": "raw-shard", "nbytes": 508559360, "records": [ { "name": "model.embed_tokens.q_weight", "shape": [ 248320, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 508559360, "byteOffset": 0 } ], "md5sum": "5127ed1de2d4054e75b1b1b8a3ec6f1f" }, { "dataPath": "params_shard_3.bin", "format": "raw-shard", "nbytes": 63569920, "records": [ { "name": "model.embed_tokens.q_scale", "shape": [ 248320, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 63569920, "byteOffset": 0 } ], "md5sum": "41f4b6191bfd7405ac03b29603174095" }, { "dataPath": "params_shard_4.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.14.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "3b93c3dd02d3684fed60279f126c2bfe" }, { "dataPath": "params_shard_5.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.14.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 }, { "name": "model.layers.14.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 25165824 } ], "md5sum": "fab4a82deaa6f9ac8fca86aad9427868" }, { "dataPath": "params_shard_6.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.14.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 0 }, { "name": "model.layers.15.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 6291456 } ], "md5sum": "77b22fb154b8f96af4fabeaeaa1f38bf" }, { "dataPath": "params_shard_7.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.15.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "572fbf219c1c5faef52bc672f582566a" }, { "dataPath": "params_shard_8.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.25.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "a82e8f78499882558506a33e43f12b02" }, { "dataPath": "params_shard_9.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.26.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "937cc3ae528b7d94b7f929482eeaf59d" }, { "dataPath": "params_shard_10.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.26.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "183b61c7075fde51221d74fe6568d00b" }, { "dataPath": "params_shard_11.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.27.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "53c75bf97552e2106335f8d2a18925a5" }, { "dataPath": "params_shard_12.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.27.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "cd997ce0e438e04488d8770600e2bb9c" }, { "dataPath": "params_shard_13.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.3.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "c0f096d64bf8cd7ce1681cf91e361931" }, { "dataPath": "params_shard_14.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.15.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 0 }, { "name": "model.layers.15.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 3145728 }, { "name": "model.layers.25.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 9437184 }, { "name": "model.layers.26.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 12582912 }, { "name": "model.layers.26.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 15728640 }, { "name": "model.layers.27.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 22020096 }, { "name": "model.layers.27.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 25165824 } ], "md5sum": "5f4a0cc34df297102ec79cb48ddacbdd" }, { "dataPath": "params_shard_15.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.3.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "a9cc1b1f177fc4ad1a632b264a0bf601" }, { "dataPath": "params_shard_16.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.30.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "fb6cce5245a74b942cbcdb5ced24f8b3" }, { "dataPath": "params_shard_17.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.30.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "a88c37ed89d1877e55a8481e31eb3372" }, { "dataPath": "params_shard_18.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.31.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "778c806cdee057feabe69d0f56c9065b" }, { "dataPath": "params_shard_19.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.9.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "b681b348db7e5d1e5e6af7e211af9daa" }, { "dataPath": "params_shard_20.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.9.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "bab9342866e2177980876dd0943b20bc" }, { "dataPath": "params_shard_21.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.3.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 0 }, { "name": "model.layers.3.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 3145728 }, { "name": "model.layers.30.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 9437184 }, { "name": "model.layers.30.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 12582912 }, { "name": "model.layers.31.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 18874368 }, { "name": "model.layers.9.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 25165824 } ], "md5sum": "505a0cfee521bb5ce8a11285a3343ba8" }, { "dataPath": "params_shard_22.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.9.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 0 }, { "name": "model.layers.1.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 6291456 } ], "md5sum": "eaaf90a4b7ee747add7e7722b1927744" }, { "dataPath": "params_shard_23.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.1.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "f656b44b053321efe017ebc35a028225" }, { "dataPath": "params_shard_24.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.10.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "213a803b354be26c73ce470d20f0f276" }, { "dataPath": "params_shard_25.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.10.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "1604bcd76ff6cb062df4a4f31174fad8" }, { "dataPath": "params_shard_26.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.11.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "6c828c05a80352b96890b5bb48b57566" }, { "dataPath": "params_shard_27.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.11.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "8f2215b9fd17fac0ae46ffa9ef4af774" }, { "dataPath": "params_shard_28.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.0.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "0498f84998fafd2a18944c0ebbab1433" }, { "dataPath": "params_shard_29.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.0.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "2a520d9b76039824ecf2c891d1b726e8" }, { "dataPath": "params_shard_30.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.1.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 0 }, { "name": "model.layers.1.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 3145728 }, { "name": "model.layers.10.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 9437184 }, { "name": "model.layers.10.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 12582912 }, { "name": "model.layers.11.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 18874368 }, { "name": "model.layers.11.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 22020096 }, { "name": "model.layers.0.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 28311552 } ], "md5sum": "f06b2ab87eff109f6860bb4c40147c0d" }, { "dataPath": "params_shard_31.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.0.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 0 }, { "name": "model.layers.21.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 6291456 } ], "md5sum": "502da919a6f986628feddb943be0a95b" }, { "dataPath": "params_shard_32.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.21.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "b1ae81a13f8df49f412bbf69e606823f" }, { "dataPath": "params_shard_33.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.22.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "11a956ceb56173f89ae990d13b777049" }, { "dataPath": "params_shard_34.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.22.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "53569a976453aa1eece5c42d9b12f57b" }, { "dataPath": "params_shard_35.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.23.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "39411ad8b24ce0b7a5d11c11019a5e2e" }, { "dataPath": "params_shard_36.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.23.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "7da02773406b4a100f18adb3d310e3ab" }, { "dataPath": "params_shard_37.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.8.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "5be6893502c21759e190626e1beda0a0" }, { "dataPath": "params_shard_38.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.8.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "7d1f090a563943e673d1eb7224606144" }, { "dataPath": "params_shard_39.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.21.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 0 }, { "name": "model.layers.21.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 3145728 }, { "name": "model.layers.22.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 9437184 }, { "name": "model.layers.22.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 12582912 }, { "name": "model.layers.23.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 18874368 }, { "name": "model.layers.23.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 22020096 }, { "name": "model.layers.8.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 28311552 } ], "md5sum": "2c27b313f1434b2a4b86afc3e914a7df" }, { "dataPath": "params_shard_40.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.8.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 0 }, { "name": "model.layers.31.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 6291456 } ], "md5sum": "6fa687ce2becba1167acad27298af58d" }, { "dataPath": "params_shard_41.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.4.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "48ff7c4eb79fb31b286b0c0368b56da0" }, { "dataPath": "params_shard_42.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.31.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 0 }, { "name": "model.layers.4.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 3145728 }, { "name": "model.layers.4.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 28311552 } ], "md5sum": "caed110822905e5b7a7a88caefb8b657" }, { "dataPath": "params_shard_43.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.4.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 0 }, { "name": "model.layers.16.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 6291456 } ], "md5sum": "19386c956c743fee5b031b076e217ff5" }, { "dataPath": "params_shard_44.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.16.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "82560881445953f6a560a3df5566260d" }, { "dataPath": "params_shard_45.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.17.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "a1b0981703b390f795e766fcf3c69596" }, { "dataPath": "params_shard_46.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.17.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "38919366de3d8e4a3438b6433311fb5c" }, { "dataPath": "params_shard_47.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.18.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "73500e975649f8768233253edd091da9" }, { "dataPath": "params_shard_48.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.18.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "e619e809cd576ab874c1c4b939d52c0d" }, { "dataPath": "params_shard_49.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.19.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "48ee00f5f5745aaf7d944449aa97dc32" }, { "dataPath": "params_shard_50.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.19.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "9ad4083f7ef57b9c28475c803c237a8c" }, { "dataPath": "params_shard_51.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.16.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 0 }, { "name": "model.layers.16.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 3145728 }, { "name": "model.layers.17.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 9437184 }, { "name": "model.layers.17.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 12582912 }, { "name": "model.layers.18.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 18874368 }, { "name": "model.layers.18.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 22020096 }, { "name": "model.layers.19.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 28311552 } ], "md5sum": "9d5b6f5cd1579425d5690379d731e5e1" }, { "dataPath": "params_shard_52.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.19.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 0 }, { "name": "model.layers.24.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 6291456 } ], "md5sum": "0f77291bb46d4518ebe8a8ce208054db" }, { "dataPath": "params_shard_53.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.24.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "d6a58b4e996fc812eb572757f1138560" }, { "dataPath": "params_shard_54.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.25.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "0263fd4181a3c050b1c968c3d5ba33f6" }, { "dataPath": "params_shard_55.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.5.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "2a6719f21557bcc1775adc558875a409" }, { "dataPath": "params_shard_56.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.5.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "b32142506b3ab533d056276cb8688363" }, { "dataPath": "params_shard_57.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.6.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "9cc5ed1209bf7a67a885702581034214" }, { "dataPath": "params_shard_58.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.6.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "c924da4262a4ed14717ab0dc2a0ec812" }, { "dataPath": "params_shard_59.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.24.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 0 }, { "name": "model.layers.24.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 3145728 }, { "name": "model.layers.25.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 9437184 }, { "name": "model.layers.5.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 15728640 }, { "name": "model.layers.5.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 18874368 }, { "name": "model.layers.6.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 25165824 } ], "md5sum": "877ee350ea4071161f1773dd60448461" }, { "dataPath": "params_shard_60.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.6.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 0 }, { "name": "model.layers.7.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 6291456 } ], "md5sum": "2b87ff35657cdcb55a08d6a0c1c0efac" }, { "dataPath": "params_shard_61.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.7.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "b7e5d9d8435fa9d5ef3f2b1be0bec0dd" }, { "dataPath": "params_shard_62.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.2.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "0a1736392b8bfcddc45a1acc9b18b1fa" }, { "dataPath": "params_shard_63.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.2.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "450dd8f28b577be0cc265be086c4150f" }, { "dataPath": "params_shard_64.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.20.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "7c7fdf9f28ea2af31e18b76167679716" }, { "dataPath": "params_shard_65.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.20.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "9dd467e30c0a3702e9b6b3f242372708" }, { "dataPath": "params_shard_66.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.28.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "8440ffc2026b906f2fda3f14486ab443" }, { "dataPath": "params_shard_67.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.28.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "ea18174876506a2f84b70c0034e69142" }, { "dataPath": "params_shard_68.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.7.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 0 }, { "name": "model.layers.7.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 3145728 }, { "name": "model.layers.2.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 9437184 }, { "name": "model.layers.2.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 12582912 }, { "name": "model.layers.20.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 18874368 }, { "name": "model.layers.20.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 22020096 }, { "name": "model.layers.28.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 28311552 } ], "md5sum": "452d625bc871f158741f004fa64a098e" }, { "dataPath": "params_shard_69.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.28.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 0 }, { "name": "model.layers.29.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 6291456 } ], "md5sum": "bb60e308c4f7d06e1a98af29c4372e69" }, { "dataPath": "params_shard_70.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.29.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "99d6a220bbe214eb64fb803c12690abf" }, { "dataPath": "params_shard_71.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.12.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "ab1fcd38010c79ea6f2b1b13a1277728" }, { "dataPath": "params_shard_72.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.12.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "fb4fcdc2487a3bf2148a6cb7831485e4" }, { "dataPath": "params_shard_73.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "model.layers.13.mlp.down_proj.q_weight", "shape": [ 4096, 1536 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "8ffe607b173032de6757d55e96faf171" }, { "dataPath": "params_shard_74.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.layers.13.mlp.gate_up_proj.q_weight", "shape": [ 24576, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "e115316cac1fa949947f1bd79200c303" }, { "dataPath": "params_shard_75.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.14.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "96ad09d8fbca71a300f3504c9e4a956d" }, { "dataPath": "params_shard_76.bin", "format": "raw-shard", "nbytes": 20971520, "records": [ { "name": "model.layers.15.self_attn.c_attn.q_weight", "shape": [ 10240, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 20971520, "byteOffset": 0 } ], "md5sum": "3c89d372c51c769e9e25fead1eace1a3" }, { "dataPath": "params_shard_77.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.26.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "9fd1c6c2d0c773f077ca6b96b2a45127" }, { "dataPath": "params_shard_78.bin", "format": "raw-shard", "nbytes": 33030144, "records": [ { "name": "model.layers.29.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 0 }, { "name": "model.layers.29.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 3145728 }, { "name": "model.layers.12.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 9437184 }, { "name": "model.layers.12.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 12582912 }, { "name": "model.layers.13.mlp.down_proj.q_scale", "shape": [ 4096, 384 ], "dtype": "bfloat16", "format": "raw", "nbytes": 3145728, "byteOffset": 18874368 }, { "name": "model.layers.13.mlp.gate_up_proj.q_scale", "shape": [ 24576, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 6291456, "byteOffset": 22020096 }, { "name": "model.layers.14.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 28311552 }, { "name": "model.layers.15.self_attn.c_attn.q_scale", "shape": [ 10240, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2621440, "byteOffset": 30408704 } ], "md5sum": "2eb0dca1bdd9d30b072c0ee834a3b53a" }, { "dataPath": "params_shard_79.bin", "format": "raw-shard", "nbytes": 20971520, "records": [ { "name": "model.layers.3.self_attn.c_attn.q_weight", "shape": [ 10240, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 20971520, "byteOffset": 0 } ], "md5sum": "3046d7dc7c46a0b43dcf10704bbb16c7" }, { "dataPath": "params_shard_80.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.30.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "43f1e9c5010993bc9b62efdb4a82abd1" }, { "dataPath": "params_shard_81.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.9.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "71efdca332630176df5f5efaa3b5f50f" }, { "dataPath": "params_shard_82.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.10.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "b45fcab029ed75495923f1909d79503b" }, { "dataPath": "params_shard_83.bin", "format": "raw-shard", "nbytes": 32505856, "records": [ { "name": "model.layers.26.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 0 }, { "name": "model.layers.27.self_attn.c_attn.q_weight", "shape": [ 10240, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 20971520, "byteOffset": 2097152 }, { "name": "model.layers.27.self_attn.c_attn.q_scale", "shape": [ 10240, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2621440, "byteOffset": 23068672 }, { "name": "model.layers.3.self_attn.c_attn.q_scale", "shape": [ 10240, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2621440, "byteOffset": 25690112 }, { "name": "model.layers.30.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 28311552 }, { "name": "model.layers.9.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 30408704 } ], "md5sum": "2ea0c72296a02d4c3b096146fb27bfd2" }, { "dataPath": "params_shard_84.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.0.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "8f771a7b16b636656055681c430df47f" }, { "dataPath": "params_shard_85.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.1.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "7f63d77a249b0e9dcb2b5c6e2ca59f47" }, { "dataPath": "params_shard_86.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.22.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "7fdd85484712afb2c761d20af5d101e6" }, { "dataPath": "params_shard_87.bin", "format": "raw-shard", "nbytes": 20971520, "records": [ { "name": "model.layers.23.self_attn.c_attn.q_weight", "shape": [ 10240, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 20971520, "byteOffset": 0 } ], "md5sum": "09f5b3638a3a6531f43f567318f7a4fe" }, { "dataPath": "params_shard_88.bin", "format": "raw-shard", "nbytes": 31981568, "records": [ { "name": "model.layers.10.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 0 }, { "name": "model.layers.11.self_attn.c_attn.q_weight", "shape": [ 10240, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 20971520, "byteOffset": 2097152 }, { "name": "model.layers.11.self_attn.c_attn.q_scale", "shape": [ 10240, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2621440, "byteOffset": 23068672 }, { "name": "model.layers.0.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 25690112 }, { "name": "model.layers.1.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 27787264 }, { "name": "model.layers.22.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 29884416 } ], "md5sum": "53906a4800634adcf5aeeceeba5ceadf" }, { "dataPath": "params_shard_89.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.8.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "4fc37b2abab2d973d98b9ee5fee3f2a1" }, { "dataPath": "params_shard_90.bin", "format": "raw-shard", "nbytes": 20971520, "records": [ { "name": "model.layers.31.self_attn.c_attn.q_weight", "shape": [ 10240, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 20971520, "byteOffset": 0 } ], "md5sum": "63fbed465664200f7a4d0a9c8be6fc07" }, { "dataPath": "params_shard_91.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.4.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "cfa28475879001acdeb6b74f2ed05b69" }, { "dataPath": "params_shard_92.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.16.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "9f030b0005b157723c96b18f00538cad" }, { "dataPath": "params_shard_93.bin", "format": "raw-shard", "nbytes": 33030144, "records": [ { "name": "model.layers.23.self_attn.c_attn.q_scale", "shape": [ 10240, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2621440, "byteOffset": 0 }, { "name": "model.layers.7.self_attn.c_attn.q_weight", "shape": [ 10240, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 20971520, "byteOffset": 2621440 }, { "name": "model.layers.7.self_attn.c_attn.q_scale", "shape": [ 10240, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2621440, "byteOffset": 23592960 }, { "name": "model.layers.8.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 26214400 }, { "name": "model.layers.31.self_attn.c_attn.q_scale", "shape": [ 10240, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2621440, "byteOffset": 28311552 }, { "name": "model.layers.4.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 30932992 } ], "md5sum": "0c5dcfc9cec05003f6b718936b179539" }, { "dataPath": "params_shard_94.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.18.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "059dd4b432f4490f02ea85490d5920a5" }, { "dataPath": "params_shard_95.bin", "format": "raw-shard", "nbytes": 20971520, "records": [ { "name": "model.layers.19.self_attn.c_attn.q_weight", "shape": [ 10240, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 20971520, "byteOffset": 0 } ], "md5sum": "61c51cf34bcf73eabec36e13738bf45c" }, { "dataPath": "params_shard_96.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.2.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "e334579ce2efaa21e3d39c706880514a" }, { "dataPath": "params_shard_97.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.24.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "92daf5b46ca79ebfec1840e5912e7337" }, { "dataPath": "params_shard_98.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.25.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "d0f229e97ebbb57bbd27558e187ca91a" }, { "dataPath": "params_shard_99.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.5.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "a340c332deda22c8943b02100e5c5687" }, { "dataPath": "params_shard_100.bin", "format": "raw-shard", "nbytes": 31981568, "records": [ { "name": "model.layers.16.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 0 }, { "name": "model.layers.17.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 2097152 }, { "name": "model.layers.17.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 18874368 }, { "name": "model.layers.18.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 20971520 }, { "name": "model.layers.19.self_attn.c_attn.q_scale", "shape": [ 10240, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2621440, "byteOffset": 23068672 }, { "name": "model.layers.2.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 25690112 }, { "name": "model.layers.24.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 27787264 }, { "name": "model.layers.25.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 29884416 } ], "md5sum": "948a7ba5b5faeafe29c6b163e73bd890" }, { "dataPath": "params_shard_101.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.20.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "8369c4c8d7b22ffc2ee7edec78ff0505" }, { "dataPath": "params_shard_102.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.21.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "f7b22ab28ec338b9221dfa486a25da80" }, { "dataPath": "params_shard_103.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.28.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "8c4483d347e0eef11ff1b448f866041d" }, { "dataPath": "params_shard_104.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.29.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "6f2a8b13502d5536450b3779e19b4bd6" }, { "dataPath": "params_shard_105.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.12.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "dc305e89190030fcb70266f54172415d" }, { "dataPath": "params_shard_106.bin", "format": "raw-shard", "nbytes": 16777216, "records": [ { "name": "model.layers.13.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 0 } ], "md5sum": "4836db1b3acdd58a2a569d6d848aec70" }, { "dataPath": "params_shard_107.bin", "format": "raw-shard", "nbytes": 31457280, "records": [ { "name": "model.layers.5.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 0 }, { "name": "model.layers.6.linear_attn.in_proj_qkv.q_weight", "shape": [ 8192, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 16777216, "byteOffset": 2097152 }, { "name": "model.layers.6.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 18874368 }, { "name": "model.layers.20.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 20971520 }, { "name": "model.layers.21.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 23068672 }, { "name": "model.layers.28.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 25165824 }, { "name": "model.layers.29.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 27262976 }, { "name": "model.layers.12.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 29360128 } ], "md5sum": "b1323ad85aef3b0515e1372c5c7f163b" }, { "dataPath": "params_shard_108.bin", "format": "raw-shard", "nbytes": 30408704, "records": [ { "name": "model.layers.13.linear_attn.in_proj_qkv.q_scale", "shape": [ 8192, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 2097152, "byteOffset": 0 }, { "name": "model.layers.14.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 2097152 }, { "name": "model.layers.14.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 10485760 }, { "name": "model.layers.14.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 11534336 }, { "name": "model.layers.14.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 19922944 }, { "name": "model.layers.15.self_attn.o_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 20971520 }, { "name": "model.layers.15.self_attn.o_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 29360128 } ], "md5sum": "9a0889b5369925b81b0678d125227538" }, { "dataPath": "params_shard_109.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.26.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.26.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.26.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.26.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.27.self_attn.o_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.27.self_attn.o_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "2a2889cef63566a074f5ed5bbcc59ff4" }, { "dataPath": "params_shard_110.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.3.self_attn.o_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.3.self_attn.o_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.30.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.30.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.30.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.30.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "55ba9632fc1a91fc5613682ced3e2f3c" }, { "dataPath": "params_shard_111.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.9.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.9.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.9.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.9.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.10.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.10.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "b25670140842e74e785c470fae04d5d2" }, { "dataPath": "params_shard_112.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.10.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.10.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.11.self_attn.o_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.11.self_attn.o_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.0.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.0.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "883b3d228389b0de02055d403538fca5" }, { "dataPath": "params_shard_113.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.0.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.0.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.1.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.1.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.1.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.1.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "1ed723d9516ed2186e1f9996f87788e4" }, { "dataPath": "params_shard_114.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.21.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.21.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.22.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.22.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.22.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.22.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "c52dc0532c2c1d70d812e21b3affd503" }, { "dataPath": "params_shard_115.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.23.self_attn.o_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.23.self_attn.o_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.7.self_attn.o_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.7.self_attn.o_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.8.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.8.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "36d1407e12793efed55ef630fa0098ac" }, { "dataPath": "params_shard_116.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.8.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.8.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.31.self_attn.o_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.31.self_attn.o_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.4.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.4.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "525a0dd49d3092983034c2bb6a9af294" }, { "dataPath": "params_shard_117.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.4.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.4.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.16.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.16.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.16.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.16.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "e579ec23f63bcbfbec4c497197aacb07" }, { "dataPath": "params_shard_118.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.17.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.17.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.17.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.17.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.18.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.18.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "c0eb736a54275cf8ef383687c10a82d8" }, { "dataPath": "params_shard_119.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.18.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.18.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.19.self_attn.o_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.19.self_attn.o_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.2.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.2.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "2826c74eb2e9d6fede640b1a04b6b5a3" }, { "dataPath": "params_shard_120.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.24.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.24.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.24.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.24.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.25.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.25.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "9aa455a5ef98b1ec2e4381f80a2d32f7" }, { "dataPath": "params_shard_121.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.25.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.25.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.5.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.5.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.5.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.5.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "22c3b576d9e8c9c7ea779d1b04f94e57" }, { "dataPath": "params_shard_122.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.6.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.6.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.6.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.6.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.2.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.2.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "916921fb9d97a5cb73f6686926a84c30" }, { "dataPath": "params_shard_123.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.20.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.20.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.20.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.20.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.21.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.21.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "110de87284cb270f530998ca371bee3f" }, { "dataPath": "params_shard_124.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.28.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.28.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.28.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.28.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.29.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.29.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "9104d8008d9e1b72715ecebc729fe73f" }, { "dataPath": "params_shard_125.bin", "format": "raw-shard", "nbytes": 28311552, "records": [ { "name": "model.layers.29.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.29.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.12.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.12.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.12.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 18874368 }, { "name": "model.layers.12.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 27262976 } ], "md5sum": "7872687b7e0827ab2d493e3e7c33686a" }, { "dataPath": "params_shard_126.bin", "format": "raw-shard", "nbytes": 24536064, "records": [ { "name": "model.layers.13.linear_attn.in_proj_z.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "model.layers.13.linear_attn.in_proj_z.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 8388608 }, { "name": "model.layers.13.linear_attn.out_proj.q_weight", "shape": [ 4096, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 9437184 }, { "name": "model.layers.13.linear_attn.out_proj.q_scale", "shape": [ 4096, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 1048576, "byteOffset": 17825792 }, { "name": "model.layers.14.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 18874368 }, { "name": "model.layers.14.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 18939904 }, { "name": "model.layers.14.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 18948096 }, { "name": "model.layers.14.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 19013632 }, { "name": "model.layers.26.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19021824 }, { "name": "model.layers.26.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 19087360 }, { "name": "model.layers.26.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19095552 }, { "name": "model.layers.26.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 19161088 }, { "name": "model.layers.30.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19169280 }, { "name": "model.layers.30.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 19234816 }, { "name": "model.layers.30.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19243008 }, { "name": "model.layers.30.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 19308544 }, { "name": "model.layers.10.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19316736 }, { "name": "model.layers.10.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 19382272 }, { "name": "model.layers.10.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19390464 }, { "name": "model.layers.10.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 19456000 }, { "name": "model.layers.0.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19464192 }, { "name": "model.layers.0.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 19529728 }, { "name": "model.layers.0.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19537920 }, { "name": "model.layers.0.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 19603456 }, { "name": "model.layers.1.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19611648 }, { "name": "model.layers.1.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 19677184 }, { "name": "model.layers.1.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19685376 }, { "name": "model.layers.1.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 19750912 }, { "name": "model.layers.22.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19759104 }, { "name": "model.layers.22.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 19824640 }, { "name": "model.layers.22.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19832832 }, { "name": "model.layers.22.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 19898368 }, { "name": "model.layers.8.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19906560 }, { "name": "model.layers.8.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 19972096 }, { "name": "model.layers.8.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 19980288 }, { "name": "model.layers.8.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 20045824 }, { "name": "model.layers.9.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20054016 }, { "name": "model.layers.9.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 20119552 }, { "name": "model.layers.9.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20127744 }, { "name": "model.layers.9.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 20193280 }, { "name": "model.layers.4.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20201472 }, { "name": "model.layers.4.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 20267008 }, { "name": "model.layers.4.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20275200 }, { "name": "model.layers.4.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 20340736 }, { "name": "model.layers.16.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20348928 }, { "name": "model.layers.16.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 20414464 }, { "name": "model.layers.16.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20422656 }, { "name": "model.layers.16.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 20488192 }, { "name": "model.layers.17.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20496384 }, { "name": "model.layers.17.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 20561920 }, { "name": "model.layers.17.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20570112 }, { "name": "model.layers.17.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 20635648 }, { "name": "model.layers.18.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20643840 }, { "name": "model.layers.18.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 20709376 }, { "name": "model.layers.18.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20717568 }, { "name": "model.layers.18.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 20783104 }, { "name": "model.layers.2.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20791296 }, { "name": "model.layers.2.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 20856832 }, { "name": "model.layers.2.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20865024 }, { "name": "model.layers.2.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 20930560 }, { "name": "model.layers.24.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 20938752 }, { "name": "model.layers.24.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 21004288 }, { "name": "model.layers.24.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21012480 }, { "name": "model.layers.24.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 21078016 }, { "name": "model.layers.25.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21086208 }, { "name": "model.layers.25.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 21151744 }, { "name": "model.layers.25.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21159936 }, { "name": "model.layers.25.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 21225472 }, { "name": "model.layers.5.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21233664 }, { "name": "model.layers.5.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 21299200 }, { "name": "model.layers.5.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21307392 }, { "name": "model.layers.5.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 21372928 }, { "name": "model.layers.6.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21381120 }, { "name": "model.layers.6.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 21446656 }, { "name": "model.layers.6.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21454848 }, { "name": "model.layers.6.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 21520384 }, { "name": "model.layers.20.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21528576 }, { "name": "model.layers.20.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 21594112 }, { "name": "model.layers.20.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21602304 }, { "name": "model.layers.20.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 21667840 }, { "name": "model.layers.21.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21676032 }, { "name": "model.layers.21.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 21741568 }, { "name": "model.layers.21.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21749760 }, { "name": "model.layers.21.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 21815296 }, { "name": "model.layers.28.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21823488 }, { "name": "model.layers.28.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 21889024 }, { "name": "model.layers.28.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21897216 }, { "name": "model.layers.28.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 21962752 }, { "name": "model.layers.29.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 21970944 }, { "name": "model.layers.29.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 22036480 }, { "name": "model.layers.29.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 22044672 }, { "name": "model.layers.29.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 22110208 }, { "name": "model.layers.12.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 22118400 }, { "name": "model.layers.12.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 22183936 }, { "name": "model.layers.12.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 22192128 }, { "name": "model.layers.12.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 22257664 }, { "name": "model.layers.13.linear_attn.in_proj_b.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 22265856 }, { "name": "model.layers.13.linear_attn.in_proj_b.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 22331392 }, { "name": "model.layers.13.linear_attn.in_proj_a.q_weight", "shape": [ 32, 512 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 65536, "byteOffset": 22339584 }, { "name": "model.layers.13.linear_attn.in_proj_a.q_scale", "shape": [ 32, 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 22405120 }, { "name": "model.layers.14.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 22413312 }, { "name": "model.layers.16.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 22478848 }, { "name": "model.layers.26.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 22544384 }, { "name": "model.layers.30.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 22609920 }, { "name": "model.layers.10.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 22675456 }, { "name": "model.layers.0.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 22740992 }, { "name": "model.layers.1.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 22806528 }, { "name": "model.layers.22.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 22872064 }, { "name": "model.layers.8.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 22937600 }, { "name": "model.layers.9.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23003136 }, { "name": "model.layers.4.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23068672 }, { "name": "model.layers.5.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23134208 }, { "name": "model.layers.17.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23199744 }, { "name": "model.layers.18.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23265280 }, { "name": "model.layers.2.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23330816 }, { "name": "model.layers.24.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23396352 }, { "name": "model.layers.25.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23461888 }, { "name": "model.layers.6.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23527424 }, { "name": "model.layers.20.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23592960 }, { "name": "model.layers.21.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23658496 }, { "name": "model.layers.28.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23724032 }, { "name": "model.layers.29.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23789568 }, { "name": "model.layers.12.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23855104 }, { "name": "model.layers.13.linear_attn.conv1d_weight", "shape": [ 8192, 1, 4 ], "dtype": "bfloat16", "format": "raw", "nbytes": 65536, "byteOffset": 23920640 }, { "name": "model.layers.14.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 23986176 }, { "name": "model.layers.14.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 23994368 }, { "name": "model.layers.15.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24002560 }, { "name": "model.layers.15.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24010752 }, { "name": "model.layers.16.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24018944 }, { "name": "model.layers.26.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24027136 }, { "name": "model.layers.26.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24035328 }, { "name": "model.layers.27.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24043520 }, { "name": "model.layers.27.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24051712 }, { "name": "model.layers.3.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24059904 }, { "name": "model.layers.3.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24068096 }, { "name": "model.layers.30.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24076288 }, { "name": "model.layers.30.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24084480 }, { "name": "model.layers.31.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24092672 }, { "name": "model.layers.9.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24100864 }, { "name": "model.layers.10.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24109056 }, { "name": "model.layers.10.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24117248 }, { "name": "model.layers.11.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24125440 }, { "name": "model.layers.11.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24133632 }, { "name": "model.layers.0.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24141824 }, { "name": "model.layers.0.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24150016 }, { "name": "model.layers.1.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24158208 }, { "name": "model.layers.1.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24166400 }, { "name": "model.norm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24174592 }, { "name": "model.layers.21.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24182784 }, { "name": "model.layers.22.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24190976 }, { "name": "model.layers.22.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24199168 }, { "name": "model.layers.23.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24207360 }, { "name": "model.layers.23.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24215552 }, { "name": "model.layers.7.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24223744 }, { "name": "model.layers.8.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24231936 }, { "name": "model.layers.8.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24240128 }, { "name": "model.layers.9.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24248320 }, { "name": "model.layers.31.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24256512 }, { "name": "model.layers.4.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24264704 }, { "name": "model.layers.4.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24272896 }, { "name": "model.layers.5.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24281088 }, { "name": "model.layers.16.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24289280 }, { "name": "model.layers.17.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24297472 }, { "name": "model.layers.17.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24305664 }, { "name": "model.layers.18.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24313856 }, { "name": "model.layers.18.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24322048 }, { "name": "model.layers.19.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24330240 }, { "name": "model.layers.19.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24338432 }, { "name": "model.layers.2.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24346624 }, { "name": "model.layers.24.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24354816 }, { "name": "model.layers.24.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24363008 }, { "name": "model.layers.25.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24371200 }, { "name": "model.layers.25.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24379392 }, { "name": "model.layers.5.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24387584 }, { "name": "model.layers.6.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24395776 }, { "name": "model.layers.6.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24403968 }, { "name": "model.layers.7.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24412160 }, { "name": "model.layers.2.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24420352 }, { "name": "model.layers.20.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24428544 }, { "name": "model.layers.20.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24436736 }, { "name": "model.layers.21.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24444928 }, { "name": "model.layers.28.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24453120 }, { "name": "model.layers.28.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24461312 }, { "name": "model.layers.29.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24469504 }, { "name": "model.layers.29.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24477696 }, { "name": "model.layers.12.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24485888 }, { "name": "model.layers.12.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24494080 }, { "name": "model.layers.13.input_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24502272 }, { "name": "model.layers.13.post_attention_layernorm.weight", "shape": [ 4096 ], "dtype": "bfloat16", "format": "raw", "nbytes": 8192, "byteOffset": 24510464 }, { "name": "model.layers.14.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24518656 }, { "name": "model.layers.15.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24518912 }, { "name": "model.layers.15.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24519424 }, { "name": "model.layers.26.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24519936 }, { "name": "model.layers.27.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24520192 }, { "name": "model.layers.27.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24520704 }, { "name": "model.layers.3.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24521216 }, { "name": "model.layers.3.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24521728 }, { "name": "model.layers.30.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24522240 }, { "name": "model.layers.9.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24522496 }, { "name": "model.layers.10.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24522752 }, { "name": "model.layers.11.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24523008 }, { "name": "model.layers.11.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24523520 }, { "name": "model.layers.0.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24524032 }, { "name": "model.layers.1.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24524288 }, { "name": "model.layers.22.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24524544 }, { "name": "model.layers.23.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24524800 }, { "name": "model.layers.7.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24525312 }, { "name": "model.layers.7.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24525824 }, { "name": "model.layers.8.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24526336 }, { "name": "model.layers.31.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24526592 }, { "name": "model.layers.31.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24527104 }, { "name": "model.layers.4.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24527616 }, { "name": "model.layers.16.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24527872 }, { "name": "model.layers.17.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24528128 }, { "name": "model.layers.18.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24528384 }, { "name": "model.layers.19.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24528640 }, { "name": "model.layers.19.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24529152 }, { "name": "model.layers.2.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24529664 }, { "name": "model.layers.23.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "bfloat16", "format": "raw", "nbytes": 512, "byteOffset": 24529920 }, { "name": "model.layers.24.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24530432 }, { "name": "model.layers.25.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24530688 }, { "name": "model.layers.5.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24530944 }, { "name": "model.layers.6.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24531200 }, { "name": "model.layers.20.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24531456 }, { "name": "model.layers.21.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24531712 }, { "name": "model.layers.28.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24531968 }, { "name": "model.layers.29.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24532224 }, { "name": "model.layers.12.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24532480 }, { "name": "model.layers.13.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "bfloat16", "format": "raw", "nbytes": 256, "byteOffset": 24532736 }, { "name": "model.layers.16.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24532992 }, { "name": "model.layers.26.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533056 }, { "name": "model.layers.28.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533120 }, { "name": "model.layers.30.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533184 }, { "name": "model.layers.10.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533248 }, { "name": "model.layers.12.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533312 }, { "name": "model.layers.0.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533376 }, { "name": "model.layers.1.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533440 }, { "name": "model.layers.22.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533504 }, { "name": "model.layers.8.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533568 }, { "name": "model.layers.9.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533632 }, { "name": "model.layers.4.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533696 }, { "name": "model.layers.5.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533760 }, { "name": "model.layers.17.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533824 }, { "name": "model.layers.18.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533888 }, { "name": "model.layers.2.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24533952 }, { "name": "model.layers.24.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534016 }, { "name": "model.layers.25.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534080 }, { "name": "model.layers.6.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534144 }, { "name": "model.layers.20.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534208 }, { "name": "model.layers.21.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534272 }, { "name": "model.layers.29.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534336 }, { "name": "model.layers.13.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534400 }, { "name": "model.layers.14.linear_attn.A_log", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534464 }, { "name": "model.layers.14.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534528 }, { "name": "model.layers.16.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534592 }, { "name": "model.layers.26.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534656 }, { "name": "model.layers.30.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534720 }, { "name": "model.layers.10.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534784 }, { "name": "model.layers.0.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534848 }, { "name": "model.layers.1.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534912 }, { "name": "model.layers.22.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24534976 }, { "name": "model.layers.8.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535040 }, { "name": "model.layers.9.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535104 }, { "name": "model.layers.4.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535168 }, { "name": "model.layers.5.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535232 }, { "name": "model.layers.17.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535296 }, { "name": "model.layers.18.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535360 }, { "name": "model.layers.2.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535424 }, { "name": "model.layers.24.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535488 }, { "name": "model.layers.25.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535552 }, { "name": "model.layers.6.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535616 }, { "name": "model.layers.20.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535680 }, { "name": "model.layers.21.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535744 }, { "name": "model.layers.28.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535808 }, { "name": "model.layers.29.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535872 }, { "name": "model.layers.12.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24535936 }, { "name": "model.layers.13.linear_attn.dt_bias", "shape": [ 32 ], "dtype": "bfloat16", "format": "raw", "nbytes": 64, "byteOffset": 24536000 } ], "md5sum": "ae837c02ee18f418cd9070fb314c1e82" } ] }