{ "metadata": { "ParamSize": 592, "ParamBytes": 8293242880.0, "BitsPerParam": 16.38931884531257 }, "records": [ { "dataPath": "params_shard_0.bin", "format": "raw-shard", "nbytes": 197001216, "records": [ { "name": "lm_head.weight", "shape": [ 32064, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 197001216, "byteOffset": 0 } ], "md5sum": "3bd75f787fce805f0a08dfb47e67cb24" }, { "dataPath": "params_shard_1.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.18.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "d34daea5d956ec270289eba4ac01591d" }, { "dataPath": "params_shard_2.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.18.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "81c3285fcfd94c4294a8efa052e3c900" }, { "dataPath": "params_shard_3.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.18.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "e3473ac0bf7df30d4f248ff311bb67d9" }, { "dataPath": "params_shard_4.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.19.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "13f4d30c1d1bfcf6ad4347bf9caa49ba" }, { "dataPath": "params_shard_5.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.19.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "0008476d961c936c44f63f5b59b03151" }, { "dataPath": "params_shard_6.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.19.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "9d191039939d32777cc8442abdfbf0f8" }, { "dataPath": "params_shard_7.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.20.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "feb2ebf8878533595d8d9e41f81eec7a" }, { "dataPath": "params_shard_8.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.20.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "57ea242ee2af7a6c2d5d54c876d34841" }, { "dataPath": "params_shard_9.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.20.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "ee44054e4b3673f77c5d6ace147f7850" }, { "dataPath": "params_shard_10.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.20.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "cea98c7527face4b508995f5a27c6c37" }, { "dataPath": "params_shard_11.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.21.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "6a682a57e3edc80d42b254aebf47272b" }, { "dataPath": "params_shard_12.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.21.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "4cdd6999468bc183b400be8715bd6717" }, { "dataPath": "params_shard_13.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.21.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "87468a88f1d150afd954fb4c9798bf0f" }, { "dataPath": "params_shard_14.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.21.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "94e461f68dad7b9be57478cb80bfc08f" }, { "dataPath": "params_shard_15.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.22.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "0cf18a5d69cd41b328b18be6889716f2" }, { "dataPath": "params_shard_16.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.22.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "a0beb2b96b429412caa9086b416b0132" }, { "dataPath": "params_shard_17.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.22.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "64b3e37b777572c7061e59e46a7e623c" }, { "dataPath": "params_shard_18.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.22.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "45caef336b29193358253b651a18a2a8" }, { "dataPath": "params_shard_19.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.23.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "f56d4e61b3a8deedad69a4af9a77fe3e" }, { "dataPath": "params_shard_20.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.23.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "a573301fc3aeb2e7d7d919ad7ac3a99c" }, { "dataPath": "params_shard_21.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.23.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "08313998391fac099a5adeca66fdca72" }, { "dataPath": "params_shard_22.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.23.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "fc84f73045f157d9d60358d971f269fa" }, { "dataPath": "params_shard_23.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.24.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "02802510fe53bfa2b9042c873f57eaa2" }, { "dataPath": "params_shard_24.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.24.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "760d2ba08bbb699a578d5a178ea0258d" }, { "dataPath": "params_shard_25.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.24.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "d0cca79c80e646bbdb69f3dc3475f612" }, { "dataPath": "params_shard_26.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.24.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "8d3addf9ac08b707a9958a8c0ea160ff" }, { "dataPath": "params_shard_27.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.25.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "5ef0beafce88b4a74bb29ce15ba3cc97" }, { "dataPath": "params_shard_28.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.25.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "818617db19ba5e8ecc926b549b08d4c2" }, { "dataPath": "params_shard_29.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.25.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "ff41221dd37cf0cd69690e993ff89b09" }, { "dataPath": "params_shard_30.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.25.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "48c002c82892f1fa84804e5e9f5e2ad8" }, { "dataPath": "params_shard_31.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.26.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "0c46c3a9db54465f73e8293220efd16e" }, { "dataPath": "params_shard_32.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.26.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "966d33f3925327b263eba597566f74e1" }, { "dataPath": "params_shard_33.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.26.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "e6049873bcaefb87ced688abe05f05cf" }, { "dataPath": "params_shard_34.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.26.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "d61163ef23fd758a858706ab249879bd" }, { "dataPath": "params_shard_35.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.27.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "42a9557556242794c22a01bdae6b0eef" }, { "dataPath": "params_shard_36.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.27.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "590a094a2a364072da309c98d5b9a5da" }, { "dataPath": "params_shard_37.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.27.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "d81430f2a5f7b2c721bf5c0057562213" }, { "dataPath": "params_shard_38.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.27.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "bab528045f745b4edde880754402f966" }, { "dataPath": "params_shard_39.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.28.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "43c7e6909cf27fab9befa20c400d5668" }, { "dataPath": "params_shard_40.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.28.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "34b032dfb8027f61c2c47bd4cbfbd5f8" }, { "dataPath": "params_shard_41.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.28.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "1870913879b764554021eaac5bfed6c5" }, { "dataPath": "params_shard_42.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.28.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "76063422dd741b0f8ff2b28b964bd37e" }, { "dataPath": "params_shard_43.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.29.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "39b655fd36e2a1712bdc2a5d4d2c98af" }, { "dataPath": "params_shard_44.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.29.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "b3f6ee8f62444b4069fabc3ae2c65864" }, { "dataPath": "params_shard_45.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.29.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "da1a3abb0aefaebd34ab65253b31025a" }, { "dataPath": "params_shard_46.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.29.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "96293b17c0f81ded4633feba9d6ada27" }, { "dataPath": "params_shard_47.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.30.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "595755d292fc58915645256a8ef5e27f" }, { "dataPath": "params_shard_48.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.30.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "fffe778efce0d712df2a2c21f022d46b" }, { "dataPath": "params_shard_49.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.30.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "59dcaf51faf1d70a61cbc053c734f2f7" }, { "dataPath": "params_shard_50.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.30.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "c4ce8e6f5c3d819b1903e3e4f08fe865" }, { "dataPath": "params_shard_51.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.31.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "e05a4b46048ff5cb480b3d002ad0784e" }, { "dataPath": "params_shard_52.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.31.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "3c0eee071fc8ae99d0cb02a1836735e5" }, { "dataPath": "params_shard_53.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.31.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "28880ebdb7ac668a17d19be77eec864f" }, { "dataPath": "params_shard_54.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.31.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "a5697ccfcf6f86729c1ac893dfef0422" }, { "dataPath": "params_shard_55.bin", "format": "raw-shard", "nbytes": 197001216, "records": [ { "name": "model.embd.weight", "shape": [ 32064, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 197001216, "byteOffset": 0 } ], "md5sum": "29dc6d1096d086d6b0c8c7aec7aad0a0" }, { "dataPath": "params_shard_56.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.0.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "b2da976ce349ade2d1c6e93d35ebb5b3" }, { "dataPath": "params_shard_57.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.0.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "f4ff0eabf3ed88414b8dd3c351f176d6" }, { "dataPath": "params_shard_58.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.0.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "5e17603551688c8df6a63996820ed3b8" }, { "dataPath": "params_shard_59.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.0.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "a18a2eb823911c4acce89efd23c4811a" }, { "dataPath": "params_shard_60.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.1.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "3502768e75e32df1f9b3b9cce70d5a50" }, { "dataPath": "params_shard_61.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.1.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "7842a0096f2e99acbb60bce5ce2b328a" }, { "dataPath": "params_shard_62.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.1.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "d57ee3889abf229d665e0e6008c9df19" }, { "dataPath": "params_shard_63.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.1.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "f08e7406226ac953f11fa8284786e240" }, { "dataPath": "params_shard_64.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.10.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "0edfd9348b35db8cb6f2c01efa757230" }, { "dataPath": "params_shard_65.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.10.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "7a677db32eebb14ff9cbd47e972e9ec4" }, { "dataPath": "params_shard_66.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.10.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "fa21cabe2ec37d51c5197a7dff51c9f0" }, { "dataPath": "params_shard_67.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.10.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "bad4ab8fa1e5d7d8134ed79ad3bb8c32" }, { "dataPath": "params_shard_68.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.11.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "bcaa7bf6df59f0a22820959c3262ac5a" }, { "dataPath": "params_shard_69.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.11.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "0dd3de2554de730a35aac641633be892" }, { "dataPath": "params_shard_70.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.11.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "153cb83832de0db654c19a77eec544a5" }, { "dataPath": "params_shard_71.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.11.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "3619d05de7d340adccdfb4c8f584b13f" }, { "dataPath": "params_shard_72.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.12.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "771d0ddde92088fff219ab7036480b6d" }, { "dataPath": "params_shard_73.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.12.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "4504bee6b790e57cdaf00768fea9d0af" }, { "dataPath": "params_shard_74.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.12.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "1d9c73464e19684d575bcc7897c70de8" }, { "dataPath": "params_shard_75.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.12.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "9e9643b067257e91cf1343bd6f11a5b1" }, { "dataPath": "params_shard_76.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.13.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "72795ecfe7e8a88b00c4aac24445661c" }, { "dataPath": "params_shard_77.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.13.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "1058b88374d1094fe3b7360053bd0517" }, { "dataPath": "params_shard_78.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.13.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "eb24e5dfd04e11dec1a03064ba33e87b" }, { "dataPath": "params_shard_79.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.13.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "ee2e84daa7adf938af97df93fa38c050" }, { "dataPath": "params_shard_80.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.14.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "c04732e5d765d2c83f959842998c217d" }, { "dataPath": "params_shard_81.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.14.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "1f80bccc2f7d41b4080542a3547439a0" }, { "dataPath": "params_shard_82.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.14.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "ed54a8e4b1995e7bdc2488aa17082480" }, { "dataPath": "params_shard_83.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.14.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "7772416580dd709c30c293ff98e45318" }, { "dataPath": "params_shard_84.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.15.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "49795dd8867199988952e201e1c0d0a6" }, { "dataPath": "params_shard_85.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.15.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "58df2c89d5eb3b8439337c8e0e7a8d56" }, { "dataPath": "params_shard_86.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.15.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "1f0711c628e61e9fffcc272fdddedf7d" }, { "dataPath": "params_shard_87.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.15.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "5c8e65f304144bc226de5ff3b9cec33b" }, { "dataPath": "params_shard_88.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.16.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "86f99f9d2ed99f375e9b94cb945c531a" }, { "dataPath": "params_shard_89.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.16.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "4a51af9590a8ce254af4aa2950a1627d" }, { "dataPath": "params_shard_90.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.16.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "e66dea67a723301dbb3d5a22c81e0c12" }, { "dataPath": "params_shard_91.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.16.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "d9739a78ac3ec0c571de0a18dca9cb90" }, { "dataPath": "params_shard_92.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.17.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "f2c1fcc3c190fb743676aa1a83c5d422" }, { "dataPath": "params_shard_93.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.17.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "dae737fc4f0aa9445a832d00d96f6823" }, { "dataPath": "params_shard_94.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.17.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "f4c6cada510246d0578816c2b51ab774" }, { "dataPath": "params_shard_95.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.17.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "e51f97c01c67f0ecb6b8eee826456a51" }, { "dataPath": "params_shard_96.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.18.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "9806daf8abd1f34de1729f16efb3e1a0" }, { "dataPath": "params_shard_97.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.2.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "e220dbaebc67287f7ba00850f862b94b" }, { "dataPath": "params_shard_98.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.2.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "cc514440ec43805290b553205652ef53" }, { "dataPath": "params_shard_99.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.2.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "5b42ea8fd5a207493c85c47100e00fdd" }, { "dataPath": "params_shard_100.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.2.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "15147819a1dba504e7a598bf3d8a93f0" }, { "dataPath": "params_shard_101.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.3.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "37cff2d1ef415c3f873388a68e0b99d9" }, { "dataPath": "params_shard_102.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.3.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "1b2ad76eb207be3e704de327c6b453bd" }, { "dataPath": "params_shard_103.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.3.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "10f810ae12e500e1ca1bf425853e1b3a" }, { "dataPath": "params_shard_104.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.3.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "5fcbbffc3b516405d69a72f7da411617" }, { "dataPath": "params_shard_105.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.4.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "e71c80600dfe8031ba8c7f72bcf5a454" }, { "dataPath": "params_shard_106.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.4.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "314b851d14795c2f53a624854d85b636" }, { "dataPath": "params_shard_107.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.4.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "a3a6d67126595ad98cb784473b58c8ef" }, { "dataPath": "params_shard_108.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.4.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "f4db4847b76df85aa65a9edc1ac55f60" }, { "dataPath": "params_shard_109.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.5.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "d27ab670f2043a80e645c8d8ae480d8f" }, { "dataPath": "params_shard_110.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.5.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "acb1327a8f6196f7c971795e250119b3" }, { "dataPath": "params_shard_111.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.5.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "bd398d78b26319d158e94614d6d0bb8c" }, { "dataPath": "params_shard_112.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.5.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "8b2d21dafda9e00bb4bf62ec6547438d" }, { "dataPath": "params_shard_113.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.6.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "341504f05efd0ecec46276a541c36f99" }, { "dataPath": "params_shard_114.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.6.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "2dfe5aaf49bfd16bd06ddee75c50efe8" }, { "dataPath": "params_shard_115.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.6.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "4a3cb786c5d59797a0dd35d143609aed" }, { "dataPath": "params_shard_116.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.6.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "151ec9a7f5cbda7de5797df9474e1dd0" }, { "dataPath": "params_shard_117.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.7.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "5a08341676dc12966486ba1db0938b01" }, { "dataPath": "params_shard_118.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.7.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "2c4cfbb3f4e562b960d53459ab98f9db" }, { "dataPath": "params_shard_119.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.7.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "3677fc1a2990c1f5ee213c31df271003" }, { "dataPath": "params_shard_120.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.7.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "34773a73852f650698da6f6839ef67f5" }, { "dataPath": "params_shard_121.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.8.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "d91c5c2a364e87b0c155386ddda0de3d" }, { "dataPath": "params_shard_122.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.8.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "9de4e9c9b24ec7ca1d74e056476a57de" }, { "dataPath": "params_shard_123.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.8.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "ae04eb0dbbca5c04a5bd647e04145ba1" }, { "dataPath": "params_shard_124.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.8.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "ffb28fb4ce7a409e942f8d7606fadfdd" }, { "dataPath": "params_shard_125.bin", "format": "raw-shard", "nbytes": 50331648, "records": [ { "name": "model.h.9.mlp.down_proj.weight", "shape": [ 3072, 8192 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 50331648, "byteOffset": 0 } ], "md5sum": "4ce1f0c7ca1e34c12283ca041e97b17a" }, { "dataPath": "params_shard_126.bin", "format": "raw-shard", "nbytes": 100663296, "records": [ { "name": "model.h.9.mlp.gate_up_proj.weight", "shape": [ 16384, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 100663296, "byteOffset": 0 } ], "md5sum": "112168d6ef7b921ff071f7fbbef78868" }, { "dataPath": "params_shard_127.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "model.h.9.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "7fecb95ebb57d905a714267f959ac366" }, { "dataPath": "params_shard_128.bin", "format": "raw-shard", "nbytes": 56623104, "records": [ { "name": "model.h.9.mixer.qkv_proj.weight", "shape": [ 9216, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 56623104, "byteOffset": 0 } ], "md5sum": "19f233adcafaf74af768eaea87801b5b" }, { "dataPath": "params_shard_129.bin", "format": "raw-shard", "nbytes": 30076928, "records": [ { "name": "model.h.18.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 0 }, { "name": "model.h.18.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 6144 }, { "name": "model.h.19.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 12288 }, { "name": "model.h.19.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18432 }, { "name": "model.h.19.mixer.out_proj.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 24576 }, { "name": "model.h.20.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18898944 }, { "name": "model.h.20.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18905088 }, { "name": "model.h.21.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18911232 }, { "name": "model.h.21.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18917376 }, { "name": "model.h.22.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18923520 }, { "name": "model.h.22.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18929664 }, { "name": "model.h.23.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18935808 }, { "name": "model.h.23.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18941952 }, { "name": "model.h.24.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18948096 }, { "name": "model.h.24.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18954240 }, { "name": "model.h.25.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18960384 }, { "name": "model.h.25.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18966528 }, { "name": "model.h.26.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18972672 }, { "name": "model.h.26.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18978816 }, { "name": "model.h.27.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18984960 }, { "name": "model.h.27.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18991104 }, { "name": "model.h.28.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 18997248 }, { "name": "model.h.28.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19003392 }, { "name": "model.h.29.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19009536 }, { "name": "model.h.29.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19015680 }, { "name": "model.h.30.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19021824 }, { "name": "model.h.30.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19027968 }, { "name": "model.h.31.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19034112 }, { "name": "model.h.31.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19040256 }, { "name": "model.norm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19046400 }, { "name": "model.h.0.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19052544 }, { "name": "model.h.0.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19058688 }, { "name": "model.h.1.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19064832 }, { "name": "model.h.1.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19070976 }, { "name": "model.h.10.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19077120 }, { "name": "model.h.10.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19083264 }, { "name": "model.h.11.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19089408 }, { "name": "model.h.11.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19095552 }, { "name": "model.h.12.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19101696 }, { "name": "model.h.12.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19107840 }, { "name": "model.h.13.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19113984 }, { "name": "model.h.13.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19120128 }, { "name": "model.h.14.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19126272 }, { "name": "model.h.14.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19132416 }, { "name": "model.h.15.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19138560 }, { "name": "model.h.15.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19144704 }, { "name": "model.h.16.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19150848 }, { "name": "model.h.16.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19156992 }, { "name": "model.h.17.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19163136 }, { "name": "model.h.17.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19169280 }, { "name": "model.h.2.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19175424 }, { "name": "model.h.2.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19181568 }, { "name": "model.h.3.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19187712 }, { "name": "model.h.3.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19193856 }, { "name": "model.h.4.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19200000 }, { "name": "model.h.4.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19206144 }, { "name": "model.h.5.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19212288 }, { "name": "model.h.5.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19218432 }, { "name": "model.h.6.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19224576 }, { "name": "model.h.6.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19230720 }, { "name": "model.h.7.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19236864 }, { "name": "model.h.7.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19243008 }, { "name": "model.h.8.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19249152 }, { "name": "model.h.8.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19255296 }, { "name": "model.h.9.ln.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19261440 }, { "name": "model.h.9.post_attention_layernorm.weight", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 19267584 }, { "name": "vision_embed_tokens.glb_GN", "shape": [ 1, 1, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 19273728 }, { "name": "vision_embed_tokens.img_processor.vision_model.embeddings.class_embedding", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 19281920 }, { "name": "vision_embed_tokens.img_processor.vision_model.embeddings.patch_embedding.weight", "shape": [ 1024, 3, 14, 14 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1204224, "byteOffset": 19283968 }, { "name": "vision_embed_tokens.img_processor.vision_model.embeddings.position_embedding.weight", "shape": [ 577, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1181696, "byteOffset": 20488192 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 21669888 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 21671936 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 21673984 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 21676032 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 21678080 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 21686272 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 30074880 } ], "md5sum": "e28856c7ab4de7364a4bd796301f9197" }, { "dataPath": "params_shard_130.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.0.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "12ccc3786467895471d31bb5b1d2610c" }, { "dataPath": "params_shard_131.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.1.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "dcc2adf61aa07fe189bb15874110e845" }, { "dataPath": "params_shard_132.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.10.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "9f8d49d8894a05f4179ba707925e9d59" }, { "dataPath": "params_shard_133.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.11.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "19b6f5d2cfc839108f9156b8b521d3ce" }, { "dataPath": "params_shard_134.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.12.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "3e6322c0673873144d67964c709fa3be" }, { "dataPath": "params_shard_135.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.13.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "f04370015a40ae2073f5d89ad86f1c42" }, { "dataPath": "params_shard_136.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.14.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "e2c99d455ee851ede30d282481166453" }, { "dataPath": "params_shard_137.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.15.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "11c75f74757caccc61b9f6c3f39a9737" }, { "dataPath": "params_shard_138.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.16.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "a7f57fd25793b85aa6027a514efc39fb" }, { "dataPath": "params_shard_139.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.17.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "e13228942bc433e638fe3903001715f3" }, { "dataPath": "params_shard_140.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.18.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "eab0a6fc1a1e3beac991a0be01fdefe8" }, { "dataPath": "params_shard_141.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.19.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "afcfd0263f305fcc63e7b86393753bf7" }, { "dataPath": "params_shard_142.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.2.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "460a8bc76edf812f85886ec9f950f395" }, { "dataPath": "params_shard_143.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.20.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "bcac984764bba0a4eb9891d2043e1d89" }, { "dataPath": "params_shard_144.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.21.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "3197e582d745d84c2ab4347a1f2d4e2d" }, { "dataPath": "params_shard_145.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.22.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "e6d67ae9f4dab5ea49fac79bceac7a86" }, { "dataPath": "params_shard_146.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.23.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "1a74510bf14d6fc06b53f09ee6a22fdd" }, { "dataPath": "params_shard_147.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.3.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "75a9ee23e852e5562e2176a5ece04f24" }, { "dataPath": "params_shard_148.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.4.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "da3f4d64dad38aa57a47fd6fbca1a334" }, { "dataPath": "params_shard_149.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.5.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "ca1eb7481171d37cc9fab4b3432d81d1" }, { "dataPath": "params_shard_150.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.6.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "03be4317a6855a15625866d5dde56763" }, { "dataPath": "params_shard_151.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.7.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "7464cccd529310def6e3920c637ad62c" }, { "dataPath": "params_shard_152.bin", "format": "raw-shard", "nbytes": 25192448, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.8.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.layer_norm1.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.layer_norm1.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.layer_norm2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.layer_norm2.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.mlp.fc1.bias", "shape": [ 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.mlp.fc1.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 16801792 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.mlp.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 25190400 } ], "md5sum": "8be48481c922751c9b723a99e5dcd727" }, { "dataPath": "params_shard_153.bin", "format": "raw-shard", "nbytes": 25165824, "records": [ { "name": "vision_embed_tokens.img_projection.linear_1.weight", "shape": [ 3072, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 25165824, "byteOffset": 0 } ], "md5sum": "b52ad1d28d6d50e91e604896e7c1a3dd" }, { "dataPath": "params_shard_154.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "vision_embed_tokens.img_projection.linear_2.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "745579b9575b427cef55901285d42f30" }, { "dataPath": "params_shard_155.bin", "format": "raw-shard", "nbytes": 16814080, "records": [ { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.mlp.fc2.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8388608, "byteOffset": 0 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.self_attn.k_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 8388608 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.self_attn.k_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 8390656 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.self_attn.out_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 10487808 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.self_attn.out_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 10489856 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.self_attn.q_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 12587008 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.self_attn.q_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 12589056 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.self_attn.v_proj.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 14686208 }, { "name": "vision_embed_tokens.img_processor.vision_model.encoder.layers.9.self_attn.v_proj.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2097152, "byteOffset": 14688256 }, { "name": "vision_embed_tokens.img_processor.vision_model.post_layernorm.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16785408 }, { "name": "vision_embed_tokens.img_processor.vision_model.post_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16787456 }, { "name": "vision_embed_tokens.img_processor.vision_model.pre_layrnorm.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16789504 }, { "name": "vision_embed_tokens.img_processor.vision_model.pre_layrnorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 16791552 }, { "name": "vision_embed_tokens.img_projection.linear_1.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 16793600 }, { "name": "vision_embed_tokens.img_projection.linear_2.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 16799744 }, { "name": "vision_embed_tokens.sub_GN", "shape": [ 1, 1, 1, 4096 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 16805888 } ], "md5sum": "972b4880db9d084f3b3c73e3dbd68847" } ] }