{ "metadata": { "ParamSize": 480, "ParamBytes": 621610240.0, "BitsPerParam": 5.693452146956644 }, "records": [ { "dataPath": "params_shard_0.bin", "format": "raw-shard", "nbytes": 127139840, "records": [ { "name": "language_model.model.embed_tokens.q_weight", "shape": [ 248320, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 127139840, "byteOffset": 0 } ], "md5sum": "072b42975051fdb2012ff1bbaaca7b24" }, { "dataPath": "params_shard_1.bin", "format": "raw-shard", "nbytes": 18874368, "records": [ { "name": "visual.merger.fc1.weight", "shape": [ 3072, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 18874368, "byteOffset": 0 } ], "md5sum": "8aa81859a325849ce0e1902d0eee793f" }, { "dataPath": "params_shard_2.bin", "format": "raw-shard", "nbytes": 30103552, "records": [ { "name": "language_model.model.embed_tokens.q_scale", "shape": [ 248320, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15892480, "byteOffset": 0 }, { "name": "language_model.model.layers.14.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 15892480 }, { "name": "language_model.model.layers.14.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 20103168 }, { "name": "language_model.model.layers.12.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 20629504 }, { "name": "language_model.model.layers.12.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 24840192 }, { "name": "language_model.model.layers.18.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 25366528 }, { "name": "language_model.model.layers.18.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 29577216 } ], "md5sum": "0fe93be56f57b54750a8a126a8f21c6b" }, { "dataPath": "params_shard_3.bin", "format": "raw-shard", "nbytes": 33159168, "records": [ { "name": "language_model.model.layers.13.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 0 }, { "name": "language_model.model.layers.13.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 4210688 }, { "name": "language_model.model.layers.2.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 4737024 }, { "name": "language_model.model.layers.2.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 8947712 }, { "name": "language_model.model.layers.4.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 9474048 }, { "name": "language_model.model.layers.4.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 13684736 }, { "name": "language_model.model.layers.20.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 14211072 }, { "name": "language_model.model.layers.20.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 18421760 }, { "name": "language_model.model.layers.1.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 18948096 }, { "name": "language_model.model.layers.1.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 23158784 }, { "name": "language_model.model.layers.10.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 23685120 }, { "name": "language_model.model.layers.10.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 27895808 }, { "name": "language_model.model.layers.16.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 28422144 }, { "name": "language_model.model.layers.16.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 32632832 } ], "md5sum": "44f02752d4e7f3237005c12c08390c33" }, { "dataPath": "params_shard_4.bin", "format": "raw-shard", "nbytes": 33159168, "records": [ { "name": "language_model.model.layers.17.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 0 }, { "name": "language_model.model.layers.17.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 4210688 }, { "name": "language_model.model.layers.5.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 4737024 }, { "name": "language_model.model.layers.5.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 8947712 }, { "name": "language_model.model.layers.6.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 9474048 }, { "name": "language_model.model.layers.6.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 13684736 }, { "name": "language_model.model.layers.8.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 14211072 }, { "name": "language_model.model.layers.8.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 18421760 }, { "name": "language_model.model.layers.9.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 18948096 }, { "name": "language_model.model.layers.9.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 23158784 }, { "name": "language_model.model.layers.21.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 23685120 }, { "name": "language_model.model.layers.21.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 27895808 }, { "name": "language_model.model.layers.22.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 28422144 }, { "name": "language_model.model.layers.22.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 32632832 } ], "md5sum": "a65ca3d15926417f2a49043c89e2d6fe" }, { "dataPath": "params_shard_5.bin", "format": "raw-shard", "nbytes": 30689280, "records": [ { "name": "language_model.model.layers.0.linear_attn.in_proj_qzab.q_weight", "shape": [ 8224, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 4210688, "byteOffset": 0 }, { "name": "language_model.model.layers.0.linear_attn.in_proj_qzab.q_scale", "shape": [ 8224, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 526336, "byteOffset": 4210688 }, { "name": "language_model.model.layers.23.self_attn.c_attn.q_weight", "shape": [ 5120, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 2621440, "byteOffset": 4737024 }, { "name": "language_model.model.layers.23.self_attn.c_attn.q_scale", "shape": [ 5120, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 327680, "byteOffset": 7358464 }, { "name": "language_model.model.layers.15.self_attn.c_attn.q_weight", "shape": [ 5120, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 2621440, "byteOffset": 7686144 }, { "name": "language_model.model.layers.15.self_attn.c_attn.q_scale", "shape": [ 5120, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 327680, "byteOffset": 10307584 }, { "name": "language_model.model.layers.7.self_attn.c_attn.q_weight", "shape": [ 5120, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 2621440, "byteOffset": 10635264 }, { "name": "language_model.model.layers.7.self_attn.c_attn.q_scale", "shape": [ 5120, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 327680, "byteOffset": 13256704 }, { "name": "language_model.model.layers.11.self_attn.c_attn.q_weight", "shape": [ 5120, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 2621440, "byteOffset": 13584384 }, { "name": "language_model.model.layers.11.self_attn.c_attn.q_scale", "shape": [ 5120, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 327680, "byteOffset": 16205824 }, { "name": "language_model.model.layers.19.self_attn.c_attn.q_weight", "shape": [ 5120, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 2621440, "byteOffset": 16533504 }, { "name": "language_model.model.layers.19.self_attn.c_attn.q_scale", "shape": [ 5120, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 327680, "byteOffset": 19154944 }, { "name": "language_model.model.layers.3.self_attn.c_attn.q_weight", "shape": [ 5120, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 2621440, "byteOffset": 19482624 }, { "name": "language_model.model.layers.3.self_attn.c_attn.q_scale", "shape": [ 5120, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 327680, "byteOffset": 22104064 }, { "name": "language_model.model.layers.22.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 22431744 }, { "name": "language_model.model.layers.22.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 24266752 }, { "name": "language_model.model.layers.22.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 24496128 }, { "name": "language_model.model.layers.22.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 28166144 }, { "name": "language_model.model.layers.23.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 28624896 }, { "name": "language_model.model.layers.23.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 30459904 } ], "md5sum": "189a3417609526f533949c1b946c983c" }, { "dataPath": "params_shard_6.bin", "format": "raw-shard", "nbytes": 30965760, "records": [ { "name": "language_model.model.layers.23.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 0 }, { "name": "language_model.model.layers.23.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 3670016 }, { "name": "language_model.model.layers.3.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 4128768 }, { "name": "language_model.model.layers.3.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 5963776 }, { "name": "language_model.model.layers.3.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 6193152 }, { "name": "language_model.model.layers.3.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 9863168 }, { "name": "language_model.model.layers.14.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 10321920 }, { "name": "language_model.model.layers.14.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 12156928 }, { "name": "language_model.model.layers.14.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 12386304 }, { "name": "language_model.model.layers.14.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 16056320 }, { "name": "language_model.model.layers.15.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 16515072 }, { "name": "language_model.model.layers.15.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 18350080 }, { "name": "language_model.model.layers.15.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 18579456 }, { "name": "language_model.model.layers.15.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 22249472 }, { "name": "language_model.model.layers.6.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 22708224 }, { "name": "language_model.model.layers.6.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 24543232 }, { "name": "language_model.model.layers.6.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 24772608 }, { "name": "language_model.model.layers.6.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 28442624 }, { "name": "language_model.model.layers.7.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 28901376 }, { "name": "language_model.model.layers.7.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 30736384 } ], "md5sum": "f3c14d97eb2d253cbe2a26476f145622" }, { "dataPath": "params_shard_7.bin", "format": "raw-shard", "nbytes": 33030144, "records": [ { "name": "language_model.model.layers.7.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 0 }, { "name": "language_model.model.layers.7.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 3670016 }, { "name": "language_model.model.layers.10.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 4128768 }, { "name": "language_model.model.layers.10.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 5963776 }, { "name": "language_model.model.layers.11.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 6193152 }, { "name": "language_model.model.layers.11.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 8028160 }, { "name": "language_model.model.layers.11.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 8257536 }, { "name": "language_model.model.layers.11.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 11927552 }, { "name": "language_model.model.layers.9.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 12386304 }, { "name": "language_model.model.layers.9.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 14221312 }, { "name": "language_model.model.layers.9.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 14450688 }, { "name": "language_model.model.layers.9.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 18120704 }, { "name": "language_model.model.layers.17.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 18579456 }, { "name": "language_model.model.layers.17.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 20414464 }, { "name": "language_model.model.layers.17.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 20643840 }, { "name": "language_model.model.layers.17.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 24313856 }, { "name": "language_model.model.layers.18.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 24772608 }, { "name": "language_model.model.layers.18.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 26607616 }, { "name": "language_model.model.layers.18.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 26836992 }, { "name": "language_model.model.layers.18.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 30507008 }, { "name": "language_model.model.layers.12.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 30965760 }, { "name": "language_model.model.layers.12.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 32800768 } ], "md5sum": "5d04bee292ce2ce525f23eb2d28a674f" }, { "dataPath": "params_shard_8.bin", "format": "raw-shard", "nbytes": 30965760, "records": [ { "name": "language_model.model.layers.12.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 0 }, { "name": "language_model.model.layers.12.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 3670016 }, { "name": "language_model.model.layers.13.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 4128768 }, { "name": "language_model.model.layers.13.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 5963776 }, { "name": "language_model.model.layers.13.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 6193152 }, { "name": "language_model.model.layers.13.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 9863168 }, { "name": "language_model.model.layers.19.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 10321920 }, { "name": "language_model.model.layers.19.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 12156928 }, { "name": "language_model.model.layers.19.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 12386304 }, { "name": "language_model.model.layers.19.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 16056320 }, { "name": "language_model.model.layers.4.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 16515072 }, { "name": "language_model.model.layers.4.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 18350080 }, { "name": "language_model.model.layers.4.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 18579456 }, { "name": "language_model.model.layers.4.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 22249472 }, { "name": "language_model.model.layers.2.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 22708224 }, { "name": "language_model.model.layers.2.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 24543232 }, { "name": "language_model.model.layers.2.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 24772608 }, { "name": "language_model.model.layers.2.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 28442624 }, { "name": "language_model.model.layers.20.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 28901376 }, { "name": "language_model.model.layers.20.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 30736384 } ], "md5sum": "49b8d8ce66fa9986f6329cc4db5351c6" }, { "dataPath": "params_shard_9.bin", "format": "raw-shard", "nbytes": 33030144, "records": [ { "name": "language_model.model.layers.20.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 0 }, { "name": "language_model.model.layers.20.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 3670016 }, { "name": "language_model.model.layers.1.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 4128768 }, { "name": "language_model.model.layers.1.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 5963776 }, { "name": "language_model.model.layers.1.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 6193152 }, { "name": "language_model.model.layers.1.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 9863168 }, { "name": "language_model.model.layers.10.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 10321920 }, { "name": "language_model.model.layers.10.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 13991936 }, { "name": "language_model.model.layers.16.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 14450688 }, { "name": "language_model.model.layers.16.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 16285696 }, { "name": "language_model.model.layers.16.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 16515072 }, { "name": "language_model.model.layers.16.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 20185088 }, { "name": "language_model.model.layers.5.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 20643840 }, { "name": "language_model.model.layers.5.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 22478848 }, { "name": "language_model.model.layers.5.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 22708224 }, { "name": "language_model.model.layers.5.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 26378240 }, { "name": "language_model.model.layers.8.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 26836992 }, { "name": "language_model.model.layers.8.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 28672000 }, { "name": "language_model.model.layers.8.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 28901376 }, { "name": "language_model.model.layers.8.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 32571392 } ], "md5sum": "cd3cc6cb0d236dc2238ea1d62f9bb5ed" }, { "dataPath": "params_shard_10.bin", "format": "raw-shard", "nbytes": 32833536, "records": [ { "name": "language_model.model.layers.21.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 0 }, { "name": "language_model.model.layers.21.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 1835008 }, { "name": "language_model.model.layers.21.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 2064384 }, { "name": "language_model.model.layers.21.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 5734400 }, { "name": "language_model.model.layers.0.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 6193152 }, { "name": "language_model.model.layers.0.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 8028160 }, { "name": "language_model.model.layers.0.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 8257536 }, { "name": "language_model.model.layers.0.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 11927552 }, { "name": "visual.merger.fc2.weight", "shape": [ 1024, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6291456, "byteOffset": 12386304 }, { "name": "visual.blocks.0.mlp.fc1.weight", "shape": [ 3072, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 18677760 }, { "name": "visual.blocks.0.mlp.fc2.weight", "shape": [ 768, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 23396352 }, { "name": "visual.blocks.1.mlp.fc1.weight", "shape": [ 3072, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 28114944 } ], "md5sum": "1bfa7e1ff088c06dc118c4fc07763af5" }, { "dataPath": "params_shard_11.bin", "format": "raw-shard", "nbytes": 33030144, "records": [ { "name": "visual.blocks.1.mlp.fc2.weight", "shape": [ 768, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 0 }, { "name": "visual.blocks.10.mlp.fc1.weight", "shape": [ 3072, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 4718592 }, { "name": "visual.blocks.10.mlp.fc2.weight", "shape": [ 768, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 9437184 }, { "name": "visual.blocks.11.mlp.fc1.weight", "shape": [ 3072, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 14155776 }, { "name": "visual.blocks.11.mlp.fc2.weight", "shape": [ 768, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 18874368 }, { "name": "visual.blocks.2.mlp.fc1.weight", "shape": [ 3072, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 23592960 }, { "name": "visual.blocks.2.mlp.fc2.weight", "shape": [ 768, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 28311552 } ], "md5sum": "c4b491048843146e209539ae26e4c3dd" }, { "dataPath": "params_shard_12.bin", "format": "raw-shard", "nbytes": 33030144, "records": [ { "name": "visual.blocks.3.mlp.fc1.weight", "shape": [ 3072, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 0 }, { "name": "visual.blocks.3.mlp.fc2.weight", "shape": [ 768, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 4718592 }, { "name": "visual.blocks.4.mlp.fc1.weight", "shape": [ 3072, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 9437184 }, { "name": "visual.blocks.4.mlp.fc2.weight", "shape": [ 768, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 14155776 }, { "name": "visual.blocks.5.mlp.fc1.weight", "shape": [ 3072, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 18874368 }, { "name": "visual.blocks.5.mlp.fc2.weight", "shape": [ 768, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 23592960 }, { "name": "visual.blocks.6.mlp.fc1.weight", "shape": [ 3072, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 28311552 } ], "md5sum": "dd4a95252f792de20f9b024693bb2b83" }, { "dataPath": "params_shard_13.bin", "format": "raw-shard", "nbytes": 33030144, "records": [ { "name": "visual.blocks.6.mlp.fc2.weight", "shape": [ 768, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 0 }, { "name": "visual.blocks.7.mlp.fc1.weight", "shape": [ 3072, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 4718592 }, { "name": "visual.blocks.7.mlp.fc2.weight", "shape": [ 768, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 9437184 }, { "name": "visual.blocks.8.mlp.fc1.weight", "shape": [ 3072, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 14155776 }, { "name": "visual.blocks.8.mlp.fc2.weight", "shape": [ 768, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 18874368 }, { "name": "visual.blocks.9.mlp.fc1.weight", "shape": [ 3072, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 23592960 }, { "name": "visual.blocks.9.mlp.fc2.weight", "shape": [ 768, 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4718592, "byteOffset": 28311552 } ], "md5sum": "d90fffcf69966b41e29c245806e3015c" }, { "dataPath": "params_shard_14.bin", "format": "raw-shard", "nbytes": 31850496, "records": [ { "name": "language_model.model.layers.23.self_attn.o_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 0 }, { "name": "language_model.model.layers.23.self_attn.o_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 1048576 }, { "name": "language_model.model.layers.3.self_attn.o_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 1179648 }, { "name": "language_model.model.layers.3.self_attn.o_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 2228224 }, { "name": "language_model.model.layers.14.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 2359296 }, { "name": "language_model.model.layers.14.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 3407872 }, { "name": "language_model.model.layers.15.self_attn.o_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 3538944 }, { "name": "language_model.model.layers.15.self_attn.o_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 4587520 }, { "name": "language_model.model.layers.6.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 4718592 }, { "name": "language_model.model.layers.6.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 5767168 }, { "name": "language_model.model.layers.7.self_attn.o_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 5898240 }, { "name": "language_model.model.layers.7.self_attn.o_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 6946816 }, { "name": "language_model.model.layers.11.self_attn.o_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 7077888 }, { "name": "language_model.model.layers.11.self_attn.o_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 8126464 }, { "name": "language_model.model.layers.12.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 8257536 }, { "name": "language_model.model.layers.12.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 9306112 }, { "name": "language_model.model.layers.17.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 9437184 }, { "name": "language_model.model.layers.17.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 10485760 }, { "name": "language_model.model.layers.18.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 10616832 }, { "name": "language_model.model.layers.18.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 11665408 }, { "name": "language_model.model.layers.13.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 11796480 }, { "name": "language_model.model.layers.13.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 12845056 }, { "name": "language_model.model.layers.19.self_attn.o_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 12976128 }, { "name": "language_model.model.layers.19.self_attn.o_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 14024704 }, { "name": "language_model.model.layers.2.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 14155776 }, { "name": "language_model.model.layers.2.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 15204352 }, { "name": "language_model.model.layers.4.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 15335424 }, { "name": "language_model.model.layers.4.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 16384000 }, { "name": "language_model.model.layers.20.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 16515072 }, { "name": "language_model.model.layers.20.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 17563648 }, { "name": "language_model.model.layers.1.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 17694720 }, { "name": "language_model.model.layers.1.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 18743296 }, { "name": "language_model.model.layers.10.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 18874368 }, { "name": "language_model.model.layers.10.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 19922944 }, { "name": "language_model.model.layers.16.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 20054016 }, { "name": "language_model.model.layers.16.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 21102592 }, { "name": "language_model.model.layers.5.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 21233664 }, { "name": "language_model.model.layers.5.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 22282240 }, { "name": "language_model.model.layers.8.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 22413312 }, { "name": "language_model.model.layers.8.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 23461888 }, { "name": "language_model.model.layers.9.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 23592960 }, { "name": "language_model.model.layers.9.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 24641536 }, { "name": "language_model.model.layers.21.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 24772608 }, { "name": "language_model.model.layers.21.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 25821184 }, { "name": "language_model.model.layers.22.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 25952256 }, { "name": "language_model.model.layers.22.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 27000832 }, { "name": "language_model.model.layers.0.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 27131904 }, { "name": "language_model.model.layers.0.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 28180480 }, { "name": "visual.blocks.0.attn.qkv.weight", "shape": [ 2304, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 3538944, "byteOffset": 28311552 } ], "md5sum": "f261436ac6d520268fd424883bd337c0" }, { "dataPath": "params_shard_15.bin", "format": "raw-shard", "nbytes": 31850496, "records": [ { "name": "visual.blocks.1.attn.qkv.weight", "shape": [ 2304, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 3538944, "byteOffset": 0 }, { "name": "visual.blocks.10.attn.qkv.weight", "shape": [ 2304, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 3538944, "byteOffset": 3538944 }, { "name": "visual.blocks.11.attn.qkv.weight", "shape": [ 2304, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 3538944, "byteOffset": 7077888 }, { "name": "visual.blocks.2.attn.qkv.weight", "shape": [ 2304, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 3538944, "byteOffset": 10616832 }, { "name": "visual.blocks.3.attn.qkv.weight", "shape": [ 2304, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 3538944, "byteOffset": 14155776 }, { "name": "visual.blocks.4.attn.qkv.weight", "shape": [ 2304, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 3538944, "byteOffset": 17694720 }, { "name": "visual.blocks.5.attn.qkv.weight", "shape": [ 2304, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 3538944, "byteOffset": 21233664 }, { "name": "visual.blocks.6.attn.qkv.weight", "shape": [ 2304, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 3538944, "byteOffset": 24772608 }, { "name": "visual.blocks.7.attn.qkv.weight", "shape": [ 2304, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 3538944, "byteOffset": 28311552 } ], "md5sum": "a79065b5141b31920df97083ce363986" }, { "dataPath": "params_shard_16.bin", "format": "raw-shard", "nbytes": 24866944, "records": [ { "name": "visual.blocks.8.attn.qkv.weight", "shape": [ 2304, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 3538944, "byteOffset": 0 }, { "name": "visual.blocks.9.attn.qkv.weight", "shape": [ 2304, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 3538944, "byteOffset": 3538944 }, { "name": "visual.pos_embed", "shape": [ 784, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1204224, "byteOffset": 7077888 }, { "name": "visual.patch_embed.proj.weight", "shape": [ 768, 3, 16, 16 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1179648, "byteOffset": 8282112 }, { "name": "visual.blocks.0.attn.proj.weight", "shape": [ 768, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1179648, "byteOffset": 9461760 }, { "name": "visual.blocks.1.attn.proj.weight", "shape": [ 768, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1179648, "byteOffset": 10641408 }, { "name": "visual.blocks.10.attn.proj.weight", "shape": [ 768, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1179648, "byteOffset": 11821056 }, { "name": "visual.blocks.11.attn.proj.weight", "shape": [ 768, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1179648, "byteOffset": 13000704 }, { "name": "visual.blocks.2.attn.proj.weight", "shape": [ 768, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1179648, "byteOffset": 14180352 }, { "name": "visual.blocks.3.attn.proj.weight", "shape": [ 768, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1179648, "byteOffset": 15360000 }, { "name": "visual.blocks.4.attn.proj.weight", "shape": [ 768, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1179648, "byteOffset": 16539648 }, { "name": "visual.blocks.5.attn.proj.weight", "shape": [ 768, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1179648, "byteOffset": 17719296 }, { "name": "visual.blocks.6.attn.proj.weight", "shape": [ 768, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1179648, "byteOffset": 18898944 }, { "name": "visual.blocks.7.attn.proj.weight", "shape": [ 768, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1179648, "byteOffset": 20078592 }, { "name": "visual.blocks.8.attn.proj.weight", "shape": [ 768, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1179648, "byteOffset": 21258240 }, { "name": "visual.blocks.9.attn.proj.weight", "shape": [ 768, 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1179648, "byteOffset": 22437888 }, { "name": "language_model.model.layers.8.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 23617536 }, { "name": "language_model.model.layers.12.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 23666688 }, { "name": "language_model.model.layers.18.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 23715840 }, { "name": "language_model.model.layers.13.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 23764992 }, { "name": "language_model.model.layers.14.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 23814144 }, { "name": "language_model.model.layers.2.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 23863296 }, { "name": "language_model.model.layers.4.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 23912448 }, { "name": "language_model.model.layers.5.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 23961600 }, { "name": "language_model.model.layers.20.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 24010752 }, { "name": "language_model.model.layers.21.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 24059904 }, { "name": "language_model.model.layers.10.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 24109056 }, { "name": "language_model.model.layers.16.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 24158208 }, { "name": "language_model.model.layers.17.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 24207360 }, { "name": "language_model.model.layers.6.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 24256512 }, { "name": "language_model.model.layers.9.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 24305664 }, { "name": "language_model.model.layers.22.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 24354816 }, { "name": "language_model.model.layers.0.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 24403968 }, { "name": "language_model.model.layers.1.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 24453120 }, { "name": "visual.blocks.0.mlp.fc1.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 24502272 }, { "name": "visual.blocks.1.mlp.fc1.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 24508416 }, { "name": "visual.blocks.10.mlp.fc1.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 24514560 }, { "name": "visual.blocks.11.mlp.fc1.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 24520704 }, { "name": "visual.blocks.2.mlp.fc1.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 24526848 }, { "name": "visual.blocks.3.mlp.fc1.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 24532992 }, { "name": "visual.blocks.4.mlp.fc1.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 24539136 }, { "name": "visual.blocks.5.mlp.fc1.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 24545280 }, { "name": "visual.blocks.6.mlp.fc1.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 24551424 }, { "name": "visual.blocks.7.mlp.fc1.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 24557568 }, { "name": "visual.blocks.8.mlp.fc1.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 24563712 }, { "name": "visual.blocks.9.mlp.fc1.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 24569856 }, { "name": "visual.merger.fc1.bias", "shape": [ 3072 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 6144, "byteOffset": 24576000 }, { "name": "visual.blocks.0.attn.qkv.bias", "shape": [ 2304 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 24582144 }, { "name": "visual.blocks.1.attn.qkv.bias", "shape": [ 2304 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 24586752 }, { "name": "visual.blocks.10.attn.qkv.bias", "shape": [ 2304 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 24591360 }, { "name": "visual.blocks.11.attn.qkv.bias", "shape": [ 2304 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 24595968 }, { "name": "visual.blocks.2.attn.qkv.bias", "shape": [ 2304 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 24600576 }, { "name": "visual.blocks.3.attn.qkv.bias", "shape": [ 2304 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 24605184 }, { "name": "visual.blocks.4.attn.qkv.bias", "shape": [ 2304 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 24609792 }, { "name": "visual.blocks.5.attn.qkv.bias", "shape": [ 2304 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 24614400 }, { "name": "visual.blocks.6.attn.qkv.bias", "shape": [ 2304 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 24619008 }, { "name": "visual.blocks.7.attn.qkv.bias", "shape": [ 2304 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 24623616 }, { "name": "visual.blocks.8.attn.qkv.bias", "shape": [ 2304 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 24628224 }, { "name": "visual.blocks.9.attn.qkv.bias", "shape": [ 2304 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 4608, "byteOffset": 24632832 }, { "name": "language_model.model.layers.22.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24637440 }, { "name": "language_model.model.layers.23.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24639488 }, { "name": "language_model.model.layers.23.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24641536 }, { "name": "language_model.model.layers.3.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24643584 }, { "name": "language_model.model.layers.14.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24645632 }, { "name": "language_model.model.layers.14.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24647680 }, { "name": "language_model.model.layers.15.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24649728 }, { "name": "language_model.model.layers.15.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24651776 }, { "name": "language_model.model.layers.6.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24653824 }, { "name": "language_model.model.layers.7.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24655872 }, { "name": "language_model.model.layers.7.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24657920 }, { "name": "language_model.model.layers.8.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24659968 }, { "name": "language_model.model.layers.11.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24662016 }, { "name": "language_model.model.layers.11.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24664064 }, { "name": "language_model.model.layers.12.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24666112 }, { "name": "language_model.model.layers.17.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24668160 }, { "name": "language_model.model.layers.18.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24670208 }, { "name": "language_model.model.layers.18.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24672256 }, { "name": "language_model.model.layers.19.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24674304 }, { "name": "language_model.model.layers.12.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24676352 }, { "name": "language_model.model.layers.13.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24678400 }, { "name": "language_model.model.layers.13.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24680448 }, { "name": "language_model.model.layers.19.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24682496 }, { "name": "language_model.model.layers.2.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24684544 }, { "name": "language_model.model.layers.2.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24686592 }, { "name": "language_model.model.layers.3.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24688640 }, { "name": "language_model.model.layers.4.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24690688 }, { "name": "language_model.model.layers.4.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24692736 }, { "name": "language_model.model.layers.20.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24694784 }, { "name": "language_model.model.layers.20.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24696832 }, { "name": "language_model.model.layers.1.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24698880 }, { "name": "language_model.model.layers.10.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24700928 }, { "name": "language_model.model.layers.10.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24702976 }, { "name": "language_model.model.layers.16.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24705024 }, { "name": "language_model.model.layers.16.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24707072 }, { "name": "language_model.model.layers.17.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24709120 }, { "name": "language_model.model.layers.5.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24711168 }, { "name": "language_model.model.layers.5.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24713216 }, { "name": "language_model.model.layers.6.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24715264 }, { "name": "language_model.model.layers.8.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24717312 }, { "name": "language_model.model.layers.9.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24719360 }, { "name": "language_model.model.layers.9.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24721408 }, { "name": "language_model.model.layers.21.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24723456 }, { "name": "language_model.model.layers.21.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24725504 }, { "name": "language_model.model.layers.22.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24727552 }, { "name": "language_model.model.layers.0.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24729600 }, { "name": "language_model.model.layers.0.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24731648 }, { "name": "language_model.model.layers.1.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24733696 }, { "name": "language_model.model.norm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24735744 }, { "name": "visual.merger.fc2.bias", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 24737792 }, { "name": "visual.blocks.0.attn.proj.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24739840 }, { "name": "visual.blocks.0.mlp.fc2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24741376 }, { "name": "visual.blocks.0.norm1.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24742912 }, { "name": "visual.blocks.0.norm1.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24744448 }, { "name": "visual.blocks.0.norm2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24745984 }, { "name": "visual.blocks.0.norm2.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24747520 }, { "name": "visual.blocks.1.attn.proj.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24749056 }, { "name": "visual.blocks.1.mlp.fc2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24750592 }, { "name": "visual.blocks.1.norm1.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24752128 }, { "name": "visual.blocks.1.norm1.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24753664 }, { "name": "visual.blocks.1.norm2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24755200 }, { "name": "visual.blocks.1.norm2.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24756736 }, { "name": "visual.blocks.10.attn.proj.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24758272 }, { "name": "visual.blocks.10.mlp.fc2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24759808 }, { "name": "visual.blocks.10.norm1.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24761344 }, { "name": "visual.blocks.10.norm1.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24762880 }, { "name": "visual.blocks.10.norm2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24764416 }, { "name": "visual.blocks.10.norm2.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24765952 }, { "name": "visual.blocks.11.attn.proj.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24767488 }, { "name": "visual.blocks.11.mlp.fc2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24769024 }, { "name": "visual.blocks.11.norm1.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24770560 }, { "name": "visual.blocks.11.norm1.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24772096 }, { "name": "visual.blocks.11.norm2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24773632 }, { "name": "visual.blocks.11.norm2.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24775168 }, { "name": "visual.blocks.2.attn.proj.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24776704 }, { "name": "visual.blocks.2.mlp.fc2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24778240 }, { "name": "visual.blocks.2.norm1.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24779776 }, { "name": "visual.blocks.2.norm1.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24781312 }, { "name": "visual.blocks.2.norm2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24782848 }, { "name": "visual.blocks.2.norm2.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24784384 }, { "name": "visual.blocks.3.attn.proj.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24785920 }, { "name": "visual.blocks.3.mlp.fc2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24787456 }, { "name": "visual.blocks.3.norm1.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24788992 }, { "name": "visual.blocks.3.norm1.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24790528 }, { "name": "visual.blocks.3.norm2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24792064 }, { "name": "visual.blocks.3.norm2.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24793600 }, { "name": "visual.blocks.4.attn.proj.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24795136 }, { "name": "visual.blocks.4.mlp.fc2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24796672 }, { "name": "visual.blocks.4.norm1.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24798208 }, { "name": "visual.blocks.4.norm1.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24799744 }, { "name": "visual.blocks.4.norm2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24801280 }, { "name": "visual.blocks.4.norm2.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24802816 }, { "name": "visual.blocks.5.attn.proj.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24804352 }, { "name": "visual.blocks.5.mlp.fc2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24805888 }, { "name": "visual.blocks.5.norm1.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24807424 }, { "name": "visual.blocks.5.norm1.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24808960 }, { "name": "visual.blocks.5.norm2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24810496 }, { "name": "visual.blocks.5.norm2.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24812032 }, { "name": "visual.blocks.6.attn.proj.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24813568 }, { "name": "visual.blocks.6.mlp.fc2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24815104 }, { "name": "visual.blocks.6.norm1.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24816640 }, { "name": "visual.blocks.6.norm1.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24818176 }, { "name": "visual.blocks.6.norm2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24819712 }, { "name": "visual.blocks.6.norm2.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24821248 }, { "name": "visual.blocks.7.attn.proj.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24822784 }, { "name": "visual.blocks.7.mlp.fc2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24824320 }, { "name": "visual.blocks.7.norm1.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24825856 }, { "name": "visual.blocks.7.norm1.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24827392 }, { "name": "visual.blocks.7.norm2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24828928 }, { "name": "visual.blocks.7.norm2.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24830464 }, { "name": "visual.blocks.8.attn.proj.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24832000 }, { "name": "visual.blocks.8.mlp.fc2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24833536 }, { "name": "visual.blocks.8.norm1.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24835072 }, { "name": "visual.blocks.8.norm1.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24836608 }, { "name": "visual.blocks.8.norm2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24838144 }, { "name": "visual.blocks.8.norm2.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24839680 }, { "name": "visual.blocks.9.attn.proj.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24841216 }, { "name": "visual.blocks.9.mlp.fc2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24842752 }, { "name": "visual.blocks.9.norm1.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24844288 }, { "name": "visual.blocks.9.norm1.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24845824 }, { "name": "visual.blocks.9.norm2.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24847360 }, { "name": "visual.blocks.9.norm2.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24848896 }, { "name": "visual.merger.norm.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24850432 }, { "name": "visual.merger.norm.weight", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24851968 }, { "name": "visual.patch_embed.proj.bias", "shape": [ 768 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1536, "byteOffset": 24853504 }, { "name": "language_model.model.layers.23.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 24855040 }, { "name": "language_model.model.layers.23.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 24855552 }, { "name": "language_model.model.layers.3.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 24856064 }, { "name": "language_model.model.layers.14.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24856576 }, { "name": "language_model.model.layers.15.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 24856832 }, { "name": "language_model.model.layers.15.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 24857344 }, { "name": "language_model.model.layers.7.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 24857856 }, { "name": "language_model.model.layers.7.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 24858368 }, { "name": "language_model.model.layers.11.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 24858880 }, { "name": "language_model.model.layers.11.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 24859392 }, { "name": "language_model.model.layers.12.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24859904 }, { "name": "language_model.model.layers.18.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24860160 }, { "name": "language_model.model.layers.13.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24860416 }, { "name": "language_model.model.layers.19.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 24860672 }, { "name": "language_model.model.layers.19.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 24861184 }, { "name": "language_model.model.layers.2.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24861696 }, { "name": "language_model.model.layers.3.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 24861952 }, { "name": "language_model.model.layers.4.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24862464 }, { "name": "language_model.model.layers.20.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24862720 }, { "name": "language_model.model.layers.1.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24862976 }, { "name": "language_model.model.layers.10.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24863232 }, { "name": "language_model.model.layers.16.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24863488 }, { "name": "language_model.model.layers.17.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24863744 }, { "name": "language_model.model.layers.5.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24864000 }, { "name": "language_model.model.layers.6.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24864256 }, { "name": "language_model.model.layers.8.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24864512 }, { "name": "language_model.model.layers.9.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24864768 }, { "name": "language_model.model.layers.21.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24865024 }, { "name": "language_model.model.layers.22.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24865280 }, { "name": "language_model.model.layers.0.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 24865536 }, { "name": "language_model.model.layers.16.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24865792 }, { "name": "language_model.model.layers.8.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24865824 }, { "name": "language_model.model.layers.12.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24865856 }, { "name": "language_model.model.layers.18.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24865888 }, { "name": "language_model.model.layers.13.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24865920 }, { "name": "language_model.model.layers.14.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24865952 }, { "name": "language_model.model.layers.2.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24865984 }, { "name": "language_model.model.layers.4.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866016 }, { "name": "language_model.model.layers.5.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866048 }, { "name": "language_model.model.layers.20.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866080 }, { "name": "language_model.model.layers.21.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866112 }, { "name": "language_model.model.layers.10.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866144 }, { "name": "language_model.model.layers.17.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866176 }, { "name": "language_model.model.layers.6.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866208 }, { "name": "language_model.model.layers.9.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866240 }, { "name": "language_model.model.layers.22.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866272 }, { "name": "language_model.model.layers.0.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866304 }, { "name": "language_model.model.layers.1.linear_attn.A_log", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866336 }, { "name": "language_model.model.layers.8.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866368 }, { "name": "language_model.model.layers.12.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866400 }, { "name": "language_model.model.layers.18.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866432 }, { "name": "language_model.model.layers.13.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866464 }, { "name": "language_model.model.layers.14.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866496 }, { "name": "language_model.model.layers.2.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866528 }, { "name": "language_model.model.layers.4.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866560 }, { "name": "language_model.model.layers.20.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866592 }, { "name": "language_model.model.layers.21.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866624 }, { "name": "language_model.model.layers.10.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866656 }, { "name": "language_model.model.layers.16.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866688 }, { "name": "language_model.model.layers.17.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866720 }, { "name": "language_model.model.layers.5.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866752 }, { "name": "language_model.model.layers.6.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866784 }, { "name": "language_model.model.layers.9.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866816 }, { "name": "language_model.model.layers.22.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866848 }, { "name": "language_model.model.layers.0.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866880 }, { "name": "language_model.model.layers.1.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "bfloat16", "format": "raw", "nbytes": 32, "byteOffset": 24866912 } ], "md5sum": "697e3d14152a3634e917b90528ff4e26" } ] }