{ "metadata": { "ParamSize": 435, "ParamBytes": 423938816.0, "BitsPerParam": 3.8829401557694054 }, "records": [ { "dataPath": "params_shard_0.bin", "format": "raw-shard", "nbytes": 127139840, "records": [ { "name": "model.embed_tokens.q_weight", "shape": [ 248320, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 127139840, "byteOffset": 0 } ], "md5sum": "ce4f3be3c1676c1e8a5baad97445fe99" }, { "dataPath": "params_shard_1.bin", "format": "raw-shard", "nbytes": 33193984, "records": [ { "name": "model.embed_tokens.q_scale", "shape": [ 248320, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15892480, "byteOffset": 0 }, { "name": "model.layers.14.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 15892480 }, { "name": "model.layers.14.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 19038208 }, { "name": "model.layers.12.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 19431424 }, { "name": "model.layers.12.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 22577152 }, { "name": "model.layers.18.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 22970368 }, { "name": "model.layers.18.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 26116096 }, { "name": "model.layers.13.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 26509312 }, { "name": "model.layers.13.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 29655040 }, { "name": "model.layers.2.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 30048256 } ], "md5sum": "2e2be8b0f1679f80f8860a1780c94cba" }, { "dataPath": "params_shard_2.bin", "format": "raw-shard", "nbytes": 32243712, "records": [ { "name": "model.layers.2.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 0 }, { "name": "model.layers.4.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 393216 }, { "name": "model.layers.4.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 3538944 }, { "name": "model.layers.20.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 3932160 }, { "name": "model.layers.20.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 7077888 }, { "name": "model.layers.1.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 7471104 }, { "name": "model.layers.1.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 10616832 }, { "name": "model.layers.10.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 11010048 }, { "name": "model.layers.10.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 14155776 }, { "name": "model.layers.16.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 14548992 }, { "name": "model.layers.16.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 17694720 }, { "name": "model.layers.17.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 18087936 }, { "name": "model.layers.17.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 21233664 }, { "name": "model.layers.5.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 21626880 }, { "name": "model.layers.5.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 24772608 }, { "name": "model.layers.6.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 25165824 }, { "name": "model.layers.6.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 28311552 }, { "name": "model.layers.8.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 28704768 }, { "name": "model.layers.8.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 31850496 } ], "md5sum": "ba01a4bb2e39970504a981742ec47e48" }, { "dataPath": "params_shard_3.bin", "format": "raw-shard", "nbytes": 31850496, "records": [ { "name": "model.layers.9.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 0 }, { "name": "model.layers.9.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 3145728 }, { "name": "model.layers.21.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 3538944 }, { "name": "model.layers.21.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 6684672 }, { "name": "model.layers.22.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 7077888 }, { "name": "model.layers.22.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 10223616 }, { "name": "model.layers.0.linear_attn.in_proj_qkv.q_weight", "shape": [ 6144, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3145728, "byteOffset": 10616832 }, { "name": "model.layers.0.linear_attn.in_proj_qkv.q_scale", "shape": [ 6144, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 393216, "byteOffset": 13762560 }, { "name": "model.layers.23.self_attn.c_attn.q_weight", "shape": [ 5120, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 2621440, "byteOffset": 14155776 }, { "name": "model.layers.23.self_attn.c_attn.q_scale", "shape": [ 5120, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 327680, "byteOffset": 16777216 }, { "name": "model.layers.15.self_attn.c_attn.q_weight", "shape": [ 5120, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 2621440, "byteOffset": 17104896 }, { "name": "model.layers.15.self_attn.c_attn.q_scale", "shape": [ 5120, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 327680, "byteOffset": 19726336 }, { "name": "model.layers.7.self_attn.c_attn.q_weight", "shape": [ 5120, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 2621440, "byteOffset": 20054016 }, { "name": "model.layers.7.self_attn.c_attn.q_scale", "shape": [ 5120, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 327680, "byteOffset": 22675456 }, { "name": "model.layers.11.self_attn.c_attn.q_weight", "shape": [ 5120, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 2621440, "byteOffset": 23003136 }, { "name": "model.layers.11.self_attn.c_attn.q_scale", "shape": [ 5120, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 327680, "byteOffset": 25624576 }, { "name": "model.layers.19.self_attn.c_attn.q_weight", "shape": [ 5120, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 2621440, "byteOffset": 25952256 }, { "name": "model.layers.19.self_attn.c_attn.q_scale", "shape": [ 5120, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 327680, "byteOffset": 28573696 }, { "name": "model.layers.3.self_attn.c_attn.q_weight", "shape": [ 5120, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 2621440, "byteOffset": 28901376 }, { "name": "model.layers.3.self_attn.c_attn.q_scale", "shape": [ 5120, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 327680, "byteOffset": 31522816 } ], "md5sum": "b6d5f3109f3c081eb9b56f19c482e485" }, { "dataPath": "params_shard_4.bin", "format": "raw-shard", "nbytes": 33030144, "records": [ { "name": "model.layers.22.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 0 }, { "name": "model.layers.22.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 1835008 }, { "name": "model.layers.22.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 2064384 }, { "name": "model.layers.22.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 5734400 }, { "name": "model.layers.23.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 6193152 }, { "name": "model.layers.23.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 8028160 }, { "name": "model.layers.23.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 8257536 }, { "name": "model.layers.23.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 11927552 }, { "name": "model.layers.3.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 12386304 }, { "name": "model.layers.3.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 14221312 }, { "name": "model.layers.3.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 14450688 }, { "name": "model.layers.3.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 18120704 }, { "name": "model.layers.14.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 18579456 }, { "name": "model.layers.14.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 20414464 }, { "name": "model.layers.14.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 20643840 }, { "name": "model.layers.14.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 24313856 }, { "name": "model.layers.15.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 24772608 }, { "name": "model.layers.15.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 26607616 }, { "name": "model.layers.15.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 26836992 }, { "name": "model.layers.15.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 30507008 }, { "name": "model.layers.6.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 30965760 }, { "name": "model.layers.6.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 32800768 } ], "md5sum": "9f25a52e9442635c3a2f1ce6c3bd9ba3" }, { "dataPath": "params_shard_5.bin", "format": "raw-shard", "nbytes": 33030144, "records": [ { "name": "model.layers.6.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 0 }, { "name": "model.layers.6.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 3670016 }, { "name": "model.layers.7.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 4128768 }, { "name": "model.layers.7.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 5963776 }, { "name": "model.layers.7.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 6193152 }, { "name": "model.layers.7.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 9863168 }, { "name": "model.layers.10.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 10321920 }, { "name": "model.layers.10.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 12156928 }, { "name": "model.layers.11.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 12386304 }, { "name": "model.layers.11.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 14221312 }, { "name": "model.layers.11.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 14450688 }, { "name": "model.layers.11.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 18120704 }, { "name": "model.layers.9.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 18579456 }, { "name": "model.layers.9.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 20414464 }, { "name": "model.layers.9.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 20643840 }, { "name": "model.layers.9.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 24313856 }, { "name": "model.layers.17.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 24772608 }, { "name": "model.layers.17.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 26607616 }, { "name": "model.layers.17.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 26836992 }, { "name": "model.layers.17.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 30507008 }, { "name": "model.layers.18.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 30965760 }, { "name": "model.layers.18.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 32800768 } ], "md5sum": "7064d6f6168e4ff3d55e490036daf06d" }, { "dataPath": "params_shard_6.bin", "format": "raw-shard", "nbytes": 30965760, "records": [ { "name": "model.layers.18.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 0 }, { "name": "model.layers.18.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 3670016 }, { "name": "model.layers.12.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 4128768 }, { "name": "model.layers.12.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 5963776 }, { "name": "model.layers.12.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 6193152 }, { "name": "model.layers.12.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 9863168 }, { "name": "model.layers.13.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 10321920 }, { "name": "model.layers.13.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 12156928 }, { "name": "model.layers.13.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 12386304 }, { "name": "model.layers.13.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 16056320 }, { "name": "model.layers.19.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 16515072 }, { "name": "model.layers.19.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 18350080 }, { "name": "model.layers.19.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 18579456 }, { "name": "model.layers.19.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 22249472 }, { "name": "model.layers.4.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 22708224 }, { "name": "model.layers.4.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 24543232 }, { "name": "model.layers.4.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 24772608 }, { "name": "model.layers.4.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 28442624 }, { "name": "model.layers.2.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 28901376 }, { "name": "model.layers.2.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 30736384 } ], "md5sum": "d2b3cd51ac5f45f85ec003d7d5efc597" }, { "dataPath": "params_shard_7.bin", "format": "raw-shard", "nbytes": 33030144, "records": [ { "name": "model.layers.2.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 0 }, { "name": "model.layers.2.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 3670016 }, { "name": "model.layers.20.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 4128768 }, { "name": "model.layers.20.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 5963776 }, { "name": "model.layers.20.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 6193152 }, { "name": "model.layers.20.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 9863168 }, { "name": "model.layers.1.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 10321920 }, { "name": "model.layers.1.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 12156928 }, { "name": "model.layers.1.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 12386304 }, { "name": "model.layers.1.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 16056320 }, { "name": "model.layers.10.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 16515072 }, { "name": "model.layers.10.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 20185088 }, { "name": "model.layers.16.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 20643840 }, { "name": "model.layers.16.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 22478848 }, { "name": "model.layers.16.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 22708224 }, { "name": "model.layers.16.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 26378240 }, { "name": "model.layers.5.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 26836992 }, { "name": "model.layers.5.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 28672000 }, { "name": "model.layers.5.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 28901376 }, { "name": "model.layers.5.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 32571392 } ], "md5sum": "cc1a1ec57404c6f583702179d2c66693" }, { "dataPath": "params_shard_8.bin", "format": "raw-shard", "nbytes": 32735232, "records": [ { "name": "model.layers.8.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 0 }, { "name": "model.layers.8.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 1835008 }, { "name": "model.layers.8.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 2064384 }, { "name": "model.layers.8.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 5734400 }, { "name": "model.layers.21.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 6193152 }, { "name": "model.layers.21.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 8028160 }, { "name": "model.layers.21.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 8257536 }, { "name": "model.layers.21.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 11927552 }, { "name": "model.layers.0.mlp.down_proj.q_weight", "shape": [ 1024, 448 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1835008, "byteOffset": 12386304 }, { "name": "model.layers.0.mlp.down_proj.q_scale", "shape": [ 1024, 112 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 229376, "byteOffset": 14221312 }, { "name": "model.layers.0.mlp.gate_up_proj.q_weight", "shape": [ 7168, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3670016, "byteOffset": 14450688 }, { "name": "model.layers.0.mlp.gate_up_proj.q_scale", "shape": [ 7168, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 458752, "byteOffset": 18120704 }, { "name": "model.layers.23.self_attn.o_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 18579456 }, { "name": "model.layers.23.self_attn.o_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 19628032 }, { "name": "model.layers.3.self_attn.o_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 19759104 }, { "name": "model.layers.3.self_attn.o_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 20807680 }, { "name": "model.layers.14.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 20938752 }, { "name": "model.layers.14.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 21987328 }, { "name": "model.layers.14.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 22118400 }, { "name": "model.layers.14.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 23166976 }, { "name": "model.layers.15.self_attn.o_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 23298048 }, { "name": "model.layers.15.self_attn.o_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 24346624 }, { "name": "model.layers.6.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 24477696 }, { "name": "model.layers.6.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 25526272 }, { "name": "model.layers.7.self_attn.o_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 25657344 }, { "name": "model.layers.7.self_attn.o_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 26705920 }, { "name": "model.layers.11.self_attn.o_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 26836992 }, { "name": "model.layers.11.self_attn.o_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 27885568 }, { "name": "model.layers.12.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 28016640 }, { "name": "model.layers.12.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 29065216 }, { "name": "model.layers.12.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 29196288 }, { "name": "model.layers.12.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 30244864 }, { "name": "model.layers.17.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 30375936 }, { "name": "model.layers.17.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 31424512 }, { "name": "model.layers.18.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 31555584 }, { "name": "model.layers.18.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 32604160 } ], "md5sum": "3ae376ffcc7d7f5bb6611b3a60f06892" }, { "dataPath": "params_shard_9.bin", "format": "raw-shard", "nbytes": 33030144, "records": [ { "name": "model.layers.18.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 0 }, { "name": "model.layers.18.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 1048576 }, { "name": "model.layers.13.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 1179648 }, { "name": "model.layers.13.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 2228224 }, { "name": "model.layers.13.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 2359296 }, { "name": "model.layers.13.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 3407872 }, { "name": "model.layers.19.self_attn.o_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 3538944 }, { "name": "model.layers.19.self_attn.o_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 4587520 }, { "name": "model.layers.2.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 4718592 }, { "name": "model.layers.2.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 5767168 }, { "name": "model.layers.2.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 5898240 }, { "name": "model.layers.2.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 6946816 }, { "name": "model.layers.4.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 7077888 }, { "name": "model.layers.4.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 8126464 }, { "name": "model.layers.4.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 8257536 }, { "name": "model.layers.4.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 9306112 }, { "name": "model.layers.20.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 9437184 }, { "name": "model.layers.20.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 10485760 }, { "name": "model.layers.20.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 10616832 }, { "name": "model.layers.20.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 11665408 }, { "name": "model.layers.1.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 11796480 }, { "name": "model.layers.1.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 12845056 }, { "name": "model.layers.1.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 12976128 }, { "name": "model.layers.1.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 14024704 }, { "name": "model.layers.10.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 14155776 }, { "name": "model.layers.10.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 15204352 }, { "name": "model.layers.10.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 15335424 }, { "name": "model.layers.10.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 16384000 }, { "name": "model.layers.16.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 16515072 }, { "name": "model.layers.16.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 17563648 }, { "name": "model.layers.16.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 17694720 }, { "name": "model.layers.16.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 18743296 }, { "name": "model.layers.17.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 18874368 }, { "name": "model.layers.17.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 19922944 }, { "name": "model.layers.5.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 20054016 }, { "name": "model.layers.5.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 21102592 }, { "name": "model.layers.5.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 21233664 }, { "name": "model.layers.5.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 22282240 }, { "name": "model.layers.6.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 22413312 }, { "name": "model.layers.6.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 23461888 }, { "name": "model.layers.8.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 23592960 }, { "name": "model.layers.8.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 24641536 }, { "name": "model.layers.8.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 24772608 }, { "name": "model.layers.8.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 25821184 }, { "name": "model.layers.9.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 25952256 }, { "name": "model.layers.9.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 27000832 }, { "name": "model.layers.9.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 27131904 }, { "name": "model.layers.9.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 28180480 }, { "name": "model.layers.21.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 28311552 }, { "name": "model.layers.21.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 29360128 }, { "name": "model.layers.21.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 29491200 }, { "name": "model.layers.21.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 30539776 }, { "name": "model.layers.22.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 30670848 }, { "name": "model.layers.22.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 31719424 }, { "name": "model.layers.22.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 31850496 }, { "name": "model.layers.22.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 32899072 } ], "md5sum": "5332ab8d42fd513dc02a939b30d53747" }, { "dataPath": "params_shard_10.bin", "format": "raw-shard", "nbytes": 3688064, "records": [ { "name": "model.layers.0.linear_attn.in_proj_z.q_weight", "shape": [ 2048, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 0 }, { "name": "model.layers.0.linear_attn.in_proj_z.q_scale", "shape": [ 2048, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 1048576 }, { "name": "model.layers.0.linear_attn.out_proj.q_weight", "shape": [ 1024, 256 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 1048576, "byteOffset": 1179648 }, { "name": "model.layers.0.linear_attn.out_proj.q_scale", "shape": [ 1024, 64 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 131072, "byteOffset": 2228224 }, { "name": "model.layers.8.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 2359296 }, { "name": "model.layers.12.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 2408448 }, { "name": "model.layers.18.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 2457600 }, { "name": "model.layers.13.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 2506752 }, { "name": "model.layers.14.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 2555904 }, { "name": "model.layers.2.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 2605056 }, { "name": "model.layers.4.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 2654208 }, { "name": "model.layers.5.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 2703360 }, { "name": "model.layers.20.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 2752512 }, { "name": "model.layers.21.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 2801664 }, { "name": "model.layers.10.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 2850816 }, { "name": "model.layers.16.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 2899968 }, { "name": "model.layers.17.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 2949120 }, { "name": "model.layers.6.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 2998272 }, { "name": "model.layers.9.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 3047424 }, { "name": "model.layers.22.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 3096576 }, { "name": "model.layers.0.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 3145728 }, { "name": "model.layers.1.linear_attn.conv1d_weight", "shape": [ 6144, 1, 4 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 49152, "byteOffset": 3194880 }, { "name": "model.layers.14.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3244032 }, { "name": "model.layers.14.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3252224 }, { "name": "model.layers.14.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3253248 }, { "name": "model.layers.14.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3261440 }, { "name": "model.layers.12.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3262464 }, { "name": "model.layers.12.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3270656 }, { "name": "model.layers.12.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3271680 }, { "name": "model.layers.12.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3279872 }, { "name": "model.layers.18.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3280896 }, { "name": "model.layers.18.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3289088 }, { "name": "model.layers.18.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3290112 }, { "name": "model.layers.18.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3298304 }, { "name": "model.layers.13.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3299328 }, { "name": "model.layers.13.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3307520 }, { "name": "model.layers.13.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3308544 }, { "name": "model.layers.13.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3316736 }, { "name": "model.layers.2.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3317760 }, { "name": "model.layers.2.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3325952 }, { "name": "model.layers.2.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3326976 }, { "name": "model.layers.2.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3335168 }, { "name": "model.layers.4.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3336192 }, { "name": "model.layers.4.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3344384 }, { "name": "model.layers.4.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3345408 }, { "name": "model.layers.4.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3353600 }, { "name": "model.layers.20.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3354624 }, { "name": "model.layers.20.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3362816 }, { "name": "model.layers.20.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3363840 }, { "name": "model.layers.20.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3372032 }, { "name": "model.layers.1.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3373056 }, { "name": "model.layers.1.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3381248 }, { "name": "model.layers.1.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3382272 }, { "name": "model.layers.1.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3390464 }, { "name": "model.layers.10.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3391488 }, { "name": "model.layers.10.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3399680 }, { "name": "model.layers.10.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3400704 }, { "name": "model.layers.10.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3408896 }, { "name": "model.layers.16.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3409920 }, { "name": "model.layers.16.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3418112 }, { "name": "model.layers.16.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3419136 }, { "name": "model.layers.16.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3427328 }, { "name": "model.layers.17.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3428352 }, { "name": "model.layers.17.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3436544 }, { "name": "model.layers.17.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3437568 }, { "name": "model.layers.17.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3445760 }, { "name": "model.layers.5.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3446784 }, { "name": "model.layers.5.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3454976 }, { "name": "model.layers.5.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3456000 }, { "name": "model.layers.5.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3464192 }, { "name": "model.layers.6.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3465216 }, { "name": "model.layers.6.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3473408 }, { "name": "model.layers.6.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3474432 }, { "name": "model.layers.6.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3482624 }, { "name": "model.layers.8.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3483648 }, { "name": "model.layers.8.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3491840 }, { "name": "model.layers.8.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3492864 }, { "name": "model.layers.8.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3501056 }, { "name": "model.layers.9.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3502080 }, { "name": "model.layers.9.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3510272 }, { "name": "model.layers.9.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3511296 }, { "name": "model.layers.9.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3519488 }, { "name": "model.layers.21.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3520512 }, { "name": "model.layers.21.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3528704 }, { "name": "model.layers.21.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3529728 }, { "name": "model.layers.21.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3537920 }, { "name": "model.layers.22.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3538944 }, { "name": "model.layers.22.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3547136 }, { "name": "model.layers.22.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3548160 }, { "name": "model.layers.22.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3556352 }, { "name": "model.layers.0.linear_attn.in_proj_b.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3557376 }, { "name": "model.layers.0.linear_attn.in_proj_b.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3565568 }, { "name": "model.layers.0.linear_attn.in_proj_a.q_weight", "shape": [ 16, 128 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 8192, "byteOffset": 3566592 }, { "name": "model.layers.0.linear_attn.in_proj_a.q_scale", "shape": [ 16, 32 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1024, "byteOffset": 3574784 }, { "name": "model.layers.22.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3575808 }, { "name": "model.layers.23.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3577856 }, { "name": "model.layers.23.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3579904 }, { "name": "model.layers.3.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3581952 }, { "name": "model.layers.14.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3584000 }, { "name": "model.layers.14.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3586048 }, { "name": "model.layers.15.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3588096 }, { "name": "model.layers.15.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3590144 }, { "name": "model.layers.6.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3592192 }, { "name": "model.layers.7.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3594240 }, { "name": "model.layers.7.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3596288 }, { "name": "model.layers.8.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3598336 }, { "name": "model.layers.11.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3600384 }, { "name": "model.layers.11.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3602432 }, { "name": "model.layers.12.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3604480 }, { "name": "model.layers.17.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3606528 }, { "name": "model.layers.18.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3608576 }, { "name": "model.layers.18.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3610624 }, { "name": "model.layers.19.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3612672 }, { "name": "model.layers.12.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3614720 }, { "name": "model.layers.13.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3616768 }, { "name": "model.layers.13.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3618816 }, { "name": "model.layers.19.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3620864 }, { "name": "model.layers.2.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3622912 }, { "name": "model.layers.2.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3624960 }, { "name": "model.layers.3.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3627008 }, { "name": "model.layers.4.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3629056 }, { "name": "model.layers.4.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3631104 }, { "name": "model.layers.20.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3633152 }, { "name": "model.layers.20.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3635200 }, { "name": "model.layers.1.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3637248 }, { "name": "model.layers.10.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3639296 }, { "name": "model.layers.10.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3641344 }, { "name": "model.layers.16.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3643392 }, { "name": "model.layers.16.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3645440 }, { "name": "model.layers.17.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3647488 }, { "name": "model.layers.5.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3649536 }, { "name": "model.layers.5.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3651584 }, { "name": "model.layers.6.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3653632 }, { "name": "model.layers.8.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3655680 }, { "name": "model.layers.9.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3657728 }, { "name": "model.layers.9.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3659776 }, { "name": "model.layers.21.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3661824 }, { "name": "model.layers.21.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3663872 }, { "name": "model.layers.22.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3665920 }, { "name": "model.layers.0.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3667968 }, { "name": "model.layers.0.post_attention_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3670016 }, { "name": "model.layers.1.input_layernorm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3672064 }, { "name": "model.norm.weight", "shape": [ 1024 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 2048, "byteOffset": 3674112 }, { "name": "model.layers.23.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 3676160 }, { "name": "model.layers.23.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 3676672 }, { "name": "model.layers.3.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 3677184 }, { "name": "model.layers.14.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3677696 }, { "name": "model.layers.15.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 3677952 }, { "name": "model.layers.15.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 3678464 }, { "name": "model.layers.7.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 3678976 }, { "name": "model.layers.7.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 3679488 }, { "name": "model.layers.11.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 3680000 }, { "name": "model.layers.11.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 3680512 }, { "name": "model.layers.12.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3681024 }, { "name": "model.layers.18.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3681280 }, { "name": "model.layers.13.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3681536 }, { "name": "model.layers.19.self_attn.k_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 3681792 }, { "name": "model.layers.19.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 3682304 }, { "name": "model.layers.2.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3682816 }, { "name": "model.layers.3.self_attn.q_norm.weight", "shape": [ 256 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 512, "byteOffset": 3683072 }, { "name": "model.layers.4.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3683584 }, { "name": "model.layers.20.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3683840 }, { "name": "model.layers.1.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3684096 }, { "name": "model.layers.10.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3684352 }, { "name": "model.layers.16.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3684608 }, { "name": "model.layers.17.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3684864 }, { "name": "model.layers.5.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3685120 }, { "name": "model.layers.6.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3685376 }, { "name": "model.layers.8.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3685632 }, { "name": "model.layers.9.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3685888 }, { "name": "model.layers.21.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3686144 }, { "name": "model.layers.22.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3686400 }, { "name": "model.layers.0.linear_attn.norm.weight", "shape": [ 128 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 256, "byteOffset": 3686656 }, { "name": "model.layers.16.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3686912 }, { "name": "model.layers.8.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3686944 }, { "name": "model.layers.12.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3686976 }, { "name": "model.layers.18.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687008 }, { "name": "model.layers.13.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687040 }, { "name": "model.layers.14.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687072 }, { "name": "model.layers.2.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687104 }, { "name": "model.layers.4.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687136 }, { "name": "model.layers.5.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687168 }, { "name": "model.layers.20.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687200 }, { "name": "model.layers.21.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687232 }, { "name": "model.layers.10.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687264 }, { "name": "model.layers.17.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687296 }, { "name": "model.layers.6.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687328 }, { "name": "model.layers.9.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687360 }, { "name": "model.layers.22.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687392 }, { "name": "model.layers.0.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687424 }, { "name": "model.layers.1.linear_attn.A_log", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687456 }, { "name": "model.layers.8.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687488 }, { "name": "model.layers.12.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687520 }, { "name": "model.layers.18.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687552 }, { "name": "model.layers.13.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687584 }, { "name": "model.layers.14.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687616 }, { "name": "model.layers.2.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687648 }, { "name": "model.layers.4.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687680 }, { "name": "model.layers.20.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687712 }, { "name": "model.layers.21.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687744 }, { "name": "model.layers.10.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687776 }, { "name": "model.layers.16.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687808 }, { "name": "model.layers.17.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687840 }, { "name": "model.layers.5.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687872 }, { "name": "model.layers.6.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687904 }, { "name": "model.layers.9.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687936 }, { "name": "model.layers.22.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3687968 }, { "name": "model.layers.0.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3688000 }, { "name": "model.layers.1.linear_attn.dt_bias", "shape": [ 16 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 32, "byteOffset": 3688032 } ], "md5sum": "d1291e75d11e70131b8c954733f83272" } ] }