| { |
| "tensor_count": 93, |
| "total_tensor_bytes": 2325648, |
| "tensors": [ |
| { |
| "name": "block.0.attn.norm.scale", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "eccf6206b565034cbae911aefab9c5a7dc483288cafd3f454089bc37ea39334d" |
| }, |
| { |
| "name": "block.0.attn.out.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "c3f9fca0599156bb298857bcb3085fda3f768064d702bd1b8014cf532bd9f715" |
| }, |
| { |
| "name": "block.0.attn.out.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128, |
| 128 |
| ], |
| "numel": 16384, |
| "nbytes": 32768, |
| "tensor_sha256": "d95bf0cad74e0854831f86981303760e10ec3c2f4ccf4b027b145b8d251a51c3" |
| }, |
| { |
| "name": "block.0.attn.qkv.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 192 |
| ], |
| "numel": 192, |
| "nbytes": 384, |
| "tensor_sha256": "87074fba354a69d4c62022329006b2013c49686bc497b126e20f3a8e71fd1e9b" |
| }, |
| { |
| "name": "block.0.attn.qkv.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 192, |
| 128 |
| ], |
| "numel": 24576, |
| "nbytes": 49152, |
| "tensor_sha256": "e1944b4f33dafdf6f86d59a78aa6093a20c71c5ca044be20992bb9df146ba1cc" |
| }, |
| { |
| "name": "block.0.attn.sinks", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 4 |
| ], |
| "numel": 4, |
| "nbytes": 8, |
| "tensor_sha256": "295ddf9b80c106900373a47e8bc3283c7d46496cb85a80ce9beecf919bffa36d" |
| }, |
| { |
| "name": "block.0.mlp.gate.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8 |
| ], |
| "numel": 8, |
| "nbytes": 16, |
| "tensor_sha256": "a39a0ec9a48054ed024cf42fd0348cf417b140f67a2c35657fd4e927a0db0e72" |
| }, |
| { |
| "name": "block.0.mlp.gate.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 128 |
| ], |
| "numel": 1024, |
| "nbytes": 2048, |
| "tensor_sha256": "4bd377300b228d0fdfec98a8d48fbb37e4f86cddfb20333dd063162d30b4d327" |
| }, |
| { |
| "name": "block.0.mlp.mlp1_bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 256 |
| ], |
| "numel": 2048, |
| "nbytes": 4096, |
| "tensor_sha256": "d9a0ab2d73984206682035017d3cd8fe5a775059df2d9ac1f595af5ec2bc9ebd" |
| }, |
| { |
| "name": "block.0.mlp.mlp1_weight.blocks", |
| "profile_dtype": "FP4", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 256, |
| 4, |
| 16 |
| ], |
| "numel": 131072, |
| "nbytes": 131072, |
| "tensor_sha256": "d0ac79ceb22dc812565c5ba380d391b03dc8ca5294827eadc479c880543aacee" |
| }, |
| { |
| "name": "block.0.mlp.mlp1_weight.scales", |
| "profile_dtype": "UE8", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 256, |
| 4 |
| ], |
| "numel": 8192, |
| "nbytes": 8192, |
| "tensor_sha256": "90262d9132d9834b218ab379ba3edf958c880ffc604f3b45cb8bbb0d90d33130" |
| }, |
| { |
| "name": "block.0.mlp.mlp2_bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 128 |
| ], |
| "numel": 1024, |
| "nbytes": 2048, |
| "tensor_sha256": "449d510ec646b8bcc3a79ed5f71a7161c3e56f6db720bc4accf91c93e91a41e3" |
| }, |
| { |
| "name": "block.0.mlp.mlp2_weight.blocks", |
| "profile_dtype": "FP4", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 128, |
| 4, |
| 16 |
| ], |
| "numel": 65536, |
| "nbytes": 65536, |
| "tensor_sha256": "5e5dc1deac2d388d49843cc7585b6ae14ad51bbe623102e4a225dfa3e38b6f8d" |
| }, |
| { |
| "name": "block.0.mlp.mlp2_weight.scales", |
| "profile_dtype": "UE8", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 128, |
| 4 |
| ], |
| "numel": 4096, |
| "nbytes": 4096, |
| "tensor_sha256": "ac8cb8c5ef0cf4d742f23d9b054fd08b6dd4b9f74838631d42c3090ef585cbdd" |
| }, |
| { |
| "name": "block.0.mlp.norm.scale", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "b63ff56f26e563de022aab5369580eda5d74578a5f22cc9dd56bf6db0f1d7082" |
| }, |
| { |
| "name": "block.1.attn.norm.scale", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "83cf4c38d54a727a4c7fb56ae3cc75ecff6bcf9ead14607ada8b0be293eec882" |
| }, |
| { |
| "name": "block.1.attn.out.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "fbfa71e7dc3809307c8501fec973d7a1ab81ebedcbbb4b1b9b40cec5522f70d5" |
| }, |
| { |
| "name": "block.1.attn.out.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128, |
| 128 |
| ], |
| "numel": 16384, |
| "nbytes": 32768, |
| "tensor_sha256": "6dd7263990966265cb55a5e8ea08a72853c012a9e64248d907699d28a862fb51" |
| }, |
| { |
| "name": "block.1.attn.qkv.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 192 |
| ], |
| "numel": 192, |
| "nbytes": 384, |
| "tensor_sha256": "5e1e38393fa6ab1456475ae99c4f39dace4c2a2f6a60de0d66685fb633d00d8e" |
| }, |
| { |
| "name": "block.1.attn.qkv.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 192, |
| 128 |
| ], |
| "numel": 24576, |
| "nbytes": 49152, |
| "tensor_sha256": "d10d95738b787c9f0f4e40c8889a734e91e010b94631aa4b0430d9f29d606d4d" |
| }, |
| { |
| "name": "block.1.attn.sinks", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 4 |
| ], |
| "numel": 4, |
| "nbytes": 8, |
| "tensor_sha256": "a117079fbf84f8c53350f712ee5ec94f14ef7f60310448c1c52c08b481f44ba2" |
| }, |
| { |
| "name": "block.1.mlp.gate.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8 |
| ], |
| "numel": 8, |
| "nbytes": 16, |
| "tensor_sha256": "e5af7ddf38f2ae2b2cb5afeaeac8ffd8ae4db992998f0e9bdd08471de8386f00" |
| }, |
| { |
| "name": "block.1.mlp.gate.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 128 |
| ], |
| "numel": 1024, |
| "nbytes": 2048, |
| "tensor_sha256": "8e5c37e0f732ce2bb0810948716dc8efb24959cd60f3fc36bab8e5c0054000d4" |
| }, |
| { |
| "name": "block.1.mlp.mlp1_bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 256 |
| ], |
| "numel": 2048, |
| "nbytes": 4096, |
| "tensor_sha256": "ead631eded681e83c149d5ad987504ef57d4d852698b1e332f50e937bf61ce2b" |
| }, |
| { |
| "name": "block.1.mlp.mlp1_weight.blocks", |
| "profile_dtype": "FP4", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 256, |
| 4, |
| 16 |
| ], |
| "numel": 131072, |
| "nbytes": 131072, |
| "tensor_sha256": "e87b455f4612e44c79fc1a6af39899f6773beee29d2477ae4dbb8b5d514c3dcd" |
| }, |
| { |
| "name": "block.1.mlp.mlp1_weight.scales", |
| "profile_dtype": "UE8", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 256, |
| 4 |
| ], |
| "numel": 8192, |
| "nbytes": 8192, |
| "tensor_sha256": "7ec5077d46ca6072fc767e945f3de37522182e754732dc2230fa972dc416c5c8" |
| }, |
| { |
| "name": "block.1.mlp.mlp2_bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 128 |
| ], |
| "numel": 1024, |
| "nbytes": 2048, |
| "tensor_sha256": "61febd0af1c19417497e9d0db0cf6063fe5f581c4232197ac5d17d59685e32ff" |
| }, |
| { |
| "name": "block.1.mlp.mlp2_weight.blocks", |
| "profile_dtype": "FP4", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 128, |
| 4, |
| 16 |
| ], |
| "numel": 65536, |
| "nbytes": 65536, |
| "tensor_sha256": "66df91b2652cf14f96f2b1542ad45d3c6742b3d4986f347493d4b9cf52c5e8ac" |
| }, |
| { |
| "name": "block.1.mlp.mlp2_weight.scales", |
| "profile_dtype": "UE8", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 128, |
| 4 |
| ], |
| "numel": 4096, |
| "nbytes": 4096, |
| "tensor_sha256": "91d42170629b6c5ddd387ddd24cd59ea6b434ed567c081c54d858aa8e4558ba7" |
| }, |
| { |
| "name": "block.1.mlp.norm.scale", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "68363b0cc5e0c20ff11f9b0107039584f3aa8bb166779428f6df0b6f3f562fee" |
| }, |
| { |
| "name": "block.2.attn.norm.scale", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "6e2b8342c7466915587ff9d08464018674d63e0cec5bfdeae8b9268cf12b6de8" |
| }, |
| { |
| "name": "block.2.attn.out.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "366efbc20599e24880cb6cb7104caf9b795a879d0b3a04486353b2ca402ef1b1" |
| }, |
| { |
| "name": "block.2.attn.out.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128, |
| 128 |
| ], |
| "numel": 16384, |
| "nbytes": 32768, |
| "tensor_sha256": "566031b008db7b923f19583f172c30ec6ba016cc542c696bb28d3f7fc5ab589e" |
| }, |
| { |
| "name": "block.2.attn.qkv.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 192 |
| ], |
| "numel": 192, |
| "nbytes": 384, |
| "tensor_sha256": "f1098142b9f8e484df961e76053c0e5185643d91808fc0b36eb6bc0f5b03175a" |
| }, |
| { |
| "name": "block.2.attn.qkv.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 192, |
| 128 |
| ], |
| "numel": 24576, |
| "nbytes": 49152, |
| "tensor_sha256": "e4711d5191b1858069796141f4c39c384bf45f62f6d2386bb81d6af13ea4b53b" |
| }, |
| { |
| "name": "block.2.attn.sinks", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 4 |
| ], |
| "numel": 4, |
| "nbytes": 8, |
| "tensor_sha256": "92dfb6f47e247ad094f13290f11fc41e788f4b9f9ace9dd69151bf77701da41e" |
| }, |
| { |
| "name": "block.2.mlp.gate.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8 |
| ], |
| "numel": 8, |
| "nbytes": 16, |
| "tensor_sha256": "a25f01749a28dcb48c5b1bba82915073fc0725405167bc48e65a5f1c5168650a" |
| }, |
| { |
| "name": "block.2.mlp.gate.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 128 |
| ], |
| "numel": 1024, |
| "nbytes": 2048, |
| "tensor_sha256": "d6a5332f558e60c6ab499182817dbb69be4d5ccc54f49c26ab0ac2e41688c9b3" |
| }, |
| { |
| "name": "block.2.mlp.mlp1_bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 256 |
| ], |
| "numel": 2048, |
| "nbytes": 4096, |
| "tensor_sha256": "62283e7f26a3806807c9ac37e2e3fdb0cce7775227db14c66a1414f6fed89c53" |
| }, |
| { |
| "name": "block.2.mlp.mlp1_weight.blocks", |
| "profile_dtype": "FP4", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 256, |
| 4, |
| 16 |
| ], |
| "numel": 131072, |
| "nbytes": 131072, |
| "tensor_sha256": "022584a26dd938b7659332af69b2047539911c800b08474ba9e5ab2bdc784300" |
| }, |
| { |
| "name": "block.2.mlp.mlp1_weight.scales", |
| "profile_dtype": "UE8", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 256, |
| 4 |
| ], |
| "numel": 8192, |
| "nbytes": 8192, |
| "tensor_sha256": "e380540cad368470ffabf8725de2d74d56d3fc05117a36218ba9a1df73e21788" |
| }, |
| { |
| "name": "block.2.mlp.mlp2_bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 128 |
| ], |
| "numel": 1024, |
| "nbytes": 2048, |
| "tensor_sha256": "ff9e10da6f5e16f91876ef5724e9825f6a59e057e063fbf51bd0850d0524d694" |
| }, |
| { |
| "name": "block.2.mlp.mlp2_weight.blocks", |
| "profile_dtype": "FP4", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 128, |
| 4, |
| 16 |
| ], |
| "numel": 65536, |
| "nbytes": 65536, |
| "tensor_sha256": "178904bea25592b069946c8cadd1301715ef0a5c8082ba0d9f817e039c991d63" |
| }, |
| { |
| "name": "block.2.mlp.mlp2_weight.scales", |
| "profile_dtype": "UE8", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 128, |
| 4 |
| ], |
| "numel": 4096, |
| "nbytes": 4096, |
| "tensor_sha256": "d84dcdd10ce6ee20c75b3a46cf29540dfc4c5f223d69321897f59bd681ba8515" |
| }, |
| { |
| "name": "block.2.mlp.norm.scale", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "be8f537c77171cbcbefbd987ff1501ae5bed6fb883c16d6a34b698876f1ee3b4" |
| }, |
| { |
| "name": "block.3.attn.norm.scale", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "a63a9daec9ce2d3eeac2fbbeae74b0138e6ba1ab9dda9ad4fa2554e0c2ca2de1" |
| }, |
| { |
| "name": "block.3.attn.out.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "342ea41e39091ddea3dceb699eae5a3f06b2d925bbc4c67ad9886d9caa4d53da" |
| }, |
| { |
| "name": "block.3.attn.out.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128, |
| 128 |
| ], |
| "numel": 16384, |
| "nbytes": 32768, |
| "tensor_sha256": "057794a40c48b05ad393b17db48203dbcb28cb67f2060645f02a98a6c6829f1c" |
| }, |
| { |
| "name": "block.3.attn.qkv.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 192 |
| ], |
| "numel": 192, |
| "nbytes": 384, |
| "tensor_sha256": "856e44029cca502df4abbc3995a1470ee25fb33f39890136806af561beb9088e" |
| }, |
| { |
| "name": "block.3.attn.qkv.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 192, |
| 128 |
| ], |
| "numel": 24576, |
| "nbytes": 49152, |
| "tensor_sha256": "e56cc7e6c0e132cea4fa41cf4815d742521974bc10d67154faea7a50f69dcad4" |
| }, |
| { |
| "name": "block.3.attn.sinks", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 4 |
| ], |
| "numel": 4, |
| "nbytes": 8, |
| "tensor_sha256": "61cb27dc684d0a0547f12e76b164aa50ace8f3d63ba6c830da694c92fa5f097b" |
| }, |
| { |
| "name": "block.3.mlp.gate.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8 |
| ], |
| "numel": 8, |
| "nbytes": 16, |
| "tensor_sha256": "2a05530d4e67ca5c305a49f2b4f3dfa35214d2d6bf2980b1d39deb62c4696b1c" |
| }, |
| { |
| "name": "block.3.mlp.gate.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 128 |
| ], |
| "numel": 1024, |
| "nbytes": 2048, |
| "tensor_sha256": "0254ec6589ab9894a11d4b7fe11900dba9f5715aa7f1a29dfc5aad6c7fbdb01e" |
| }, |
| { |
| "name": "block.3.mlp.mlp1_bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 256 |
| ], |
| "numel": 2048, |
| "nbytes": 4096, |
| "tensor_sha256": "a4b8b349adfaf60be1e8f5dd89cb9dc236e596c6045f7d3bdd7573e0fdcca4e9" |
| }, |
| { |
| "name": "block.3.mlp.mlp1_weight.blocks", |
| "profile_dtype": "FP4", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 256, |
| 4, |
| 16 |
| ], |
| "numel": 131072, |
| "nbytes": 131072, |
| "tensor_sha256": "ad0f0d6726de843bf2c766a40b1fd3f3ea5f0844bce9238193455b3ada7357cc" |
| }, |
| { |
| "name": "block.3.mlp.mlp1_weight.scales", |
| "profile_dtype": "UE8", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 256, |
| 4 |
| ], |
| "numel": 8192, |
| "nbytes": 8192, |
| "tensor_sha256": "63973340238aceb7e6158b3b2a1b227bac28ab2924a07e16f25cfa4ea03e0e72" |
| }, |
| { |
| "name": "block.3.mlp.mlp2_bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 128 |
| ], |
| "numel": 1024, |
| "nbytes": 2048, |
| "tensor_sha256": "b4c9587a5885563ff92e98626e16e859bab5ca2a22061fb0eb4a478def07ed97" |
| }, |
| { |
| "name": "block.3.mlp.mlp2_weight.blocks", |
| "profile_dtype": "FP4", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 128, |
| 4, |
| 16 |
| ], |
| "numel": 65536, |
| "nbytes": 65536, |
| "tensor_sha256": "8c8ce501f6b10800c56f9053a71e3aadc4ea2cdbd1cdf711ac70f39e9b2493e5" |
| }, |
| { |
| "name": "block.3.mlp.mlp2_weight.scales", |
| "profile_dtype": "UE8", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 128, |
| 4 |
| ], |
| "numel": 4096, |
| "nbytes": 4096, |
| "tensor_sha256": "ced5ca3bc9bf9bf8060326484a79514cb17478b54a056b6bade9c936af146726" |
| }, |
| { |
| "name": "block.3.mlp.norm.scale", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "e3f47dc61b2798a760d584319069246b56891571eadd2a51ee8fadc1fc83aac0" |
| }, |
| { |
| "name": "block.4.attn.norm.scale", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "411109db41ab85e6668a07b60509aa321745699e7c2acbba38ed6b2cb77b99d9" |
| }, |
| { |
| "name": "block.4.attn.out.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "1d145798ca557f230c7a5559243173723ac013b408dca1a5af37ae95f81fcd40" |
| }, |
| { |
| "name": "block.4.attn.out.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128, |
| 128 |
| ], |
| "numel": 16384, |
| "nbytes": 32768, |
| "tensor_sha256": "8af8f74e0b4a220b62680b3a26d429a531bd7db70b5a1d77cf7cb4895eb72484" |
| }, |
| { |
| "name": "block.4.attn.qkv.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 192 |
| ], |
| "numel": 192, |
| "nbytes": 384, |
| "tensor_sha256": "6dee9f3beeef2fe7699ad3d6297b97a86be3fb42761141c2301018fa95710a74" |
| }, |
| { |
| "name": "block.4.attn.qkv.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 192, |
| 128 |
| ], |
| "numel": 24576, |
| "nbytes": 49152, |
| "tensor_sha256": "a7669d41f636b05f66d4e3ab3db8a7e8e1584cba7915dc930226760a63f9cf28" |
| }, |
| { |
| "name": "block.4.attn.sinks", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 4 |
| ], |
| "numel": 4, |
| "nbytes": 8, |
| "tensor_sha256": "e4876b0316d2a8ef7f4e3fa6dcdfb7b04114d8a0f8aedc8605a3f8d1ef0e83cf" |
| }, |
| { |
| "name": "block.4.mlp.gate.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8 |
| ], |
| "numel": 8, |
| "nbytes": 16, |
| "tensor_sha256": "a46518dc141801777ea50b8723d71cfebb64c4f0524b54e8d553b52f7439135c" |
| }, |
| { |
| "name": "block.4.mlp.gate.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 128 |
| ], |
| "numel": 1024, |
| "nbytes": 2048, |
| "tensor_sha256": "c6b3682df23d9c07674962210066010f90755836d948c437561b933e1d168e8b" |
| }, |
| { |
| "name": "block.4.mlp.mlp1_bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 256 |
| ], |
| "numel": 2048, |
| "nbytes": 4096, |
| "tensor_sha256": "22ee8aaffbdb4421b6cd54a7ed58d6f439567c7e7c443a52e95bcc084b74edab" |
| }, |
| { |
| "name": "block.4.mlp.mlp1_weight.blocks", |
| "profile_dtype": "FP4", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 256, |
| 4, |
| 16 |
| ], |
| "numel": 131072, |
| "nbytes": 131072, |
| "tensor_sha256": "a50822355f9c1b4c625294f6f0ba162e09c6ca4b308dc9b52a5d5311682dde25" |
| }, |
| { |
| "name": "block.4.mlp.mlp1_weight.scales", |
| "profile_dtype": "UE8", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 256, |
| 4 |
| ], |
| "numel": 8192, |
| "nbytes": 8192, |
| "tensor_sha256": "b9d1b50f963382e1ed34355421c8091642183d0856482149cbef32f4f32627ef" |
| }, |
| { |
| "name": "block.4.mlp.mlp2_bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 128 |
| ], |
| "numel": 1024, |
| "nbytes": 2048, |
| "tensor_sha256": "c6bde1e2894a2a4ea5555faccdea0f1787ad56f39acb552e1097dfda3b91d1a6" |
| }, |
| { |
| "name": "block.4.mlp.mlp2_weight.blocks", |
| "profile_dtype": "FP4", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 128, |
| 4, |
| 16 |
| ], |
| "numel": 65536, |
| "nbytes": 65536, |
| "tensor_sha256": "2b9f453cac8c3bc740b552cfcdf5dac414d4ecf5594bf5cc7ba72bd5135f9595" |
| }, |
| { |
| "name": "block.4.mlp.mlp2_weight.scales", |
| "profile_dtype": "UE8", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 128, |
| 4 |
| ], |
| "numel": 4096, |
| "nbytes": 4096, |
| "tensor_sha256": "51f2aed84152cd33655ecc33bc1aaf895955ab01b90c28540d8fcaeaa4b2c75c" |
| }, |
| { |
| "name": "block.4.mlp.norm.scale", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "c583f057e81a175b0f8faebba532f0b775138c3b29019fd8d28685d3ebfb9b6c" |
| }, |
| { |
| "name": "block.5.attn.norm.scale", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "89676d9c580766cfd9172b5d7050cb43628e127102ab766a18ddcb7235093620" |
| }, |
| { |
| "name": "block.5.attn.out.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "a9d2eea381c89d51d7c96374e9c0e2cb082b87783aee002e62d86065eecc0fb2" |
| }, |
| { |
| "name": "block.5.attn.out.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128, |
| 128 |
| ], |
| "numel": 16384, |
| "nbytes": 32768, |
| "tensor_sha256": "f2ca3c84225d7acf3386a7c3d4a9b07b2d717f379b645564a8a84137199b07e1" |
| }, |
| { |
| "name": "block.5.attn.qkv.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 192 |
| ], |
| "numel": 192, |
| "nbytes": 384, |
| "tensor_sha256": "e0ad88cc16c2daf0db1dfe0d6909fea440fa304d9b716cc1cc445276765d71d9" |
| }, |
| { |
| "name": "block.5.attn.qkv.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 192, |
| 128 |
| ], |
| "numel": 24576, |
| "nbytes": 49152, |
| "tensor_sha256": "81d4cfba44bab6d7914f27ab61896bcc84a2d814059ce000f5e12657d4a41866" |
| }, |
| { |
| "name": "block.5.attn.sinks", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 4 |
| ], |
| "numel": 4, |
| "nbytes": 8, |
| "tensor_sha256": "16e389e81866d6ec7e85ad3e2dd70e714428f4d92f5be84fca0b5e518ee33325" |
| }, |
| { |
| "name": "block.5.mlp.gate.bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8 |
| ], |
| "numel": 8, |
| "nbytes": 16, |
| "tensor_sha256": "e913d05270793f26e74da05493c26165a8ce46afb19f44031d109db53ac341a7" |
| }, |
| { |
| "name": "block.5.mlp.gate.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 128 |
| ], |
| "numel": 1024, |
| "nbytes": 2048, |
| "tensor_sha256": "782d4ddcc67b12bda908b0a4355b468df94e87d576ceae9db61bf35e5fc11f0b" |
| }, |
| { |
| "name": "block.5.mlp.mlp1_bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 256 |
| ], |
| "numel": 2048, |
| "nbytes": 4096, |
| "tensor_sha256": "1280153cf4f5b8d1a248abb7a080f8770792f58e5374ff1a3d0a16f04e38f191" |
| }, |
| { |
| "name": "block.5.mlp.mlp1_weight.blocks", |
| "profile_dtype": "FP4", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 256, |
| 4, |
| 16 |
| ], |
| "numel": 131072, |
| "nbytes": 131072, |
| "tensor_sha256": "19c88563037e63e3d3ebb0fd9ba6edbba2b9b4ad1ae314fccfe44a396e8ab904" |
| }, |
| { |
| "name": "block.5.mlp.mlp1_weight.scales", |
| "profile_dtype": "UE8", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 256, |
| 4 |
| ], |
| "numel": 8192, |
| "nbytes": 8192, |
| "tensor_sha256": "f7ddf93e3c0ab64ef3224f8e3ea6047c9e37681302e2a0f256784839059fa422" |
| }, |
| { |
| "name": "block.5.mlp.mlp2_bias", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 8, |
| 128 |
| ], |
| "numel": 1024, |
| "nbytes": 2048, |
| "tensor_sha256": "368c68c9d3b6d499d1dc53ad5f1b730817a589554af94bf6bf6334636977634f" |
| }, |
| { |
| "name": "block.5.mlp.mlp2_weight.blocks", |
| "profile_dtype": "FP4", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 128, |
| 4, |
| 16 |
| ], |
| "numel": 65536, |
| "nbytes": 65536, |
| "tensor_sha256": "896d96385e51bed7a0ccd2538525787564ad235f32d7ee389d02f6dd5dba7efe" |
| }, |
| { |
| "name": "block.5.mlp.mlp2_weight.scales", |
| "profile_dtype": "UE8", |
| "storage_dtype": "uint8", |
| "shape": [ |
| 8, |
| 128, |
| 4 |
| ], |
| "numel": 4096, |
| "nbytes": 4096, |
| "tensor_sha256": "e6d33e18970c9a8ba0ab4862464ed1e88b61519b4d9637ff2acc405d7cd28aae" |
| }, |
| { |
| "name": "block.5.mlp.norm.scale", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "50323ecbb641619272668d71d2c2bdd0b2c59048bf4dba9913fa136faa6a0454" |
| }, |
| { |
| "name": "embedding.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 1024, |
| 128 |
| ], |
| "numel": 131072, |
| "nbytes": 262144, |
| "tensor_sha256": "175c0ba5dab18fbf041cdb19362eed8820aec016a83b7dd144e965d5c0869f07" |
| }, |
| { |
| "name": "norm.scale", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 128 |
| ], |
| "numel": 128, |
| "nbytes": 256, |
| "tensor_sha256": "8740cb20ea1a531ef2f4cf3c5e3244d108727c24b48050f51a5ba63131082906" |
| }, |
| { |
| "name": "unembedding.weight", |
| "profile_dtype": "BF16", |
| "storage_dtype": "bfloat16", |
| "shape": [ |
| 1024, |
| 128 |
| ], |
| "numel": 131072, |
| "nbytes": 262144, |
| "tensor_sha256": "27061dff957ec99df5ccbc99fe4aa89c6645676611fbbf2bcdd80203c5fda931" |
| } |
| ] |
| } |
|
|
