| { |
| "metadata": { |
| "algorithm": "sha256", |
| "total_parameters": 17711116288, |
| "total_size": 35422232576, |
| "tensor_count": 723, |
| "shard_count": 6 |
| }, |
| "tensors": { |
| "lm_head.weight": { |
| "shape": [ |
| 32001, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 131076096, |
| "nbytes": 262152192, |
| "sha256": "86232bf5f9a204c3f7723717c048c4362313fb15cb0dd62d481c6c10d2c0e136", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.embed_tokens.weight": { |
| "shape": [ |
| 32001, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 131076096, |
| "nbytes": 262152192, |
| "sha256": "d6a3261fadd35150bc889fa926ee72f80bf421704de70f47a130fb8c902ad914", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.0.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "7cd6d1ff65dc35d324ec8a57fe9976a650e9a015e5b5b103462feaa2940fbb56", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.0.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "9f5b8b015a11199b95c290b586c32117775f22033ae82535de04ae5c3de9c457", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.0.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "425dd6e65810e4ee46b23752f397a28d9a67ad7a6d6ca605b77548d7569bb0a6", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.0.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "47fd00a06f47338fd68fecd443fe5f7dfeaa5f18f616fb389de388d94493041c", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.0.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "b915c418616a98533365d630032b4e202cff222fa352cd05eb516e8a2eec87e7", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.0.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "43969ff621535b7c8f3323aa76e43c01576b830106701e3e5e42daf4a08c97f6", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.0.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "5e8752dfecf08734f1502faf246de2118d7c53cb61e8a85c312d6e99dda76746", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.0.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "c78718b1a1cffbd88083e1dd363fa933b9c929c5e1117e00fb992b97a72bb463", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.0.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "c2332b9bfb2d1fbed4248048d3fde7a69f2b249559af6f1b55b239dfc6c2717f", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.1.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "4e2a337a0c066937d6e4bdeceb0f37c5cb41a9ab3bace9f6422d66fc340102d2", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.1.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "0f84ccf6bbbd06cdedfcaa8ba3d533fd7aeb149ba65155cf05a1ff067be0c9b2", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.1.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "3f2af1141dd8dd33d68d0fa150164f22166dda0e0a3b1dd4d696068a5d788144", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.1.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "4f8f2e3334231719a712f525fbe33515a28c6efb81a52e29eebb8dd66cc37f5a", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.1.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "bc9d11955a78f3fcce33f483f54a42db5db99c99400dc48915f868785db8ac03", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.1.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "2a89dc02ffe4f7d95a9f5f5324dc2bbd9c6905085f0392e5e927f7e3103b80b9", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.1.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "8ee68986e45519ad931bc3213533f114192b81552050e6859ae9afec73e9aba9", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.1.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "b7d08c22e065725abbb44604098db8968b9138c4656084cbdc4940d55fed9832", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.1.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "e33ccc6394b3ec4c86b9ce9f7d0f8f47e22e292a58a215bc9f14f61d272baa32", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.10.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "b3e0050131bbf2844d1693b28fc1a5cd4278605725f3f826ce6c48acbeea2a26", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.2.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "0dcf0c8a28783cfef6928bdc7dbb0b06d7b7ebf5a1f3a67f4aff5ca1488b5a55", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.2.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "c4d38f3a6e8e413afe957776d18df8fdf816f5e0dd59d497d1230b0ffb091cd0", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.2.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "e22403db6e6c11167f48d6cef9e17a1253fe30c3088f03955a9f561678f4b96e", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.2.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "a49f7a4e56a4e5847dad3749256508cd7f754260955ddfdef2e907629544950e", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.2.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "381aca6329ae1e21b3ca715156893556c2f5158c3910350d6f7be133deb3517c", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.2.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "10a065ea73d33139bfcad10483c9c6cb14f01f40b1b28b19bdc151763ae12ad9", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.2.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "0d5c32822ad8e8c83b092bd543b7098407227ddc322f6056ee5f4e3530c5bffa", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.2.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "da7e5ec9da21e1c53ea6d9976d6fd55a000c808d7cc9c9583ecbd0fbc3759cc2", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.2.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "84449f1fcc156b12e077b226e975bf032a1bb13794cff18903396ad7f6ce819a", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.3.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "5bada1cb63bf326325f60387116a01dbf58684eab0a21e67143a42008af73a32", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.3.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "e42f7925a62d998f8e200653aeb9135e3a0bdbacb9656f5753eb6dfa535e7476", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.3.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "01793d8ac933d813ce16d259be7aa6e50c1b901c00f987dd33a938c56d0eb56c", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.3.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "4155b744aa5ef74bcde6905e80ecce6bad3845ac50bbfa3ac2be78eb3757b8b3", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.3.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "70fe86f8979a8af81236cdb78d150927cf785a7188d8cbdfb2b2f4355cff7441", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.3.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "f332230e659eb96642450528d029726184257f47409b5927c6c2c91085fd003d", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.3.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "70e72edd2df206b15a6b53b782d69b80698defe8f0ed3d58299707c5e7ac3921", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.3.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "5ee40d7a333b5ec6412b65bd80f5fd0074753d3b36b5ca7efd47e563e951f89f", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.3.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "83ea13646614f0451f5f478d09f2ec53034e0db11ff077c390f1dbff800736da", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.4.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "15bc7f5a43fdeb931dbec8d205a5f7aef1545ad40eed8cff257212540945b300", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.4.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "72ea52f663d8dc8d5d864465c546b0ce34cbfd693d07738c7e093889f9968f44", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.4.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "6ec8ac55c9b0c1e3d06600c2634a8403415653974096a13714b8a6a7cd701e07", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.4.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "6a097522e728be8ab7b2bfba6619a3de07f7b172e95b1601ccd34ac785b1154d", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.4.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "25a2226c4513288374400d609678e3863af780c19f43767ae93adac2f117d710", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.4.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "a7dfd6d1ef9a6773716844073500d34d32d607a57f48b70c1c25e9c8e8e614fb", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.4.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "9f4764b789cff7c00e97d19457a7b78824fc1bf4a1c833a761975fed8b4310a6", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.4.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "df46ca366ee201dd8af7c03bb4fb40f1b3a110e4dc790303f68668302afe73fb", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.4.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "7ff58fee6e168cb149fc3174b2b2e1649d4cc3fcf71f37144f25ab92c01eeb49", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.5.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "d4ebcfc4d4137753f434da91eddb3bbf756f3df60226f86d73b7c5033b607e4c", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.5.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "9a4d540dbf53c2c7ed9a8f197ead3b0a643979b86c7fb22cd648cd5a8aecda11", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.5.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "35ad28f6619f7d1e733608c88bfd6c34cefa1ab1a832e7fea5e851ce29ad7db4", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.5.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "12c6f47fc9f9cc09606c24061cd232eacf62637ffd030803cf81d8d4899aa3e9", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.5.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "cabad238c87eec2915b9ff630395b11d297dee067b1d1c438114d801afa5b3a4", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.5.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "fc103c63e7084ced3b601155af84ce7d6f1cdd0939e77babc5581661b33f0a5a", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.5.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "6683d59a7a18e0d77ffdf4af79557eb518df3929e0384e476c6750d067467e49", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.5.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "5252a325ac1b259094674c209f20ad44c3e1c6070081f30a15c3ea999dfa624b", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.5.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "d1e256b18d147135b254f959367346d5f252b7cce353c425fa35f7fd1a01dae7", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.6.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "1027015bbf82171d90a9c299f91efab4277a7534aad9afa5dc14365051ba08ab", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.6.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d5dda43add013781de0db3c642b38ad062228e13ab09c58d768f5f569446571f", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.6.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "e75881bbae056a532a856d1bf88cd3fecb32e4ab348c8caccc7ceea38a3c091e", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.6.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d1aa817049c6ec0416c65dba671601c83d21bb1e99cfd9adb254ae980afa0311", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.6.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "ab835c73706c71ee83caeff8e8aa38b67833ecb3ebe607b970b7600752fb20b4", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.6.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "81eed13e6e461f5c1c2e57ec31868ecac934fb3d44005e48c287cbbdb73b5928", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.6.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "79c8bdf6873f18ff974e6d918871d48899e02aecb168d704cd8f0840866802d9", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.6.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "ec977cec69fb7934ed13e1f57edd61e32367c2e1a3cb036906fc3359465de94f", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.6.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "6f30e1ad6816302374e888cc9980d61b0f8b15c6b1bafb3ded274923f60e435a", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.7.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "7918407bce8ae774d9125181c251e7ce6efb11dd834e959b96c4b05704ce32f1", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.7.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d48fdae41dff0fb93ea2fe3369090bf915e070a8479c7599ddc547a72babaa89", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.7.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "5f8bfedaeb70f78c6d8e002974d388026452c56d9109c8a9a9b895e1bf09a499", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.7.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "ab91f4704bd8134488463518212fbf9286c6bcce88908ec4f677756273083d48", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.7.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "2532c57c67e5fbbfef0b55bc99f8fe1d7aa39246010d5e9632842725aeff2f8a", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.7.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "7ec2ce4a8e33568b4f3d959b8a559f209bcb5472a36496e522bd0ea360cd43b5", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.7.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "55416d9d7225bd2b980bdac392b2227e670bb5442d4c20ccc148df1cac811e3c", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.7.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "2e85f7d5a4d05640c1c94526af1778b969e4f3c29f6798974ac697091af8b259", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.7.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "0db19889b9c8e1f85ed8ced05104ed1d75c14e18fe56f15c64f11950c1f29880", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.8.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "b5e41af029dd0ae99c54692dd25583d992cf094080001c4d881fbe6d1d6a366f", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.8.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "039c60983e31a6989636d25f63a9f40a3472f28d46e13041fc6aab8f88b189f9", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.8.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "bd81e30e959ed34e3ae036c5c28028ae771198c241a4334edd7baab9aaf9bc3e", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.8.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "6ebeb34532c6b41f519db42b5ecb886063a7420b7e329ce612750354ae10df6f", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.8.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "340bbd11d9fb5dda319b4c70f7b0d897352fd033dd0802e533e2a3c2316a4316", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.8.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "0c1c5bbf169a56df568b5dd33c409e293174f37a2760e939f33c5725fef66f54", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.8.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "90dbb397e714fee4826e5a1ad9977e3809aa91a6df978b6157cfffc484a5a7d1", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.8.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "6efe62ed7df551754715883437d836354b7f892985593cee00814562ab6f6e45", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.8.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "706551c1ad1d3e1e8beaac91b69084751ba628c8f684b2275b8870115389f06b", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.9.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "725f196407ce9338e41633312f5a90408fea64c4b09585d80ebc303dbeecd288", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.9.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "b27e5d60aa4064702611ec1d432fc9e28d4ee6033d5252dd94a88de290675924", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.9.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "4462de750f9806298ab79077a0f8f7ed694b5384a07dc154bbc97a1d74943d59", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.9.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "500c86eefed2abdfb4d461129af423fd4d0fadaa3d1d14c0c64d8e16e3f46ae7", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.9.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "a61a7b82fdffff03f4344b5c1715e77a5b38afb17af0e759ae668a90cb0922c4", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.9.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "99c15d874f0297ca4c0613a54e0725536f33e50eab56ad3e9933d22a8b191afb", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.9.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "1da48e03dee8c05c60ea91770d2d83e6b50b2ae8311e9960fbb86926a6137a75", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.9.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "97bd03e31202c1c71c2a202ee66c72794736ba9db3c7140374674d5269f4d657", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.9.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "a1a3c5504824b6c6243be4e0a8ceb99dbd8a264cf6e3a9fbd6b144352b2d2391", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.10.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "352478740da3847d5d7938bbcc34665d0a06c6793657608a449d79293851b36b", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.10.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "61e0980d0bd3168a92211170cc1a1020a9df35c9d25b210fb8b5687efb918b4d", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.10.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "db3760f775a1c9ad120749fca33c39abe31b692c0967670e96e4046e829ef15b", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.10.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "52da019b7b4adf2b34656cd104e9692eae688ca1ad1f5da95e20156d801064ce", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.10.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "175e34239fb41728d88a47d84c8f96160211e432ba03b4d9e6d0962ad62bb28d", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.10.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "ba071f5491b800060af66a409231d56fdd1b828ef20b896b4832dcb806740b7a", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.10.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "1624876806cc6ed27d0b7a7170c119e816a0c4a8f08a799c58f649978e83e179", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.10.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "d5ac1e6c96cfbd491a1b870f2902be62eea6b659732e200920853b4fb3498adb", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.11.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "95ffeedf1e01ed65f3eae5dcd4c2ee9c75f23570e057b76e597711939089cd3f", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.11.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "c86362d6f025e695b05e84d111965a5d0683506083946ce8007617df2a94158d", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.11.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "cf6709fd3203f4df32d9ee77505d169fecfe891327f17bfcbff1a85500c59b0f", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.11.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "deade33894433dddd0de5c6060f21f196f2712a5ff317bb00ee15ecd31e552de", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.11.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "f92493f56c615bc60ddd77a27f01cc17bccc952d3f3599e3c41d438b0c3f9bf5", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.11.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "715427afc9b0db7d706ade5a7b5d74f1ef27c1f591710f9f00f700ed44d766d4", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.11.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "be9e93f9436b06956b572729a2dc61ba87804f57d515fbadda36d035006d57f3", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.11.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "f54a70e8e4c8650c3fe68b64ede02dd995148a6b4f89e314a293f122ffb988c9", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.11.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "36aba87adb810066f8d995bb6e74eeeeda8b61e231f1566674c5341eecaadb32", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.12.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8b02e128a08aa9949da6b1c08043c3b5cdc09541846dde0902306319710c9aac", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.12.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "691088e9dc0263e5db9aafadd632b47be15ecc6b3308900b2cbbdf02dbb723ed", |
| "shard": "model-00001-of-00006.safetensors" |
| }, |
| "model.layers.12.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "8cef34f402e59c221204ff6e091c06ec2c3e8049d4dd6e2a708b82489154096b", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.12.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "2462bfbbb804dd82c9aea57fadde5745bb065287144bbff451325325691ef90b", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.12.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "4f1af5912386c1930656f7c3831df7e0e6c12923b6d54012bedf8da762f7381c", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.12.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "7831df9f2998b7b98138e68d6f8b602cb37e55e89faad075467463a8e3bac4ae", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.12.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "9aa7bf801b5baed03bb7dad3855332d8f749e0c717e188960f445b889e716121", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.12.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "7a8846e5ee409fe70b03f29cbbe41cbb488d8a7d698983b2b2a4799f89b48a1a", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.12.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "65fa34bde38ed9e0d1dcdf3802da16e5e54ce02372584c4f9822a5e64218d752", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.13.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "1f26afbd6c6c847f1a87d3eac99702561c03b92a5714004040a275a694f40df1", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.13.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d50640ca21ce078d6bba16b72d7b6a641bc1e6815f83b5f58d81140400217061", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.13.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "7e4fb7b747840075452f0c53b6f012099c126342f50fe7334aa73ececb10c2cd", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.13.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "098f83899481cfecb4f9d9148206bb547931b9efdb69e4e285b96edb6f544c53", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.13.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "46a75a4e9ba4b6703bb8b8e63bd8e31fdb5fee7a81c10c7d9f70af5d3b36191e", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.13.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "9631cff2ec5c473a0795d5e097a9a1b010ca7ca05c1c251bc0bd0839f9f9a99b", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.13.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "221f00efdee86ea1ff6f403da8b8d0dbc46630c25e41355c6f2b996a71794d95", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.13.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "3678c348b81bfb4ca3be69e7a060e54efe9c55259c8602c70bca846a585c078e", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.13.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "285d669a2cf79774816a54bce8ec6dc5aa43c20b9bdfa77410d7f74ed9fd6d80", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.14.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "ab023c9d09b8e40a9f3615bc161d18483a547df6087f4a5167986e6eb4a2c2a5", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.14.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "25402ccd6c1df0dd3ba8f64acb755729b34cc8fc2373451a331e8d7e85a91ae2", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.14.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "8157df683ed002301c262e9844cd61fbbaa49dd5625c1f863f7cb63771bc7dcc", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.14.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "ae1c97a350095ebfaa9c946d074ef4845797588e3a70cc84d6b48666d628d5e9", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.14.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "65e13eaf21667b954524c1f8ad24551f928d470d65a4bed03d646da0b832bda8", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.14.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "09241d5d7abbdc740bea991be477cf136561afd3b2c603d488fc6003c45f5584", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.14.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "ff1665b94314a12e41c8f5884b3059c723bcf3878f6478f3a5ecae196ea490e1", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.14.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "dcc369fc32c85dbd990ed1da556e15babb0e461c7dd699b07a75d34af6d32ba4", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.14.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "79421c5f4fdc219d11cc99df0df4e26e139fb34656411a02dd3c26fd807368da", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.15.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "601fa12fdcd23664b5f33041e1ea824391faad593a124a9dc0b552c8c1b72dbd", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.15.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "bafff802eb23d7a19c2ab1049c3c64a485cf115f0e9ddb857cb535044b8ca53f", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.15.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "55fe2deaf37d24c5638d84e250c8a4bdcded86ac5ba911245a506ec38476cf0c", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.15.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "3fc3e5f8ea650fd24991693977cfeb9526b99c98aad77d409b683ecf0ece9087", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.15.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "24bb98ffb5e839078d0a6fb1d9b158dd2245f0366b4460bd37d4bf670f4e902d", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.15.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "837e23926680e3d9142913232c07bccd885d3e2d2b69fd3da84cd809c2d1fd16", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.15.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "fc467659b7480ecfae7f484842e5bc85cf72621057845ce5adafdd3afedc68c0", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.15.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "11bf0eadc47fcc88a6cf2547af27967d17ff62981718030f1c26869143b88688", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.15.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "be3f1d1f6f1906b02e394e236a47c0156d549e8d3310d1ce7c0e0dd71066546c", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.16.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "def713092b8a9ca9cbab2c4500e00456faa2808718cc5d3a20324997837116f8", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.16.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "29d301013f9b050ec0ace6b332b91696182a868fb2b5aa613e3fac35e58ee732", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.16.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d404e36490f473fe6e1aa648c6236f5a94bc6a64254fc579cfabf5523b4abf23", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.16.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "1878d77ad654e7ece8f42883adcfa50876293c8e09a6b5dcb612ababcf4ab611", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.16.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "f5e2bdab7ab3e26aa57c877b34afe72dfa5dbe93eb115e17ff215737b68230ea", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.16.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "b292ee28cc9b7d276cc5f0865aabbc6362668f299de66022c233f490e4cf169e", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.16.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "edfa8a7bcb9fcb529aca432b27fa49ac585d033f58a1ee0e27120644e64f1bd8", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.16.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "7df4c3b746eb14775f5e95ed5d49af87a5c11db38a03db1c34a4188eeffa0284", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.16.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "555315df5d635ff9efef6fc653c541a288a2244c2c6d53e6fcef0386afbb564b", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.17.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "f2272cf1f368316cdfd5b114f4450bd5ce98757e674d6761f568b2c24f5aa6e8", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.17.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d62972f3f7781ccb50e401eeb3fd00959b160650e751b1f6f33c2ba11616a3af", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.17.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "11437eec5f950f55fb99a1eb491afbe0097697c69b0f6d245fd2a398ee3640c5", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.17.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "13ede683768602b0444e1fbdb141d91f403d2b23961f38ca7782d88acef3df98", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.17.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "391bc8983beb7d7af9c7799e6e5ebbe7fab8bee6fd76b6cf9f757597fe5805e5", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.17.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "e40eba6249a9bdcbf539c80fa846eabb185858a16cc1a4535e5af3455ee01dfb", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.17.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "da49ada82253f09158c9d171c18c2ee68f117cfe8bfbe6783d14984fcf3ca5b8", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.17.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "eecf843ec7870ffe7992c583b26eb699944685ddbe2cfdc569b5ef17aa0c6f93", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.17.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "5ed54e271407a9dde3aeb4aea04ecffb4023390195637964690171b113bcad94", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.18.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "ede33bc07acc58b753bdab23b1133d8c6b4cde34594ff0e15da691c25f6ca9b8", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.18.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "f82b2b71d85a7e2dc727b150d254d9fd954fc76898f324c47dc808a891f139c2", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.18.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "7f4b5b237a142d10ef073970d45742bfb8ce23cf5e8da8c536e31168ba1df6d6", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.18.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "4a2eb7152b7ea510bf2349baf8cdbfd8b7bf748180992a7af8717b65939d96cd", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.18.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "e3ca236439b27e3b4a85774ce921c0250e5a31172b9ddb59b71e35c64446212b", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.18.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "0aec29cc62c2ee2ff126d328c1efb75b5482a40bf72b3ea1fd882e27a9831394", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.18.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "68c90d2d2bfd95cc7605feba64ffd3cc21c6e0d0e6ef128b4878335a657d84da", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.18.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "e43392127d00f1019b5f8db4020fe0d2ba295fb4d69b5f4d19ce26fa0a6b000d", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.18.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "03194fc041de034222a9065fc57efd3b0ce019022b3d4b62132e16b0bd088426", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.19.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "09a23395c32621667cd27ba893f0c1933e70caa1431f3ea0c78b36de7d5b2fd7", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.19.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "65d469fc94b7608239d67311ecf35c73ba80e95055967d144ac33fe5454e7acc", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.19.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d002d0021397325cfded54e85d319e836b4fd4ecba162085c4fd9e16291e15e8", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.19.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "62656d90fe9077595349d5fa09f9960d4dc8eaad35aa6d422f36b2544ae3a392", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.19.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "c9a1839de22e2b41d5b9e17e43e6ef0b4ff1c78b2e8fecb995e7da530ac4629a", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.19.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "d478829fcdad403b6706c0a82d131555dfdcb148ec8ad82e2a9fcc9e7b235a1c", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.19.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "c2f1ed9ccb020f051b3426922305bf10606baf8e3d824ef366d01647b9267fe0", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.19.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "fd93c62c1353db9b5eb14fc9490b48d31515596de52dc6c49c085a9cfd85cc4c", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.19.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "a5ad563aed0028324200a9044c6e1d294904da06dd8d98d2f689b1a3149ef6d5", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.20.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "ede7099735ad12ed824c3bbc6c3070b6116ad958193e2e4a2d4bdd3ee0cedd0f", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.20.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "8dc411df57da23e8e7edc631bae08b189d1748c3b2101eee0d9aeb5dda9f9afb", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.20.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "fca7b9f9ced57457c3fdc680f6ff8b0c54e5bf519754f5adea5eec1f2816d48a", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.20.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "5fe6bbc1c895f387e5dd4a9e359c9afaef9cd8714cad914a0faf7b58c86a7f6e", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.20.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "e5439a9ccf4b1d3bab05351070a46d4e7fa618ed41b36939a3a94509ab927e69", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.20.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "89b498769f4aaed84a6e82a5d5d43dc48adf1a6289be963c71580f9966b39ff6", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.20.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "30514191190a5a57a314a04e2cc6b5cc86208106f02c78ba3c8671795b1d6a64", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.20.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "c15f12d4add13acbe245d1f71a66ca1c68f23eef63bd6fd0321f5ac0f5a2e849", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.20.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "40c2fbf067ca075f9c2534eaa9742b35b50e2d09a79703842d96fe4887edbde9", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.21.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "006b280a13de7ddf7fd49acf650d586cd9589c7fdfe5e54fa9933b50a4dca3e5", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.21.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "17efd9b9fde3ae21cf9296a59557156dbe63761c23ebafdaf760c26e097b2ca1", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.21.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "95b8fe4b5fcd1e2054b3320fee9896c5006d47c59ec9685102a64133a64a5a74", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.21.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "c27480b2cc57045897e452f615dd1f07a6fd7c67097ceb1eca0156cca1a74e69", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.21.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "fc07f2a25aebe159bacbe477a5ac106f364d28be89363318450f1935bdd05c89", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.21.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "abcede66c2445f551d778aefc52b6e58860b9ef98184b0d7af703c5c3d7c65a0", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.21.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "c924b76cde1036ca38f5933538bfdd2ee527e73a15fb737715cd218db94e9fab", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.21.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "a9dd5687beb16055c3086d7f874f2194a71b8a6fb712cb043c040d3d19e7a4ba", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.21.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "44011af895564ddcaa61990d16253c2bb6fe9e65dde0c509b5ac72f6f92d14d9", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.22.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "298fa5f4a597d4e8fc01364745c570bda4b94790d7c3aa9f371c269e666b8506", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.22.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "2be89c84f7b6bb35006259deb2a973b15fb290d95001ec9c7e98bc36be5fa46f", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.22.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "f42f8c34a73737acec0c8c5c509aa3355d9e1e078c2f0989e24cb17759810c12", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.22.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "54a77b71e41d29abba229fc19af5074a8c13405ea6f9bb703382a1d4bdb191b3", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.22.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "7d887b12966c51f9bd45969ba7d6d99331aa451a9c44bce2e85b6134f7e451e8", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.22.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "7dfaf8b57ed143a3b62d4b98813069b81bebe997d16ae23bea30647bb1e391a0", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.22.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "79269a0dcd6e3c0fae034d6473e000591ea5f0aa606cb4d8b61f52a1f0f77cb0", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.22.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "8bd2cd8536f5a13fd16f2a9e2ef7a2a4cd80d6e78555b67d37de9f5dceb9a97e", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.22.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "9d5ca9a5e79f46be828a2bb4729bdc7645e46dbaa14172697097ed6e0021b33f", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.23.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "d22412918c657a1c1eb90018208aac3b79e4135b45d70b7bae3b1febba4bc3e7", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.23.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d34bd4696673736766e68d283bdfab9a9a640c128ab353adaf8749b33db0805b", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.23.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d56a6d3c70b6571539a5f5fbdb6cf0ca58ce10ea2a3edb14468ed89da1f19443", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.23.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "e1efb20ff857485897c967eb8c22dd3fe1131da2843fa3966a1a102713bc6c7a", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.23.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "9f3bd01bb46701e29e3a4a1e2ded6fc79d45993f935ee24bee644a3db8c1dde9", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.23.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "3def3428423bfff373759440c2b09aff9204fd5c3ff4882285d4ecea1c7a2b3c", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.23.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "36c5e69288592513333e97f8ee431ca26f9a6afacddf8032ca34698acee26a2c", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.23.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "512069f4dbcc910effcce6814b00c888c67c0c2b09cbf2cb359aa2e1bed20f1c", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.23.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "a063d57c631f82a9642b783fac8491f06a00915952a0c45fc30ae8f0778301e5", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.24.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "eb0d26100ae6daf57c56478d2d9fd429dde27c1d736453e9d68dd5bd6d91614c", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.24.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d227a9d4b84bf2e9ccf66a849e87b3f12329f51362e6f0443d335abd5dc2a608", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.24.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "85f7c454ffbfc444ca2cd2126b063055d2a08c2f0f180357f8cb14b8033348b7", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.24.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "0d82ee5d58033cfb44887a38a0e7b9b9b0c551202b433add4d316c5c2b507fa9", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.24.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "2ea58b7564b333fba09e5bcb7197af253979ed2897b6f6a0828ee554b3905ffb", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.24.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "ac602098d44798cb975987856d7ac7d8e31dc66de1d2d8d27a838af5209ee738", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.24.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "e4a39db651e0fce698a827ddb90bd5f6cdc0f7f0559478add9b505374fc3c49f", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.24.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "b00c606dcd8616e04b5ac772fe6917a5e290311373181aeb121087b6241794e5", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.24.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "58cd8bba9ae8f85d65a6946e48611299b96d51f95527bf94ad8bb56bc09b50e4", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.25.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "ac57959e0ec3f382a08cff6beb578c1dd073d1d90af875e7801e9f8261ab4feb", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.25.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "1b076db8da6bde3eacdda7197d83732bca26a80d43f9acdd4eec9a15c0ce75e0", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.25.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "fc924ac907c2d4fc7d8abfa1a67f3812fb82ef802cba7909e2fa515796ea7331", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.25.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "bcda6d4cb3d33c591f3bdfb258878bf77497e6138194f85198888fc1235b1651", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.25.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "3e274d98e318366924dbca962de95177ab7ee3ee3d65e2ac59c727be15ce2314", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.25.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "cba023962737f81cb026197bda459d9cb9c31a505bf88491a1d0171d9ec430bb", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.25.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "440b49bf0c70c4b9e16d7cce96f3ac10b7a8116d0a4322accf03b9c9114f3535", |
| "shard": "model-00002-of-00006.safetensors" |
| }, |
| "model.layers.25.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "3155fec58e6182d9740445a2afd929146728e52fdd4bae6eb70fef21c8f54d24", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.25.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "c54cfa193292fd6795ddb6dd3ced10ce04636595ceb6109fee26e36da045afa2", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.26.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "9b2916379b773f495634fef6f5d7710b70b0495878424d98c7b40ada696fa7fd", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.26.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "3470bc52e001b591c285b038fc5e89b49559213564130858c1e94a49ba5d4639", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.26.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "bc7e89255f61e447b2bb8d7dee61ad48277e955304ad308ac35bbbbb5b7b069a", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.26.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "541757f871836653161ee23de9924484dfff9e6eecb90234d6c2cb1bd9975945", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.26.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "d9c46479c68aa5faf57f41e39b275bb9bee88eac43137c0d7e88f9c01a7d63c1", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.26.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "ba285d59a4385783ca04cc92558bbe951085b725a552044bac1df279a5e07eeb", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.26.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "5ae3c99fc74116293c5630b7eeb00320ad3770e5f413401083f11602942a75d2", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.26.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "bdd2f6f9263630f1ff25aa2a86b043ae8aec06cc32dfe4c603931968044d417c", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.26.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "434b3ea3e1990129639b397638316c1d21287654f003fc0f7f25b45dba8463e8", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.27.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "5d96c348c65ad8ded2d0fe604c9dd62c24b9f160e851f4839ca1cec099ce3a5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.27.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "e09a1f5dece222fa04f138134dae3a193fec72a68428df3bbda0d35edeff9d84", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.27.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "bd38f4efa7bca4a5bd58089befd7e99f26aaf1b93cd006529a30a79869021aa8", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.27.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "b8419b1df0631f73b74a560eeef84fc551242a5347661d3c2384461115178da2", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.27.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "0831c809e7a7add02bc5bf640ce316c21b07407ed209922d2845dc1ec854b225", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.27.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "870ac77e5a419b70193af8ea0932293bd99a3b8ebfd1a081d92b79e4ee83912d", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.27.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "aebdac9ccf61da0d5c4abdfd45f2686bd924d462266ef71f9f3e735b6039d503", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.27.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "29fc7b8b3e1a19261a8effad285e0fd33f973925203c6451cd2a1b925bb9d73f", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.27.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "f594a8a8fc221f7d41d56ead507a57550391d1a1726a47c454497f659e554fa2", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.28.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "b19b872fc3de6d8ca6784238d7bb77260b4152842acac84af5db488f7bf51611", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.28.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "bac05624afe0f2b1f114b7969c33485a132940896b987ba5f50722a4af6ed5ba", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.28.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "f73dcd0a38e9b817df1d4d6ab955d9eec436b5efd0082371766ff83c6829a363", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.28.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "6274e7b59d52becc2120a740dcd3d7bce26cf105c86193176d6a32a9912b12c4", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.28.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "04e1f109933547f6ee72f1217219c87e4ed35f1b5801c6d2865bdda6abaa19fa", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.28.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "f475d0b46eeb0d7d7b2a76a1e2946cfe4e9d24cd7a8c4e98712e1c9dcca9bf09", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.28.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "b23600af81f2a1ca54483f88445b77124bf86d0f72bdf2f3c5c140aa7931863f", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.28.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "7347863e167ad6f5b0a5da8770431750326a4a0be47058baad6ec8c759b60aa4", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.28.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "6b5a06ce89f5e587fa85d5cec45c288712121885bd08c772697f04719ce576ec", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.29.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "93c7d9481f8a474b1717b00111c3347ace0a3be4ef98a9ad81f3d81f8b6d654e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.29.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "85b8a104e464c6db2fb3386ab856e8f90116e8dbfeb9771c7fc444c357c0488d", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.29.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "1db08c9c1e3efe7f92fb79bd1cdcb9c11269885240dd2e8aaab3a8343ae131d8", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.29.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "2958a1d97ba2d16777bd69c996f0e722d43b80a95506f748a5651f5fa90f61e3", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.29.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "4a0fdff94176bc20381c4dc8576cb72117998993213204b4fdf611109e7ec93d", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.29.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "fb79bdb171d55f7004bf25d60673ef25e3b107c37e71cae447d24a6871a3efc5", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.29.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "78c3d3102e1de13298c2b9e8a468258c3c75e3c42d679a913f58aa879a13922c", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.29.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "d0e54363381899538db5880dad517df23cf982e0075f04b47e84b559f6435fce", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.29.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "545eca500c62902606faebf44ec8de8b1baa74c1e8d3413233fea9c54eeca125", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.30.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "e6a01b3ccc96a306baa9f2c40e9951f8aa6651cbf9d837ad29444dca7950d123", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.30.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "87214b0e33f50ff2889e90bf182e8dd5f0e27b7dc2d61cec6f4473d2c50d13c9", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.30.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "97d079bf3d38f5d859aac639ff401d332c659e7de8aaeccae182cc1f12f288fe", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.30.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "ea3415ec4a3cc201127d2d2082d5f42d9832d1bf8e6a85c2dca38dc20b0d509b", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.30.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "38623814d1d3c6e973f824c5574414df2e71a0a217539704cc781d7a6e809091", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.30.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "68bb3d0754edc53b0fefbbe4e4bb6dd005d1a031fc0fe5aa625ad7de434408ff", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.30.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "b1a12b50b47b0496f7fcdf5f029409ed687b16c944755b07889d0c6e619573cd", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.30.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "2eb18b46c540cf64909ed84a5c71125ce6a84e866ab6ddaac27ac4fa5bec3702", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.30.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "665ca1091709ecf8904f193ef1580f2e60f059b99b00fe6d97c8386d5fcb9b81", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.31.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "fdf953f2573185990bed33578a55e0067209347d84fc6c9465598021f99cce0d", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.31.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "15b253c7196e0019f66ed84af9392fb087755e4bc381e0dd4251a3c4e80bd30e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.31.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "80c8879960e641c2cba1475c5a024f01df1a579de1bd217d8605278fb329110e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.31.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "35081f5df04bb3324179a561cfa0c123baafc685f6d9b645774c044fdee80ffb", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.31.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "0cbe5a13c9299e7939513a8266a917a84bfaf899add5579dc5db76cb648d7621", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.31.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "6b13b3bd0450ba5ba23532b1d9e29ba4e84127e940a15c286af2f32c3fb96dc9", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.31.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "61ebb8e3f0a2a77d482421ec9a4e8a363203533b1b1c34c693e49c4689bc8d88", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.31.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "b659abada3ff2b505b8533d4ca49bed4626ae8c03ba31d70653b6a3c924fd1e1", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.31.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "8a8b81eaba500429e80d0cc25ba8f2d5a319ed0dfccbd40df8a38c0cecf78fb7", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.32.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.32.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "8967ea4ef712bc56fba84cf86d2e846fdc4922368349b925fd756ff2a12b826c", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.32.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d9e813ba68d0cc7993adddeb389854532017c02f0043724d9f98a0d03ac74b42", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.32.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "8e92faf0378f2b28b81403174ad63084e4f215395f161acd1b896edb5331e762", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.32.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.32.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "be6980b2229a5f3d3d99b1a596951d086780e0b4da38463c2a140c9bc8ba6b88", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.32.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "c73d71c40301737642c5e19ae540aa31d00f8d2dcc070f1d6fccb3cad1675064", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.32.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "2d0b9ab59c2e1fce03ff1e57eec46cbd73d1b4f3d0b8c46f1c4fd3b5389e50e2", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.32.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "68b6baf038d2e43f9dc953dbedc0f5bdf43e017ff08d499d0e796592c7e8ea65", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.33.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.33.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "97110ebc0f49b0603f3cae35d1d0fc3f3ea6850fc5d320662fb13b98c5ac0083", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.33.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "a0f2bc193e817d8dfd4fdc9bb24f02eaf7ab818213fff9906db4d013d3a6e083", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.33.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "c1a0617a68bb8d7e7b2892b0051b89fd9469d0b7e8036fd23ea04cda17e3a45b", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.33.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.33.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "772bb754a3ea62cd0c7896baa23c1f98c531d6b91453d130aa09f48149623914", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.33.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "0ceeef101388e77330db64eb0eaff29b007b332d9e81c68445ae13c9e02a4d8a", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.33.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "a14cd62069d8cc09833ffbad00344bcc7c3eae378fe0d5dbdb177510c0a2a72a", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.33.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "9111ada83585c0f8a75b89418422370d527b9db8b7cb686334d9a69875ffc442", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.34.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.34.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "ea0d974d160ac725a0de95093ebbb45465994d1e56f490e2479d68f63b7098a8", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.34.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "724d045afd937a0792f20c87cc63bfbd84b22b91d485fafbd6dbdc4c92d52dbe", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.34.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "18de7f864fb7d30eb75f40b63c035c543f1ac13382ecf04fce7a3a4b0cc8d50d", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.34.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.34.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "86e353efacc2e6960ca3da0ce55c4cb1558f4c662320dcd07f90d7515b6dc93c", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.34.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "0854cb8750e72c96aac71bfaf122c202df2aa70ef945ba1322a4303f9009f006", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.34.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "933346949ee0c13d712f818c6fa8f95b2071c34e3865506d090e5907b56e878c", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.34.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "1b257aedb96f4c3d80608d67d53d2858d1158d5e974e1a26b5ca449cbf446c8e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.35.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.35.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "6b023185444966d9108aa1fd7d9b3a95c539b6e86994fc28e34dc8f9e2b9a83e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.35.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "02d9583860c865647f2c6622d91ab3bd4102b12ddbb6b2e8ed8657fbbfac96a2", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.35.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "eea402d39994d5a34c49551c00b1878b2f73358b83458a34d38cbcba9a5710a1", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.35.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.35.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "38fd89b05192464a42efc347c2333e9245ce0d5d58d6d53805cb321d7f6708b0", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.35.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "bfe9f1bd9afdd44baeabd397c94b7a84d823686dc5783b49678eee1a5b04f82c", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.35.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "ecdd66d77cb5195be820a75db55e26587844982ef3508e567f968902f0c10466", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.35.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "3c461c8f24ae9631903f73b7ec079bc34ef872ae60eb81221a4636b9ec98501d", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.36.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.36.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "9ab2f5a527fe133285787499d3f552e82e4bd5195e3ea1639f2f68650d844539", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.36.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "0e5303e3f03ba3cab3c356675822bcf59910361786bbfdb08d2fe92075170b93", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.36.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "97616544a8997889741ad38ea471d190c1b825f20e95c53a4c549c69f4e66be4", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.36.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.36.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "6a87be1fbbe94771d6c2ed65edc924f2351cff486cca4c8756c29c00c05ffd96", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.36.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "4c34e7112598ac37d580004d5d0dd5a261aacf8eba080c480822a9159295aa73", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.36.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "0c9c24a6cc1427afd5e0f3a46e7d394942762c9f11174003db5f20708fba4e8b", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.36.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "434741c878bb9117d3600851512d34ddc1ae34f6073df8bdbc96c7e4bea6371e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.37.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.37.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "500b453e57c83f38d854ebf85385648101537392e54b629c226abc9be198faf3", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.37.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "43f44a505960270b11fea012d4d337d71c53f881c226cee793b120ad1ac95240", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.37.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "9e103947dc05c55da631f79ea162dc01a988d5201f0a5699c04221f0d03d0b7d", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.37.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.37.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "5a4bb5f5e53370e39f6206787e65bf1be80728952fc23225ea1058a380c23baa", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.37.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "6eb8a9a45ad088847870a5883517fddf8a3682287f22b001cbfcebb4644aca06", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.37.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "7a48c62c4de355306383f8932e03d71d5dc99f231bd792a4f8418d44ba644acc", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.37.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "5afbe015fd9cdc0d8eabf4c2da54d95f00d694b6c90f3bdec941b27cb080feb3", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.38.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.38.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "ecb149f65d5aef882657b0a053721ae94adb058a438c0d682fe4cceaea87d441", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.38.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "e480b971f787eb13f71fb347631fb2fe526681938980538c53041f7272da9685", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.38.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "17edb379b076fd2a3ba9ae8eb170a708061cdeeeafb83eb7bf2551cfd7c79eb0", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.38.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.38.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "46ecf5d44ce75b2d13e898ee3d1992c03ba1545ffb8b892f0b1269da7700966d", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.38.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "2842498edd9ff5b7a5a8b49ca7ecfbe10a9accc5c1c720567486779440a51a04", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.38.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "a23bf783d2e2abfdb5e7bdddb731771737412866a985d61bad875a6b23dc3bb9", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.38.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "369919c6e75c35b73e03eb264cdf1711865f54e872555b47da90202ebe562257", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.39.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.39.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "c4953d662e42e5e4825dcf4cf036d8bafc387793a122a64b1abe6ba29582f607", |
| "shard": "model-00003-of-00006.safetensors" |
| }, |
| "model.layers.39.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "34dc0d73c00d486dfdc4476a11047babe5a22efaeb2aad1df418e1386092b6d9", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.39.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "0a36068b201a8377b64350e3a19317e9a31c640087500c8e2b85c40ab4c9e8b8", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.39.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.39.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "8f8ed2e7dd1db0841ddce9ab158215ac790bf9cd5143db2a55a21b371cd0d476", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.39.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "39adf0c30d1179174c80aafb0ddeae1e4327aee94e2ee89d8e8c7c942b2ca52f", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.39.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "767949c617457191fa44b6aaefb64846447b7097c947a81537128295c6a842a7", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.39.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "6b5cc74b542bb8f90b5eae435b52f31f9eb8feb36cb3fd2e63ad6b5e328b908a", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.40.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.40.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "e7d2a3018ab6603313f9a4bf9ea5533cfb4359e41e72a4c7ea87d90a6c75ed00", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.40.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "4d1e97ff827e2f2f922e9bfb69b20341575fc53730e8eaae03c7728cef1c34c8", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.40.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "a524f2bac5343afc1058f2fe8a798834868b25a5723b9016ef7e909a96347ee7", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.40.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.40.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "c430c0c2859a9e846a97bf5a2fa776a151285c8d8a4dca3bee8dc4d24a97fe40", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.40.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "b9d0a07bc08154d69fe952373b22bd8beced8800353f45c2bd59a1f036ad2ad7", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.40.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "d626536a38be61aaedefe4deab84c220156e235a0d32358bd42e61d07db4fe69", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.40.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "1e7bc31dceeac57fe70f31a284a5fbc62e4a33f08e8232478dc87b2c675ab1d0", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.41.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.41.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "a5ffa735834fd3d77f04de0ba2798b4e4b86c7c8d4e0cb9249b6b4b57a843e97", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.41.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "0df696c7d1cd460a4b78fb3b0bbc01e9f9ba28035aeebe1baebad016ad682276", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.41.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "30334208cc09aa1ea6eff0d3572864f0251af767f09e30ca9bb3e676853055eb", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.41.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.41.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "3359f85857ff1cc6cb1774c3ee81eb27f9cfd16572c270867f7b6b40facff203", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.41.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "deb2264642855081630d8ea002f5b55cd491d91a78e6c18483f8c9cfd0d5f705", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.41.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "21b5fef2310bd2703f622fd3b1da10c8b6e0f8db8453ee5de3d19730cae7cffc", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.41.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "0b0e03facabeed1cd670e47130d1ea5b08aebb890790a609291ac3e8a321b8a4", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.42.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.42.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "049748f67650433893b896b6b9a07763a2880e51de696ec5d7e02172f073e25d", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.42.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "ea378a632a6854d67809cf9e2945dc6e61997316b0c352b528445d5df9d519ed", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.42.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "1da34340abcbbf10150dd4e5837f021c84cd7396e8291977bcd810486cb9fc2c", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.42.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.42.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "fac42b31cb94e16350be0b18b315585742e15f6717625854440b01b4c475e817", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.42.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "00fdf5494463ffc26174ccbfdaa8cd3670a1c15e1f16792dd5f31db7b469b337", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.42.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "c90f29a395fb816ced60044a23618f14edd689ff65b7465b28c6bee1a82077f3", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.42.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "525c0a930949757e890044fe5e08d0ff2d8554d5883fe8b0bd349055a5e37a9d", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.43.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.43.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "ca1df2c1c47d286f0c4d50b295912facf5aaf897a15d16dd4bc3203f9a271eee", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.43.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "82fd0e8fca32543f1375cc5b5f8526b7efaf51bcb2c79b63b0411a66a0d12023", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.43.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "a815c849d9a82a10025b89897280c4325fa4512251d02b0b2aeb3d61e6e980d5", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.43.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.43.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "e1d30d9b9a454ff87ed1d351b0da51c806b1fcfcc95d8c78c7aea1902d9da570", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.43.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "cf5afb528ce4a45bf8cde7a06712817453778b1f07489d9f18aeaf5fa961a646", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.43.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "6e27386f24ab8bb81bf8cc81e786e72f944cb8ba849203169161a10a156dd3ee", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.43.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "6671ad93e6644ddbfad507042dc6be07a2fdbb32aa78479251f31fdc6ae4d442", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.44.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.44.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "bdee5a6ce132e1719870ed16bfda9faf969a79d8e84ec9d0db5c4a2f0bbed597", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.44.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "6ed001665d2cc8d35130551139af61ed7a5671f1a5297cbd76fed22f5fb3803a", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.44.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "2ceafe652104fe34a385228209a6b0f89b9eb17dc77d0e2ce74c5a60218d3d24", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.44.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.44.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "91bfdbf7b6ed83f633e782d7a8227a83975782b5f6d0e8afb858d080b6811c34", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.44.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "5516e717b18537329d784a09243ebccc00476c58b673dbe6178ce6400d2e250e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.44.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "d22bbc828287f82be7efb56852879a48eebf82d01a62f199b08bea01e67b50e4", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.44.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "061c2eeca1b59c4599aff9cfda098c792c70505d0fb1bac0545ab7940aba17a9", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.45.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.45.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "4ed1ac9e9a048fda9156400410b9a5efd1106cdcdf65c8e60fc78994d1b9be73", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.45.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "7f60dbdc85b23fb08b3bcd171c536c9b1b0455dca3a960cf7db8b09241a4da5d", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.45.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "eb92ddac016fa4fd0c4720cc21550cea7c9818123a26332ddf783faf542ac113", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.45.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.45.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "35661b2e45de3f730afb2ee553cac9bae28f014dde899973cd6cafed7a4496ec", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.45.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "4404642d03606f3afc0e638ebb7f2e1de92b999aec784ff73cea7148c53b827a", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.45.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "3d5094bde8e6f1855e0c4be188fca5d7f80c63f93362b9035fdad75d22b3bad2", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.45.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "3e284fd1362397cd44089a28ecefb3647bdd7e6427ad33a2511196805e8464da", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.46.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.46.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "b43014d5729c3b82bda12aaaab24fb846c774a8eec1bcbca20c1dc735a784cd6", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.46.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "0ee72858c943f61869e70fc45770bb3a1dc9860cc542320f4d957c0bdd299c67", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.46.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "ce9a7edacd9dd72901f3dc37bd218421661393c6b00d4714eb895dca3db4be02", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.46.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.46.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "45d24f684f50774d9f5b5cef37c6eaec6c21385454017f4b138ebb8677003559", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.46.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "533756d396bc0d9a9eaeb5e69f9f4c21dcfe11265f650439d382e0ab4151618c", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.46.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "df9cb95689f0e82520d5d96842e7eb66f79b014a27fca9416c8f25c429fa1ed7", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.46.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "c90582084ed72be682217fbe09f533a78fcc6b25898f35df13dd647a4fc4ebb3", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.47.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.47.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "5898b99d436b829a5e0a4b8099f02f51c9521e0514837c1fc4c2b42e5b621f77", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.47.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "ae997b67ce4da4d4a694e4112439c1594cbf7c3c1307d19606864ea8eecc26b8", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.47.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "5b738e2b7b64010b0f0a879d315094a96840187db25ad3868a68360d2463c5d0", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.47.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.47.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "a83015603c1ac142b7a8b725ce9fb81fa2580f2e62b20399fe185ac5ff00a3fe", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.47.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "82a15739fbe0b1796d1c3e6f377535e1d857e627bddaff30c8139c7dbd135093", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.47.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "56204436eeb2304e5b9d8ff43c39f16834e6a6f8892355dcf431c01568158e02", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.47.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "78fa913babfc3cd3da767e4e13bc14990baea6b5100e469f0c159c4df0552c22", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.48.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.48.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "51561bcd7b9888f074314493263d4d1daab31128c8004410b2d488a10fc38a9a", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.48.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "e6e2425ad2db047c2bb3b8413db5efb189bd87037cd4d88dd577662c69fb174d", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.48.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "8df84ae9df1806e663fe3ad00f1f0750ff641093aa7531ccbf66a3c2b5a4dda0", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.48.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.48.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "990794211ba1c8948771410f1a965ba21d8ca63d4f9f8f53568430631479e0cb", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.48.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "54954ee2b974d95e1e3b025167bc678478259728bf4491c812666a76827f52a3", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.48.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "c188bd2b9f9574d2356275f95bf463b6c90569c0d83e3d4dd003db509f732fc2", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.48.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "5251670ed060251b12ba867da2727b031cb2eb57bb2ac29bb1f9e623900810a2", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.49.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.49.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "053ae596e01ce720c0ddc339d64a1c5dca21819e7f85fc96fdb1e9b0555a0010", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.49.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "3988d426d8277c7d97e6cdfb5f0d2e0a15baf7af55cb2f18f29c649718e1f7fb", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.49.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "a965ff818515ec93b52dc42ece8f3c09031684944a66e12142eff87e76963c18", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.49.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.49.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "f9d44b8b8f2e8da8d9ced0f5440410db2891246810795eb7607d13481d5bbd30", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.49.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "c3efe65a9522e94025f24915e20e421bff495909ab88fc2293649887f3e2a4d4", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.49.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "e936f8f145b99cd129055f22a3e8fb33f8c3485ab2240259c39a9001005d2b0d", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.49.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "8509bc192ecc4f634640471a4f2a5195e5e9b6659a87b833327f1b1d895b8071", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.50.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.50.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "b82e93eca5d8d21b55cc87cee2d166980c16cb9b203051996f548fa76ec6df93", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.50.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "68f94b8fc550a2dee773adec5561227d33cb22a83484798dea34fecc11d96bcf", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.50.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d74bd58b3cadeab0c6f6b5f2cda9b51226eee689eddf2160bde5b654d5392537", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.50.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.50.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "400659eab7f711d14714d38eb12811cd0371f9882d1a70e7597e58cbaa42c154", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.50.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "e6cba477e8037bafa0f89dceee3707e4e985588ba3f1cee665583cdc8871149c", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.50.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "c6d7bb5309afa285f69a3fba92b7ebf2948d79f79c945cd73c84da90dd0fbfcf", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.50.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "8c0f919f568a1252b124433cf5e43309f3e30cc86cfa444ae177206489b53102", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.51.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.51.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "bc7a2d3d1654daf40101ea82c3c7ed53f6f3490840e8f5c060fc1cea9807abd6", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.51.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "c76abb7164beaf11a2ebadded4ac25b70a1c2e3491ff3074206e0495610fa342", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.51.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "8c8fdbbb9a2172a2a8d778c539a3bfcb40d68dbaaa7dd3583b7379e7f6118c0d", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.51.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.51.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "ca4a9ef16884b2ab694be67220f373e76ec8a5ec96c7cbd14cc98c0f471e31a0", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.51.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "b3d415bf048a1866cf0747895047d57b0fd2ad6ffb74c954cb66969c8cd45e7d", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.51.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "534500b04ba792d597d7765e072b084480915ec4af5c885317bb21a1582eb9b5", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.51.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "924d011b030fbad79adae5c3b6e5dfa5c09c15a38322cefe295beefdccbf9f6d", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.52.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.52.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "9996ae856f5b4e443aecda59924e46999fd289ed758fa51f13a811a0b254d6c0", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.52.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "ea8c3d7d24f5c4c600765f4400c1c400ccb5c5613b71c0d0014ed97e3ab0b8db", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.52.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "a10d72378ebcd80167a139b58999e30f33f6453e0dd518bf6e1a1f53003a1571", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.52.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.52.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "77f40d562c73b4c69dcf90a31e96465d28312a9ad1ed4b594e669104beee002b", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.52.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "8e853d2ce78da20aae677e8971b5c066de39c918e7bad8a71cec4d2317f681f8", |
| "shard": "model-00004-of-00006.safetensors" |
| }, |
| "model.layers.52.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "265625de28f8d94a61604d8f9ac2f44a69a7afe3c5ccef07e654b6953c90e41f", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.52.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "a6ced2619578c95aa9136fd7577e06a5e002a92261f3c1792468897c258eba8a", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.53.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.53.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "b8321e957532fa9efa91a52540a640c63f82de2084933a41ae3e2cff691d4db4", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.53.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "199a857722c8cee510da6a9359d6bd22d804f7c071bfa1f1ff9980737e1254b1", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.53.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "4ff8c1cf8fc8787bcc11f0f6924d3bcb99bc16c03f298331e9fe883ea1215c95", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.53.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.53.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "4e8f6c592f1e55e38d96ac1de9d56f7c88b5b537feae21d9b17d5dec312c0474", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.53.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "85b719a2d10fade78110124ac15973d5b3c4f379b1da9de3f40034cae23617f3", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.53.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "662ffb9dceaa5bc51996660cbd66462616c9f1cdc6e6ab2ca1cf0e7b74e4d3b0", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.53.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "f192323379bf78ce2fe6662aa475a5231f47fa42dc119bf281f2b971e53487f2", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.54.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.54.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "001c1fa9038d85c863a63e0faf304c497702ed965d8c4de0337b85e3489b3d55", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.54.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "469617b1fb350a098a7738879e1dca4e15c9e2c243a4a4bc542f6f65ebf4d378", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.54.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "9938a0b0a5eec569891d6ed74bd202f8ce1051ddceda01afae16d45f8dba708e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.54.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.54.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "a54ce7497c7deec0e277fe4a2984b4e6c8e85698aaeab7b3e3532405475749b0", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.54.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "ebbdb0bd306b4068a8e21a40f1659a2ab7f30c41f6d926b6da66971c01426d2c", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.54.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "cbc77d06d3f2b7281a05210fb2f320bd089ca08eec880da90225156367e6fd98", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.54.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "67e247c86e5398dcd9b5a4a433f5d5cedbacceb18b95d5ccd377ab052b7f07c8", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.55.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.55.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "22f382f25a0497a2bb6591cb9f91b8a00cc3bf24fb4e91aa54fe5f78df315c0b", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.55.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "241efb1f8b4a0e498a2ec5bf7b96dbf90442e28c8c876db15f81002f963e6ee5", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.55.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "1f923ef6164b7f1aa885421620bc8c167a28f0876fc6f142e7c8ac41623e3b44", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.55.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.55.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "ede6e7156c2f20124afb4fee786665be231bb887d74b13427eb840a4409f254c", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.55.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "66acad5de2e8fb64285d3737f650de10de67dd9a634236d4d4b7ecac9f536b96", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.55.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "a2257394784ad0e3d7db2bc86828cfbfef9c56987a6757e24dd5838b2600cc6d", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.55.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "79b59a88fe8ee235d8b74475bf48c9d9744ccd1f3ea67164894b86dab17dbf32", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.56.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.56.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "df00cddf5756f80d94e338404f5b02799c361787e83cc3e59b351797a58b8e26", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.56.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "bdb0d6aa51549e30dc7d01e22a885bae2aefbe970919a9fa393977c37ec181eb", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.56.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "989d67e5dddd3598a125a1520c7dc0baa1931042841f12fbe411cc7c1e19678c", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.56.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.56.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "e82caf99c2d689afa4145a5052e6157602749496f4a114120d1bc4fa5a261e66", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.56.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "9566acc9c7a3509fdc60d85a8ec106eca8064b8a57a618bc569a19b4643d49aa", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.56.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "d292e6039d098abae54c43aec3c3fc7309983fa306d26740e937390a8c2225da", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.56.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "2a31cbaeede1d4cee246e993a86cf7348d282a76d917bcf665e4106a3bc8585f", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.57.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.57.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "642fed3c72dc75e1a93c333bedb0597b0220af1165101b5c6ec3464c232d480e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.57.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "b5db423718525a2dd8f94c2775759599cffc39c8fc98f402dd25fc910b59e695", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.57.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "8b7def132b0f2f9d59d42882abed38f80a93c480930657aab6f324ea45eac643", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.57.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.57.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "2c56d17f46c38ae4d5df34d460e8d9507c257dc33b6acb33d68285f759db9e81", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.57.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "37db20a52f55959e9f4bc04ecbf1b05c914a94b10b9f13f949d815570d4cf5b4", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.57.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "c472afc29c3b42af6f6bc816b1da7fade094c1ccacf213640e926e6c13bafdf7", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.57.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "ee47799e84630907b325d08d5ed4b40cb533ba99e3821be4e4383cb7b0bd6616", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.58.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.58.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "083e955bf0b1a6007011929146b8aca3871fd914ff23d7b3bb0170b964d512f5", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.58.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "2312e7a9431681764c3b3562e183121c85e92c355e6d3c608fe6c0f37e60cc68", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.58.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d00dddc52fa95ebd3dc1a25c00605e775ab48b1f1e218c670e6d00d12d654507", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.58.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.58.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "9ac8f6175de4df0fa1b210442b14b0925129efef1394dd71cd072b55954dd5a7", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.58.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "e66fe161f840e78ac180d72db394cc7dc0f64e8737733a1afe769f4820a29e8b", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.58.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "efe48f8e4d661cc41b1f088631312d0df47d3fbfc4c3953e06255cf41efa9546", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.58.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "b940aabf4ce574c2fd4811671df1c8d9d2fe81b26d407a971b18cdabea73cee4", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.59.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.59.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "91410df28c527bae70773fa06300322f9dfda3fde6df68e853d652858b7325e3", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.59.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "91ffb9cf86f8cbf6f5b821c4611792c330f102a8cbe1752f8f3cf08f3dbf471f", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.59.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "ce2ad445627a7d30ff66d221139eb05c8831b7c45d525d21eb95e45459e35c53", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.59.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.59.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "c16aaf6d45a0271d258664c18c7858cb3505642bf8264347fd4ca8311695f45a", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.59.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "a0eb5f1a6eb5e0e4343aae178ea28e73f70a0c48c14bc0d77683b45918a67eaa", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.59.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "20aede179b3d28cb5a5b6d404adf0d9cceddfb353436aead5170d29a1091aade", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.59.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "7d5a067d909d29f199f8b1f0c86e4a878d460545c4da608b1b278597485337ae", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.60.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.60.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "bd5df30d4b4412a9fa74714164ab86116e3eb2945a5b965eb7fc58176dc0e15f", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.60.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "4061d4e28d932052dd9ce7f6d5ad2a04d3a9bd76419bf5b3998e659fe157ce7b", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.60.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "28ccecbf8a2a15d8350716f9a982d1a971f1168412aa1239bb8f8168337661a1", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.60.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.60.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "0aa71cbab91a29d2a9a2fedb2531b28510376861c86f0aede753540d91bd88a1", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.60.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "6ee2cb51a61e1b56a3e6c5a9a115b758ba1ef5ea036a0906df6ea557cd25ae44", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.60.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "2335df8e612de8e3156418a32658ed562effe909cbad4f3ad757c6de47d2cf0d", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.60.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "2e515ce435ae1928c30fa744c88664c8f8b853f5ef32014c2e5d36c461f13334", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.61.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.61.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "7c456304ef25b4e055e33144f7386cd53423a32800dcd3460c392f792bf1f029", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.61.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "2b79011af9f4c436cb9a9a8d0ddcca33fc884553cc0e74cef12c7ed94fbe630a", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.61.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "2a2fc0c3f5b1fbe9a9252006d9997fa9d402bda8f1f9b21c784e25bd2450574b", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.61.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.61.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "d9a362f32b85965e1f52ac9fb8fed2212fe77b80753fead1cb05344a6f04e91d", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.61.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "e7f013a02244dd8e14a8a197ed94ac761c24f64902dbc1b20714a74f49b9c4a1", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.61.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "0febf9b0b94b8abb5c661842d089899d876297d281c412d6aac8e91c6e5b45ba", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.61.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "6a821b887271cc94b146b86fffe37a0c34ea392738ff6b6f8f8bac01c26cc10d", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.62.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.62.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "7bda9eeae0e7f9f5239e0896f58a76ea6736f06f6ee4352b348ad03ee5d7f413", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.62.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "a5277245574a1d8c1f8dd6fe4f1ea4b0b94b980034ec2767f90f94d7f94eb60c", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.62.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "6adcf1ad5175b8a0790f5cc3127e3f9a39db1b0453759a7c5ffbf5b5cd7e6349", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.62.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.62.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "c83a7ef597ccc8be2e258a87736e6a2eb588fda7c74e4f76d4b6dc81392a23a8", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.62.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "84d8df3c7a119fc167a7b56c74dd5f952afd1947c573da8e41be835e62e0b969", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.62.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "52b8e3825c43ef6b80492475e7c1b894444016b04799091bceb3541a9b748d5b", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.62.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "05914337102004c273fe856be1614a89c178cf976817804e032c4024523ce5a8", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.63.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.63.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d44d282bc3dfe11a77911bd5f2285c8927975f0e7789d3616f3953539d13c889", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.63.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "f024aa3241765b44e9804744f14b37c7272709da5439f11599df7db10671b08e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.63.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "ecb9abbf61e34af81c9253c7b4337dff35a9592160c13b981b55d88629fdd563", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.63.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.63.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "e638d895943665d076796e5d16baeaeaac56f8e5c58082621a471b9ca8fa1c67", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.63.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "6eab231e55487bb94ac681e89b39ce1bb23ccd9d73a9c0246b788ac1d2b1692b", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.63.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "459945d3abc0cdc11654f218aade82b4bd70d49cb9c2661fb5dfb01a61dc02ef", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.63.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "c3737ef618ad0cfd7539c89843dce71097231fbd7487c751d1701a7938bba46d", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.64.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.64.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "816fd3dcd3c1edf5846c8b2fa48716578ef324ceb15a0bb2084cb494ed8ae314", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.64.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "303e974b2fa6c626edb191ef8b4d7c4d2dbca8e1836b3b3cd52ebb5c22d6f1e7", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.64.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "33db3afab869a0b59c44e02308903815b777030b0d1e533466573ea847a640b7", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.64.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.64.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "b8db65bb848982d7c19136865d1a5cfc0e21cafdb42731509c58ddbba2528256", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.64.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "0407a84040c392fb766bdbb2e82e930b5bf841f1e274e7d880e94b707aacaea3", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.64.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "69c0feb737fc29d0223f1cd3069b9c6477d37d67b21ad29b083419326eea5c5d", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.64.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "bbc98a5b67cfa7295291260b128c5744cc18b02d165edd9de48991434b9b71bc", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.65.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.65.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "eff6a8463af99c5b61ce8e4cdac8e3e168ef9200b31b3b6d2db44ec69b5089d6", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.65.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "0fba19f74a37f48decd7b44aeaec2ec6f3fffabdff8fb7e1f5b0690dae5a10f1", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.65.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "6175730703685c31522f1eb5553376f813d875c71865e39b4ccf9bccacf387e8", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.65.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.65.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "34043b778248f4c90f021b6ab02bc7d2398a43c34bd88c1ee1742560049b437a", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.65.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "8c82d8f79bdf48e7ecc9088d3a172f9af9a40c1db72e60ac55bff5fe0d2ef509", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.65.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "8e5312133ba00ec4fdbabe9c594d8d2d3a25f3ef628aa021a1bb1b8419953975", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.65.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "40a0770e1e4b4342dfc800c8dcc90697a4649c4b7e0dfd9254b13e9c57c6be9e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.66.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.66.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "a89b5d66b91a8d34a252f622ca13216b2efb9d1c75dba192371cdd6aa8d5d3ca", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.66.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d866aeab82eb15fd028bb2eefb184f12349959e39eda0f0ebca5f8bc239ddf00", |
| "shard": "model-00005-of-00006.safetensors" |
| }, |
| "model.layers.66.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "8ea10a4b99370623ffae102c839b6b6cff15be7290f0374bd8205eb558ac6e29", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.66.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.66.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "300135ba9367b190ce96257851c0427cc89580baf09249b95a20327b2e71f737", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.66.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "87c67d3d30fbb55693d33c542984df703d853020e1c3acfb060bbfeaabfeb46a", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.66.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "da975bb5b08c7627c530233447099367c2870dacf03c7b3d7cf992852db90710", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.66.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "7929836634037ae772e792f1f9f4cde2bcabccadf21bbe8b2d0623faecf844a9", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.67.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.67.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "3125f70c12729f2926dca7a8e3fa0439c9cedb1a8da2f7a7c4d587b978d573e4", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.67.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "4c78ab2790494c0b569c3b3f41cc2b00b5eab981c8202d4df99c1f0e13091509", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.67.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "0282169a7588dcbb36ee11c6574e1044b2717364b116d0d8823541ffcab56031", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.67.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.67.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "10277fbc82a689ca647ee42ce1a06d0f3c2d33d470c738b6264fca0f5438cd62", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.67.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "19302949bbf4a7f51059735aefb0d5349ea768aebba6a143ea1f5fecf13c3535", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.67.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "29ecebf5633993939bc05d6abf796d29e51109efeb78ef735536bc91e59ba2d3", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.67.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "ee5dbe493e93dc4d7ed310ee3aad35a2a19557583ed8c25f43f0d587c8dc82b2", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.68.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.68.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "993dc6a592b5833d669e783fd6c4011ae534ec10a282094ac995fca5fff9cfbf", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.68.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "f3de3c178e55b666c2fb15eaf9032711d3f4decf647a5a829ec9443ae69a8b60", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.68.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "7307f9672579cb884b4fe803f8c656620b3f213c13c46b28f3f2d781ab18ee43", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.68.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.68.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "3612819999f79014e84084a6db46321375ceeda3e2f4d07a1c973216649bcd42", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.68.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "eb1271405a26f7c8bd51a4e7d1e03c505ea8083aeedc923d94bc8d4cdcd07440", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.68.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "94af6632fcd81b503d8e7eaf839e0b4bd4cb78471da3ff39c425631a6562d12c", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.68.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "9364c504676733a86b3944a2d03fa397d42bb86125124d2aa2a896105c4f7be7", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.69.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.69.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "f1a468d8aa6e6689fef9b607e28f185714d1cffd4c6c69e4f2653a70b357887e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.69.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "1c1b74db31ef51a1631b09b2e6c746ea115f595472cf185ee9cfc81df9ee0a52", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.69.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "0348626eee37602c1a3bb0c0d1415655c5ccdb67667f280089b44ba35ceda09b", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.69.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.69.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "565b868131ad847647b768b899694e4254b5da3b73f554e67e32358dffe0ffdc", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.69.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "ba2791cc269b658a7524f22c99c2040270022ea435a0923a8315a69f939049cb", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.69.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "398f4383aed2526d0530c47a55dc61c14ea450990890951e88a225d8b94cd161", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.69.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "c62b4c3cc917b91d89354c446449c14f07e05adab7d1c31908fe634dc17a36d8", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.70.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.70.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "deb94534c935689bebf5880cc82b36ce7c64f4c50aeb17ed47b08aa505ef72f1", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.70.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "251f6de2b283036f263c9e0a223c3c1a2623e5cb7c8425bd31ab1944f880342e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.70.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "4a1bbb45d7a2eae8a11bde72155b35f3deb95be1ba910299e363ffd6ea37ab33", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.70.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.70.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "e375523b607385bea317727899e729de3da7f9801e42901593b089276aafd786", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.70.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "fb00b2a4186b5d0b165dd6cbdddd5a62f2b10488438ffd3ba1d9d8d9e8a0a5ee", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.70.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "451a69630fdfd0016427fda7c12608cf2fc85eb9aba207caf3a42be7f180351c", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.70.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "1b73e7487143c883560c09e95d44f96be46ea47adfdf9a28d7a5e8fa8a0eba84", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.71.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.71.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d9402b6372f6371861f4235be1ffc312fce822e5658bac342a07bf97e230e145", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.71.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "e5c14d15ae3b4ce51b511eb79a34c425adbfb3acb0d4c35ae3f12eb9e7d362a4", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.71.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "e8078f36263451ccf1f773c5a65aeaacdf132f505a6b816a16b998efaa7b21ea", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.71.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.71.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "8a992e8c69aae09ba9d6631b577fd5dabd5cd8d4ff85da0b67be75c86df32afa", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.71.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "34093a1dadcef5aec91f2a8e84749b75a841d0967d809f667d5b86ce715cd625", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.71.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "3c53e62704ab8e119068b3167b606e55cc40f05f7f7147eb77794d99df39717a", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.71.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "9af41c402b9edc6455fdf5a35f824204624614bfcdb117694f32e2ac27c96591", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.72.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.72.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "54d545c4deb88a05241611ba69c1a66b4418f221e25f222bdc4a66233f7ce5c1", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.72.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "011356df294a22c66feba22d3d829d5776724f94ce1ff805c392ea12707de32e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.72.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "f2fc23c06bda25bfbac48360ae8f3497ed01c0d8257b5ce8feec57eb1d18edcc", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.72.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.72.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "6e1745210eedc63fb7ad5b181d2b2f66c2df201e53e9eb8b4b0bb842df3b79c4", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.72.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "bc52e2901c81c891b6bc1d52ea70286a33bf72178e34f4edcfedda04dda6791d", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.72.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "73dd50a296ca2c69cbf12760cc3d1d09cd2a636bf3fe89fbfeb26624d7a93235", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.72.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "de3a511c75635d8566ce3e7d1212c8342f149362e5b5a9b6374823bf84e5de16", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.73.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.73.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "1ed2ef5b9ba62c8fe62b3249df7a898675f9634862981dee3705ca946f88ea1d", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.73.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "5a1a36873d947a163d5f600cec544c154343ce36097cfd126e9aa1186f84c41c", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.73.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "79d8c3933785baf44018f5d52ea70554455ed0473d07062559bcaca51cbc6cc9", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.73.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.73.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "7420f6b7468d4c0b63434c21a08c6b270dfeae26e1cca4cc30333c9855f631c4", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.73.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "d66997d212fc546ad97d132571e913aa6a0459ac0f8341f08793514fd7a27fdd", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.73.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "d00045c84f3eb52dde04127dad9de44c41a5f97c1e709f7b9d341ec3ebbeb5ff", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.73.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "be4b4ef3e624a6eeca342f40c29b57fccf33f30fb67f43bdb6a6a7d2982b5463", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.74.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.74.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "9734b3b0ddf89645e6c468fe4c2be44591aad58d9baf6e11562bf4a3d28bc8ce", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.74.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "b5025435975b82aff695fa6075011ec9fb5257541ad0158eaa656217964e44ed", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.74.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "d1feac27bca1fe3590308399b11c8b88923cc5bf2e23d9728d4422f4bd0852ce", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.74.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.74.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "3db8422bbabd625fbac53f53dbd203a27d6b00ffe0982e442fedba3fa9b9ca94", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.74.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "a112a042f942ad6b518665b5d6ed04f7738c4568841c45a09df0b59ad42a6389", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.74.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "4e10a3e8f209882f3bca23c05d4b6e867d876b5fccf1e44a8a472943bffa1305", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.74.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "5e861fe8d68dfb96a0fe2fcb01b771253ccae93535591aaff270c09bb428288d", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.75.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.75.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "7b306cf4a4182129a1888a3c0ee711ce7518b90cc5429741d0d16a7f84adad51", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.75.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "009a60e3ccf3a5efc07c70ab130bef2dae4eef13c23a8c5e0156d06cd4bd51d7", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.75.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "78ce38248cbeb9fed2eab7d707e7da42e8867731b2dc2b2eb8cdbe395103a0ad", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.75.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.75.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "fb2d3548583069fa875f0c625e2afe4b3ff39f7e7fb450a81cad61336a0b8bb1", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.75.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "b7c1f829e60d258e9e58cc2311c41ff40d8e3352afdfe80cd0d17754467d69a6", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.75.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "5657e97af5929d8eaa87118b708497f915211772dd4daa47a3fca64c859fea56", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.75.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "0e015eaefdc8027f3c29888a6917413d8713e1fd0971989722fa7a8195575781", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.76.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.76.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "96cb8f92a22786d891432fa42393fd2f8fac1e8ef38aec315cb6886201e954e8", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.76.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "1b8d3b398cccce29a966867e7b7076be0cf6f1edc463649e54379417dd426525", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.76.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "68322c72346cec91b22f5ca971aedbb94300a9b8a791d26d4a371a4733aa4fad", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.76.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.76.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "4eddd624fa4364d056a9592efb86ef6d53376daf6cc719bc22ca99d4583582c8", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.76.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "530fd68a18c7f88031f4b75c4a6247b6cc610614fd0964914b5bb12bc774c278", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.76.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "ee45924190ec1baa03fb4f20ffb6ec2f20e87283462d7dac4db612185a942366", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.76.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "c1a699e70e69a6f468d6121c7b19a5990b2bd1ced19633069e436e98f22bc293", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.77.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.77.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "2bc181654f3ffe23b089b30505b4ecab09af6476247dcbe54a96df13e6284779", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.77.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "c4dd7167d5eb2f8176c563b4ce9b9e16107bfe60a65c7c562957abac6e2b0305", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.77.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "6988c20cbc8ee0e798775b00901400d94970cf2b34cb673b9f60ae1611250899", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.77.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.77.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "084a70a31dc398e04f1e7784d095ad422fe1012df1b47154012a06949ebff804", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.77.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "68ef7a69e6394c36858f38b6493dcdfd8e3f1a131caefbee8fefde6cd9641604", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.77.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "dadc1cef3c009ed8f4cf4fe426dbc169800e1b8b02c8f4db66f89388644339ec", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.77.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "cafaa81fec4ae994aee96d68bab0610cc5c3ac6d51cc609379da6e480d30928e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.78.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.78.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "53e6603d17d1b4cc4505b5e5a46e8eb19cc25b7613bff665844c1fede88ee0d4", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.78.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "8394c7dcb86c03e9ed0ffcd365ab86baf42a4d83000e0f2125e8a5a4c88fa5fe", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.78.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "cde119c4ca8fabe02c1d84069eb151708efb4177ac93d24cd90c0cd489414f1f", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.78.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.78.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "dc1c4bfa15f2ffdabbd70d6f23c996d72e462cdd5bceea2fb2ffec44098ea47e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.78.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "12e0efcdde03fe7c09e3c39a21a0f56d53e86096fc1c0e45115843c1832d073c", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.78.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "b1f4a1838644f841545f13a9e8384197b762f2f9c7e1e868d14fb6b8d0a7cd2d", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.78.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "1225d14da5e7a315ace8c0e5c0c1f6c3a6e25fd07d0843ee78fe81cb4f74273e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.79.input_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.79.mlp.down_proj.weight": { |
| "shape": [ |
| 4096, |
| 14336 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "ed9eca61e4d7d17f7c68741a5569fdfda8c68ecea4f15e23b5a4994fa539a434", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.79.mlp.gate_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "ab59fd885e944cf6d884eaafe0e69d21bb26dc46320963aed5b8c13c1c3f9167", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.79.mlp.up_proj.weight": { |
| "shape": [ |
| 14336, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 58720256, |
| "nbytes": 117440512, |
| "sha256": "0bb001e4abd4cb3fa81d882ba6018723a77d0e17d5c4530f89eef52f96e3e518", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.79.post_attention_layernorm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "8676bed4ca36c6e3328f5eaebc71b2f1a6c760c56f7139942239a033fe776c5e", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.79.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "fb981a6e1c74803f3f5761d8e48e14c0b08bcd53540f3bbb3fe92c754bf04023", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.79.self_attn.o_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "8ab0719c3ee02db39bcbdb819ade79ac104fa7bd0a22482e2f80b3c2cbaaca41", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.79.self_attn.q_proj.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 16777216, |
| "nbytes": 33554432, |
| "sha256": "da9e4a34212d425d8554b7324f1637b8ad5c7762dd664ce585c79d7d6cb51b99", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.layers.79.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4194304, |
| "nbytes": 8388608, |
| "sha256": "6e30e9c2cf19f3d22b8c6e52fe00205093bc34321b82a03482cd26c3e3da77a9", |
| "shard": "model-00006-of-00006.safetensors" |
| }, |
| "model.norm.weight": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "BF16", |
| "elements": 4096, |
| "nbytes": 8192, |
| "sha256": "fcea373047bc9cdb432f757971d4432d07dbe20c2adc49083f59d544b1d7c90c", |
| "shard": "model-00006-of-00006.safetensors" |
| } |
| } |
| } |
|
|
