{"model.embed_tokens.weight": {"dtype": "BF16", "shape": [128000, 2048], "data_offsets": [0, 524288000]}, "model.embedding_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [524288000, 524292096]}, "model.layers.0.conv.conv.weight": {"dtype": "BF16", "shape": [2048, 1, 7], "data_offsets": [524292096, 524320768]}, "model.layers.0.conv.in_proj.weight": {"dtype": "BF16", "shape": [6144, 2048], "data_offsets": [524320768, 549486592]}, "model.layers.0.conv.out_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [549486592, 557875200]}, "model.layers.0.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [557875200, 589332480]}, "model.layers.0.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [589332480, 620789760]}, "model.layers.0.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [620789760, 652247040]}, "model.layers.0.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [652247040, 652251136]}, "model.layers.0.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [652251136, 652255232]}, "model.layers.0.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [652255232, 652259328]}, "model.layers.0.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [652259328, 652263424]}, "model.layers.1.conv.conv.weight": {"dtype": "BF16", "shape": [2048, 1, 7], "data_offsets": [652263424, 652292096]}, "model.layers.1.conv.in_proj.weight": {"dtype": "BF16", "shape": [6144, 2048], "data_offsets": [652292096, 677457920]}, "model.layers.1.conv.out_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [677457920, 685846528]}, "model.layers.1.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [685846528, 717303808]}, "model.layers.1.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [717303808, 748761088]}, "model.layers.1.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [748761088, 780218368]}, "model.layers.1.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [780218368, 780222464]}, "model.layers.1.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [780222464, 780226560]}, "model.layers.1.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [780226560, 780230656]}, "model.layers.1.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [780230656, 780234752]}, "model.layers.10.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [780234752, 811692032]}, "model.layers.10.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [811692032, 843149312]}, "model.layers.10.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [843149312, 874606592]}, "model.layers.10.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [874606592, 874610688]}, "model.layers.10.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [874610688, 874614784]}, "model.layers.10.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [874614784, 874618880]}, "model.layers.10.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [874618880, 874622976]}, "model.layers.10.self_attn.gate_proj.weight": {"dtype": "BF16", "shape": [16, 2048], "data_offsets": [874622976, 874688512]}, "model.layers.10.self_attn.k_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [874688512, 874688768]}, "model.layers.10.self_attn.k_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [874688768, 876785920]}, "model.layers.10.self_attn.o_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [876785920, 885174528]}, "model.layers.10.self_attn.q_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [885174528, 885174784]}, "model.layers.10.self_attn.q_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [885174784, 893563392]}, "model.layers.10.self_attn.v_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [893563392, 895660544]}, "model.layers.11.conv.conv.weight": {"dtype": "BF16", "shape": [2048, 1, 7], "data_offsets": [895660544, 895689216]}, "model.layers.11.conv.in_proj.weight": {"dtype": "BF16", "shape": [6144, 2048], "data_offsets": [895689216, 920855040]}, "model.layers.11.conv.out_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [920855040, 929243648]}, "model.layers.11.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [929243648, 960700928]}, "model.layers.11.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [960700928, 992158208]}, "model.layers.11.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [992158208, 1023615488]}, "model.layers.11.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1023615488, 1023619584]}, "model.layers.11.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1023619584, 1023623680]}, "model.layers.11.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1023623680, 1023627776]}, "model.layers.11.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1023627776, 1023631872]}, "model.layers.12.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1023631872, 1055089152]}, "model.layers.12.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [1055089152, 1086546432]}, "model.layers.12.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1086546432, 1118003712]}, "model.layers.12.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1118003712, 1118007808]}, "model.layers.12.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1118007808, 1118011904]}, "model.layers.12.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1118011904, 1118016000]}, "model.layers.12.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1118016000, 1118020096]}, "model.layers.12.self_attn.gate_proj.weight": {"dtype": "BF16", "shape": [16, 2048], "data_offsets": [1118020096, 1118085632]}, "model.layers.12.self_attn.k_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [1118085632, 1118085888]}, "model.layers.12.self_attn.k_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [1118085888, 1120183040]}, "model.layers.12.self_attn.o_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [1120183040, 1128571648]}, "model.layers.12.self_attn.q_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [1128571648, 1128571904]}, "model.layers.12.self_attn.q_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [1128571904, 1136960512]}, "model.layers.12.self_attn.v_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [1136960512, 1139057664]}, "model.layers.13.conv.conv.weight": {"dtype": "BF16", "shape": [2048, 1, 7], "data_offsets": [1139057664, 1139086336]}, "model.layers.13.conv.in_proj.weight": {"dtype": "BF16", "shape": [6144, 2048], "data_offsets": [1139086336, 1164252160]}, "model.layers.13.conv.out_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [1164252160, 1172640768]}, "model.layers.13.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1172640768, 1204098048]}, "model.layers.13.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [1204098048, 1235555328]}, "model.layers.13.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1235555328, 1267012608]}, "model.layers.13.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1267012608, 1267016704]}, "model.layers.13.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1267016704, 1267020800]}, "model.layers.13.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1267020800, 1267024896]}, "model.layers.13.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1267024896, 1267028992]}, "model.layers.14.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1267028992, 1298486272]}, "model.layers.14.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [1298486272, 1329943552]}, "model.layers.14.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1329943552, 1361400832]}, "model.layers.14.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1361400832, 1361404928]}, "model.layers.14.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1361404928, 1361409024]}, "model.layers.14.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1361409024, 1361413120]}, "model.layers.14.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1361413120, 1361417216]}, "model.layers.14.self_attn.gate_proj.weight": {"dtype": "BF16", "shape": [16, 2048], "data_offsets": [1361417216, 1361482752]}, "model.layers.14.self_attn.k_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [1361482752, 1361483008]}, "model.layers.14.self_attn.k_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [1361483008, 1363580160]}, "model.layers.14.self_attn.o_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [1363580160, 1371968768]}, "model.layers.14.self_attn.q_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [1371968768, 1371969024]}, "model.layers.14.self_attn.q_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [1371969024, 1380357632]}, "model.layers.14.self_attn.v_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [1380357632, 1382454784]}, "model.layers.15.conv.conv.weight": {"dtype": "BF16", "shape": [2048, 1, 7], "data_offsets": [1382454784, 1382483456]}, "model.layers.15.conv.in_proj.weight": {"dtype": "BF16", "shape": [6144, 2048], "data_offsets": [1382483456, 1407649280]}, "model.layers.15.conv.out_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [1407649280, 1416037888]}, "model.layers.15.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1416037888, 1447495168]}, "model.layers.15.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [1447495168, 1478952448]}, "model.layers.15.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1478952448, 1510409728]}, "model.layers.15.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1510409728, 1510413824]}, "model.layers.15.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1510413824, 1510417920]}, "model.layers.15.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1510417920, 1510422016]}, "model.layers.15.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1510422016, 1510426112]}, "model.layers.16.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1510426112, 1541883392]}, "model.layers.16.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [1541883392, 1573340672]}, "model.layers.16.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1573340672, 1604797952]}, "model.layers.16.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1604797952, 1604802048]}, "model.layers.16.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1604802048, 1604806144]}, "model.layers.16.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1604806144, 1604810240]}, "model.layers.16.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1604810240, 1604814336]}, "model.layers.16.self_attn.gate_proj.weight": {"dtype": "BF16", "shape": [16, 2048], "data_offsets": [1604814336, 1604879872]}, "model.layers.16.self_attn.k_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [1604879872, 1604880128]}, "model.layers.16.self_attn.k_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [1604880128, 1606977280]}, "model.layers.16.self_attn.o_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [1606977280, 1615365888]}, "model.layers.16.self_attn.q_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [1615365888, 1615366144]}, "model.layers.16.self_attn.q_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [1615366144, 1623754752]}, "model.layers.16.self_attn.v_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [1623754752, 1625851904]}, "model.layers.17.conv.conv.weight": {"dtype": "BF16", "shape": [2048, 1, 7], "data_offsets": [1625851904, 1625880576]}, "model.layers.17.conv.in_proj.weight": {"dtype": "BF16", "shape": [6144, 2048], "data_offsets": [1625880576, 1651046400]}, "model.layers.17.conv.out_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [1651046400, 1659435008]}, "model.layers.17.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1659435008, 1690892288]}, "model.layers.17.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [1690892288, 1722349568]}, "model.layers.17.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1722349568, 1753806848]}, "model.layers.17.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1753806848, 1753810944]}, "model.layers.17.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1753810944, 1753815040]}, "model.layers.17.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1753815040, 1753819136]}, "model.layers.17.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1753819136, 1753823232]}, "model.layers.18.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1753823232, 1785280512]}, "model.layers.18.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [1785280512, 1816737792]}, "model.layers.18.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1816737792, 1848195072]}, "model.layers.18.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1848195072, 1848199168]}, "model.layers.18.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1848199168, 1848203264]}, "model.layers.18.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1848203264, 1848207360]}, "model.layers.18.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1848207360, 1848211456]}, "model.layers.18.self_attn.gate_proj.weight": {"dtype": "BF16", "shape": [16, 2048], "data_offsets": [1848211456, 1848276992]}, "model.layers.18.self_attn.k_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [1848276992, 1848277248]}, "model.layers.18.self_attn.k_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [1848277248, 1850374400]}, "model.layers.18.self_attn.o_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [1850374400, 1858763008]}, "model.layers.18.self_attn.q_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [1858763008, 1858763264]}, "model.layers.18.self_attn.q_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [1858763264, 1867151872]}, "model.layers.18.self_attn.v_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [1867151872, 1869249024]}, "model.layers.19.conv.conv.weight": {"dtype": "BF16", "shape": [2048, 1, 7], "data_offsets": [1869249024, 1869277696]}, "model.layers.19.conv.in_proj.weight": {"dtype": "BF16", "shape": [6144, 2048], "data_offsets": [1869277696, 1894443520]}, "model.layers.19.conv.out_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [1894443520, 1902832128]}, "model.layers.19.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1902832128, 1934289408]}, "model.layers.19.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [1934289408, 1965746688]}, "model.layers.19.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1965746688, 1997203968]}, "model.layers.19.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1997203968, 1997208064]}, "model.layers.19.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1997208064, 1997212160]}, "model.layers.19.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1997212160, 1997216256]}, "model.layers.19.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [1997216256, 1997220352]}, "model.layers.2.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [1997220352, 2028677632]}, "model.layers.2.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [2028677632, 2060134912]}, "model.layers.2.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2060134912, 2091592192]}, "model.layers.2.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2091592192, 2091596288]}, "model.layers.2.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2091596288, 2091600384]}, "model.layers.2.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2091600384, 2091604480]}, "model.layers.2.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2091604480, 2091608576]}, "model.layers.2.self_attn.gate_proj.weight": {"dtype": "BF16", "shape": [16, 2048], "data_offsets": [2091608576, 2091674112]}, "model.layers.2.self_attn.k_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [2091674112, 2091674368]}, "model.layers.2.self_attn.k_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [2091674368, 2093771520]}, "model.layers.2.self_attn.o_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [2093771520, 2102160128]}, "model.layers.2.self_attn.q_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [2102160128, 2102160384]}, "model.layers.2.self_attn.q_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [2102160384, 2110548992]}, "model.layers.2.self_attn.v_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [2110548992, 2112646144]}, "model.layers.20.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2112646144, 2144103424]}, "model.layers.20.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [2144103424, 2175560704]}, "model.layers.20.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2175560704, 2207017984]}, "model.layers.20.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2207017984, 2207022080]}, "model.layers.20.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2207022080, 2207026176]}, "model.layers.20.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2207026176, 2207030272]}, "model.layers.20.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2207030272, 2207034368]}, "model.layers.20.self_attn.gate_proj.weight": {"dtype": "BF16", "shape": [16, 2048], "data_offsets": [2207034368, 2207099904]}, "model.layers.20.self_attn.k_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [2207099904, 2207100160]}, "model.layers.20.self_attn.k_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [2207100160, 2209197312]}, "model.layers.20.self_attn.o_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [2209197312, 2217585920]}, "model.layers.20.self_attn.q_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [2217585920, 2217586176]}, "model.layers.20.self_attn.q_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [2217586176, 2225974784]}, "model.layers.20.self_attn.v_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [2225974784, 2228071936]}, "model.layers.21.conv.conv.weight": {"dtype": "BF16", "shape": [2048, 1, 7], "data_offsets": [2228071936, 2228100608]}, "model.layers.21.conv.in_proj.weight": {"dtype": "BF16", "shape": [6144, 2048], "data_offsets": [2228100608, 2253266432]}, "model.layers.21.conv.out_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [2253266432, 2261655040]}, "model.layers.21.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2261655040, 2293112320]}, "model.layers.21.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [2293112320, 2324569600]}, "model.layers.21.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2324569600, 2356026880]}, "model.layers.21.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2356026880, 2356030976]}, "model.layers.21.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2356030976, 2356035072]}, "model.layers.21.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2356035072, 2356039168]}, "model.layers.21.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2356039168, 2356043264]}, "model.layers.22.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2356043264, 2387500544]}, "model.layers.22.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [2387500544, 2418957824]}, "model.layers.22.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2418957824, 2450415104]}, "model.layers.22.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2450415104, 2450419200]}, "model.layers.22.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2450419200, 2450423296]}, "model.layers.22.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2450423296, 2450427392]}, "model.layers.22.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2450427392, 2450431488]}, "model.layers.22.self_attn.gate_proj.weight": {"dtype": "BF16", "shape": [16, 2048], "data_offsets": [2450431488, 2450497024]}, "model.layers.22.self_attn.k_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [2450497024, 2450497280]}, "model.layers.22.self_attn.k_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [2450497280, 2452594432]}, "model.layers.22.self_attn.o_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [2452594432, 2460983040]}, "model.layers.22.self_attn.q_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [2460983040, 2460983296]}, "model.layers.22.self_attn.q_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [2460983296, 2469371904]}, "model.layers.22.self_attn.v_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [2469371904, 2471469056]}, "model.layers.23.conv.conv.weight": {"dtype": "BF16", "shape": [2048, 1, 7], "data_offsets": [2471469056, 2471497728]}, "model.layers.23.conv.in_proj.weight": {"dtype": "BF16", "shape": [6144, 2048], "data_offsets": [2471497728, 2496663552]}, "model.layers.23.conv.out_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [2496663552, 2505052160]}, "model.layers.23.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2505052160, 2536509440]}, "model.layers.23.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [2536509440, 2567966720]}, "model.layers.23.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2567966720, 2599424000]}, "model.layers.23.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2599424000, 2599428096]}, "model.layers.23.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2599428096, 2599432192]}, "model.layers.23.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2599432192, 2599436288]}, "model.layers.23.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2599436288, 2599440384]}, "model.layers.3.conv.conv.weight": {"dtype": "BF16", "shape": [2048, 1, 7], "data_offsets": [2599440384, 2599469056]}, "model.layers.3.conv.in_proj.weight": {"dtype": "BF16", "shape": [6144, 2048], "data_offsets": [2599469056, 2624634880]}, "model.layers.3.conv.out_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [2624634880, 2633023488]}, "model.layers.3.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2633023488, 2664480768]}, "model.layers.3.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [2664480768, 2695938048]}, "model.layers.3.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2695938048, 2727395328]}, "model.layers.3.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2727395328, 2727399424]}, "model.layers.3.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2727399424, 2727403520]}, "model.layers.3.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2727403520, 2727407616]}, "model.layers.3.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2727407616, 2727411712]}, "model.layers.4.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2727411712, 2758868992]}, "model.layers.4.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [2758868992, 2790326272]}, "model.layers.4.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2790326272, 2821783552]}, "model.layers.4.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2821783552, 2821787648]}, "model.layers.4.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2821787648, 2821791744]}, "model.layers.4.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2821791744, 2821795840]}, "model.layers.4.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2821795840, 2821799936]}, "model.layers.4.self_attn.gate_proj.weight": {"dtype": "BF16", "shape": [16, 2048], "data_offsets": [2821799936, 2821865472]}, "model.layers.4.self_attn.k_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [2821865472, 2821865728]}, "model.layers.4.self_attn.k_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [2821865728, 2823962880]}, "model.layers.4.self_attn.o_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [2823962880, 2832351488]}, "model.layers.4.self_attn.q_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [2832351488, 2832351744]}, "model.layers.4.self_attn.q_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [2832351744, 2840740352]}, "model.layers.4.self_attn.v_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [2840740352, 2842837504]}, "model.layers.5.conv.conv.weight": {"dtype": "BF16", "shape": [2048, 1, 7], "data_offsets": [2842837504, 2842866176]}, "model.layers.5.conv.in_proj.weight": {"dtype": "BF16", "shape": [6144, 2048], "data_offsets": [2842866176, 2868032000]}, "model.layers.5.conv.out_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [2868032000, 2876420608]}, "model.layers.5.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2876420608, 2907877888]}, "model.layers.5.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [2907877888, 2939335168]}, "model.layers.5.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2939335168, 2970792448]}, "model.layers.5.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2970792448, 2970796544]}, "model.layers.5.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2970796544, 2970800640]}, "model.layers.5.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2970800640, 2970804736]}, "model.layers.5.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [2970804736, 2970808832]}, "model.layers.6.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [2970808832, 3002266112]}, "model.layers.6.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [3002266112, 3033723392]}, "model.layers.6.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [3033723392, 3065180672]}, "model.layers.6.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3065180672, 3065184768]}, "model.layers.6.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3065184768, 3065188864]}, "model.layers.6.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3065188864, 3065192960]}, "model.layers.6.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3065192960, 3065197056]}, "model.layers.6.self_attn.gate_proj.weight": {"dtype": "BF16", "shape": [16, 2048], "data_offsets": [3065197056, 3065262592]}, "model.layers.6.self_attn.k_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [3065262592, 3065262848]}, "model.layers.6.self_attn.k_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [3065262848, 3067360000]}, "model.layers.6.self_attn.o_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [3067360000, 3075748608]}, "model.layers.6.self_attn.q_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [3075748608, 3075748864]}, "model.layers.6.self_attn.q_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [3075748864, 3084137472]}, "model.layers.6.self_attn.v_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [3084137472, 3086234624]}, "model.layers.7.conv.conv.weight": {"dtype": "BF16", "shape": [2048, 1, 7], "data_offsets": [3086234624, 3086263296]}, "model.layers.7.conv.in_proj.weight": {"dtype": "BF16", "shape": [6144, 2048], "data_offsets": [3086263296, 3111429120]}, "model.layers.7.conv.out_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [3111429120, 3119817728]}, "model.layers.7.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [3119817728, 3151275008]}, "model.layers.7.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [3151275008, 3182732288]}, "model.layers.7.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [3182732288, 3214189568]}, "model.layers.7.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3214189568, 3214193664]}, "model.layers.7.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3214193664, 3214197760]}, "model.layers.7.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3214197760, 3214201856]}, "model.layers.7.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3214201856, 3214205952]}, "model.layers.8.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [3214205952, 3245663232]}, "model.layers.8.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [3245663232, 3277120512]}, "model.layers.8.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [3277120512, 3308577792]}, "model.layers.8.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3308577792, 3308581888]}, "model.layers.8.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3308581888, 3308585984]}, "model.layers.8.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3308585984, 3308590080]}, "model.layers.8.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3308590080, 3308594176]}, "model.layers.8.self_attn.gate_proj.weight": {"dtype": "BF16", "shape": [16, 2048], "data_offsets": [3308594176, 3308659712]}, "model.layers.8.self_attn.k_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [3308659712, 3308659968]}, "model.layers.8.self_attn.k_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [3308659968, 3310757120]}, "model.layers.8.self_attn.o_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [3310757120, 3319145728]}, "model.layers.8.self_attn.q_norm.weight": {"dtype": "BF16", "shape": [128], "data_offsets": [3319145728, 3319145984]}, "model.layers.8.self_attn.q_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [3319145984, 3327534592]}, "model.layers.8.self_attn.v_proj.weight": {"dtype": "BF16", "shape": [512, 2048], "data_offsets": [3327534592, 3329631744]}, "model.layers.9.conv.conv.weight": {"dtype": "BF16", "shape": [2048, 1, 7], "data_offsets": [3329631744, 3329660416]}, "model.layers.9.conv.in_proj.weight": {"dtype": "BF16", "shape": [6144, 2048], "data_offsets": [3329660416, 3354826240]}, "model.layers.9.conv.out_proj.weight": {"dtype": "BF16", "shape": [2048, 2048], "data_offsets": [3354826240, 3363214848]}, "model.layers.9.feed_forward.w1.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [3363214848, 3394672128]}, "model.layers.9.feed_forward.w2.weight": {"dtype": "BF16", "shape": [2048, 7680], "data_offsets": [3394672128, 3426129408]}, "model.layers.9.feed_forward.w3.weight": {"dtype": "BF16", "shape": [7680, 2048], "data_offsets": [3426129408, 3457586688]}, "model.layers.9.ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3457586688, 3457590784]}, "model.layers.9.operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3457590784, 3457594880]}, "model.layers.9.post_ffn_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3457594880, 3457598976]}, "model.layers.9.post_operator_norm.weight": {"dtype": "BF16", "shape": [2048], "data_offsets": [3457598976, 3457603072]}}