| row,layer_id,short_id,layer_type,param_type,shape,nparam,nnz,sparsity,tile_shape,n_tile,n_tile_total,tile_avg,tile_min,tile_med,tile_max,col_avg,col_min,col_med,col_max,row_avg,row_min,row_med,row_max | |
| 0,model.layers.0.self_attn.q_proj,tx.0.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.240234375,0.60009765625,0.87744140625,0.599853515625,0.0,0.6875,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 1,model.layers.0.self_attn.k_proj,tx.0.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.279296875,0.59912109375,0.87158203125,0.599853515625,0.0,0.6796875,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 2,model.layers.0.self_attn.v_proj,tx.0.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.31640625,0.59912109375,0.84521484375,0.599853515625,0.0,0.5546875,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 3,model.layers.0.self_attn.o_proj,tx.0.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.10302734375,0.611328125,1.0,0.599853515625,0.0078125,0.59375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 4,model.layers.0.mlp.gate_proj,tx.0.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51708984375,0.60009765625,0.6728515625,0.599853515625,0.078125,0.609375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 5,model.layers.0.mlp.up_proj,tx.0.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.53125,0.599609375,0.67724609375,0.599853515625,0.1015625,0.609375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 6,model.layers.0.mlp.down_proj,tx.0.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.4208984375,0.60107421875,0.7314453125,0.5999581813812256,0.046875,0.6171875,0.984375,0.5999581813812256,0.0,0.625,1.0 | |
| 7,model.layers.1.self_attn.q_proj,tx.1.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.3779296875,0.60205078125,0.79345703125,0.599853515625,0.0,0.6640625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 8,model.layers.1.self_attn.k_proj,tx.1.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.37060546875,0.60107421875,0.77734375,0.599853515625,0.0,0.6640625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 9,model.layers.1.self_attn.v_proj,tx.1.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.39208984375,0.60205078125,0.7333984375,0.599853515625,0.0234375,0.65625,1.0,0.599853515625,0.125,0.625,1.0 | |
| 10,model.layers.1.self_attn.o_proj,tx.1.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.28466796875,0.62353515625,0.98974609375,0.599853515625,0.0390625,0.5859375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 11,model.layers.1.mlp.gate_proj,tx.1.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.52783203125,0.60009765625,0.677734375,0.599853515625,0.1640625,0.609375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 12,model.layers.1.mlp.up_proj,tx.1.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.52734375,0.599609375,0.66845703125,0.599853515625,0.140625,0.609375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 13,model.layers.1.mlp.down_proj,tx.1.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.48828125,0.6015625,0.7109375,0.5999581813812256,0.0,0.6328125,0.9375,0.5999581813812256,0.0,0.625,1.0 | |
| 14,model.layers.2.self_attn.q_proj,tx.2.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.45166015625,0.6025390625,0.7451171875,0.599853515625,0.0,0.6484375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 15,model.layers.2.self_attn.k_proj,tx.2.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.427734375,0.60302734375,0.7607421875,0.599853515625,0.0078125,0.65625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 16,model.layers.2.self_attn.v_proj,tx.2.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4853515625,0.6025390625,0.7099609375,0.599853515625,0.0546875,0.640625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 17,model.layers.2.self_attn.o_proj,tx.2.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.2900390625,0.54296875,0.98779296875,0.599853515625,0.0078125,0.5625,1.0,0.599853515625,0.0,0.5625,1.0 | |
| 18,model.layers.2.mlp.gate_proj,tx.2.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.52880859375,0.60009765625,0.66455078125,0.599853515625,0.1328125,0.6015625,1.0,0.599853515625,0.0,0.625,1.0 | |
| 19,model.layers.2.mlp.up_proj,tx.2.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.53076171875,0.599609375,0.6650390625,0.599853515625,0.1796875,0.6015625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 20,model.layers.2.mlp.down_proj,tx.2.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.474609375,0.60107421875,0.73974609375,0.5999581813812256,0.015625,0.625,0.9140625,0.5999581813812256,0.0625,0.625,1.0 | |
| 21,model.layers.3.self_attn.q_proj,tx.3.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.48876953125,0.60107421875,0.705078125,0.599853515625,0.0234375,0.6328125,1.0,0.599853515625,0.0,0.625,1.0 | |
| 22,model.layers.3.self_attn.k_proj,tx.3.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.46923828125,0.60205078125,0.69921875,0.599853515625,0.015625,0.640625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 23,model.layers.3.self_attn.v_proj,tx.3.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.49658203125,0.60107421875,0.6923828125,0.599853515625,0.0625,0.625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 24,model.layers.3.self_attn.o_proj,tx.3.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.32666015625,0.58544921875,0.96142578125,0.599853515625,0.046875,0.5859375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 25,model.layers.3.mlp.gate_proj,tx.3.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.53466796875,0.60009765625,0.66748046875,0.599853515625,0.109375,0.609375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 26,model.layers.3.mlp.up_proj,tx.3.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.54150390625,0.599609375,0.6611328125,0.599853515625,0.1953125,0.6015625,1.0,0.599853515625,0.0,0.625,1.0 | |
| 27,model.layers.3.mlp.down_proj,tx.3.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.4814453125,0.6005859375,0.71240234375,0.5999581813812256,0.03125,0.6171875,0.9375,0.5999581813812256,0.0,0.625,1.0 | |
| 28,model.layers.4.self_attn.q_proj,tx.4.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.4619140625,0.60107421875,0.71728515625,0.599853515625,0.0078125,0.640625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 29,model.layers.4.self_attn.k_proj,tx.4.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.46142578125,0.60107421875,0.71923828125,0.599853515625,0.0078125,0.640625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 30,model.layers.4.self_attn.v_proj,tx.4.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.49609375,0.6015625,0.71923828125,0.599853515625,0.0546875,0.625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 31,model.layers.4.self_attn.o_proj,tx.4.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.27587890625,0.56298828125,0.9892578125,0.599853515625,0.0234375,0.5703125,1.0,0.599853515625,0.0,0.5625,1.0 | |
| 32,model.layers.4.mlp.gate_proj,tx.4.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.52001953125,0.6005859375,0.66064453125,0.599853515625,0.078125,0.609375,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 33,model.layers.4.mlp.up_proj,tx.4.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.53759765625,0.6005859375,0.66357421875,0.599853515625,0.1875,0.609375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 34,model.layers.4.mlp.down_proj,tx.4.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.48779296875,0.60107421875,0.70263671875,0.5999581813812256,0.0546875,0.609375,0.984375,0.5999581813812256,0.0,0.625,1.0 | |
| 35,model.layers.5.self_attn.q_proj,tx.5.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.44970703125,0.6005859375,0.7177734375,0.599853515625,0.0078125,0.6328125,0.96875,0.599853515625,0.0625,0.625,1.0 | |
| 36,model.layers.5.self_attn.k_proj,tx.5.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.455078125,0.6005859375,0.708984375,0.599853515625,0.015625,0.640625,0.9296875,0.599853515625,0.0625,0.625,1.0 | |
| 37,model.layers.5.self_attn.v_proj,tx.5.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.49755859375,0.6005859375,0.70751953125,0.599853515625,0.0625,0.625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 38,model.layers.5.self_attn.o_proj,tx.5.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.25390625,0.5859375,0.9423828125,0.599853515625,0.0078125,0.609375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 39,model.layers.5.mlp.gate_proj,tx.5.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51953125,0.60107421875,0.658203125,0.599853515625,0.046875,0.609375,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 40,model.layers.5.mlp.up_proj,tx.5.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.52392578125,0.6005859375,0.6630859375,0.599853515625,0.140625,0.6015625,1.0,0.599853515625,0.0,0.625,1.0 | |
| 41,model.layers.5.mlp.down_proj,tx.5.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.48828125,0.6005859375,0.7041015625,0.5999581813812256,0.0,0.609375,0.953125,0.5999581813812256,0.0,0.625,1.0 | |
| 42,model.layers.6.self_attn.q_proj,tx.6.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.47412109375,0.6005859375,0.7080078125,0.599853515625,0.015625,0.6328125,0.9609375,0.599853515625,0.0625,0.625,1.0 | |
| 43,model.layers.6.self_attn.k_proj,tx.6.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.46484375,0.6005859375,0.705078125,0.599853515625,0.015625,0.6328125,0.953125,0.599853515625,0.0625,0.625,1.0 | |
| 44,model.layers.6.self_attn.v_proj,tx.6.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.498046875,0.6005859375,0.685546875,0.599853515625,0.0703125,0.6171875,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 45,model.layers.6.self_attn.o_proj,tx.6.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.3115234375,0.60302734375,0.94775390625,0.599853515625,0.0234375,0.6015625,1.0,0.599853515625,0.0,0.625,1.0 | |
| 46,model.layers.6.mlp.gate_proj,tx.6.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51416015625,0.60107421875,0.658203125,0.599853515625,0.0625,0.609375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 47,model.layers.6.mlp.up_proj,tx.6.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51318359375,0.6005859375,0.64892578125,0.599853515625,0.1171875,0.6015625,1.0,0.599853515625,0.0,0.625,1.0 | |
| 48,model.layers.6.mlp.down_proj,tx.6.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.48291015625,0.60009765625,0.70654296875,0.5999581813812256,0.0078125,0.609375,0.984375,0.5999581813812256,0.0,0.625,1.0 | |
| 49,model.layers.7.self_attn.q_proj,tx.7.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.48291015625,0.60107421875,0.68408203125,0.599853515625,0.0234375,0.625,0.9765625,0.599853515625,0.0625,0.625,1.0 | |
| 50,model.layers.7.self_attn.k_proj,tx.7.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.49072265625,0.60205078125,0.6845703125,0.599853515625,0.0078125,0.625,0.8984375,0.599853515625,0.0625,0.625,1.0 | |
| 51,model.layers.7.self_attn.v_proj,tx.7.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.50537109375,0.6005859375,0.67626953125,0.599853515625,0.0703125,0.609375,0.9921875,0.599853515625,0.0625,0.625,1.0 | |
| 52,model.layers.7.self_attn.o_proj,tx.7.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.41748046875,0.5927734375,0.798828125,0.599853515625,0.0234375,0.6171875,0.9921875,0.599853515625,0.0,0.625,1.0 | |
| 53,model.layers.7.mlp.gate_proj,tx.7.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51904296875,0.6015625,0.64892578125,0.599853515625,0.046875,0.609375,0.8203125,0.599853515625,0.0,0.625,1.0 | |
| 54,model.layers.7.mlp.up_proj,tx.7.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.529296875,0.6005859375,0.650390625,0.599853515625,0.140625,0.6015625,0.8125,0.599853515625,0.0,0.625,1.0 | |
| 55,model.layers.7.mlp.down_proj,tx.7.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.49072265625,0.60107421875,0.693359375,0.5999581813812256,0.0078125,0.609375,0.9921875,0.5999581813812256,0.0,0.625,1.0 | |
| 56,model.layers.8.self_attn.q_proj,tx.8.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.48193359375,0.6005859375,0.70654296875,0.599853515625,0.015625,0.6171875,0.9296875,0.599853515625,0.0,0.625,1.0 | |
| 57,model.layers.8.self_attn.k_proj,tx.8.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.48193359375,0.60107421875,0.68701171875,0.599853515625,0.0,0.625,0.953125,0.599853515625,0.0625,0.625,1.0 | |
| 58,model.layers.8.self_attn.v_proj,tx.8.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.50537109375,0.6005859375,0.666015625,0.599853515625,0.0390625,0.609375,0.9921875,0.599853515625,0.0,0.625,1.0 | |
| 59,model.layers.8.self_attn.o_proj,tx.8.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.31689453125,0.6064453125,0.83056640625,0.599853515625,0.03125,0.609375,0.9609375,0.599853515625,0.0,0.625,1.0 | |
| 60,model.layers.8.mlp.gate_proj,tx.8.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.5166015625,0.6015625,0.6494140625,0.599853515625,0.03125,0.609375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 61,model.layers.8.mlp.up_proj,tx.8.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.5205078125,0.60107421875,0.65234375,0.599853515625,0.1015625,0.6015625,1.0,0.599853515625,0.0,0.625,1.0 | |
| 62,model.layers.8.mlp.down_proj,tx.8.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.4775390625,0.6005859375,0.705078125,0.5999581813812256,0.0390625,0.609375,0.953125,0.5999581813812256,0.0,0.625,1.0 | |
| 63,model.layers.9.self_attn.q_proj,tx.9.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.4814453125,0.60107421875,0.6943359375,0.599853515625,0.015625,0.6171875,0.9375,0.599853515625,0.0625,0.625,1.0 | |
| 64,model.layers.9.self_attn.k_proj,tx.9.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.47607421875,0.6005859375,0.68701171875,0.599853515625,0.0078125,0.6171875,0.953125,0.599853515625,0.0625,0.625,1.0 | |
| 65,model.layers.9.self_attn.v_proj,tx.9.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.5146484375,0.60009765625,0.67529296875,0.599853515625,0.0703125,0.6015625,1.0,0.599853515625,0.125,0.625,1.0 | |
| 66,model.layers.9.self_attn.o_proj,tx.9.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.337890625,0.60888671875,0.8427734375,0.599853515625,0.0625,0.6171875,0.9453125,0.599853515625,0.0,0.625,1.0 | |
| 67,model.layers.9.mlp.gate_proj,tx.9.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51318359375,0.6015625,0.65576171875,0.599853515625,0.0234375,0.609375,0.8515625,0.599853515625,0.0,0.625,1.0 | |
| 68,model.layers.9.mlp.up_proj,tx.9.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51953125,0.60107421875,0.65185546875,0.599853515625,0.0859375,0.609375,0.8359375,0.599853515625,0.0625,0.625,1.0 | |
| 69,model.layers.9.mlp.down_proj,tx.9.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.47705078125,0.60009765625,0.70703125,0.5999581813812256,0.0078125,0.609375,0.9609375,0.5999581813812256,0.0,0.625,1.0 | |
| 70,model.layers.10.self_attn.q_proj,tx.10.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.490234375,0.6005859375,0.6982421875,0.599853515625,0.015625,0.6171875,0.953125,0.599853515625,0.0625,0.625,1.0 | |
| 71,model.layers.10.self_attn.k_proj,tx.10.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4755859375,0.6005859375,0.68701171875,0.599853515625,0.0,0.625,0.9453125,0.599853515625,0.0625,0.625,1.0 | |
| 72,model.layers.10.self_attn.v_proj,tx.10.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.49951171875,0.60009765625,0.6640625,0.599853515625,0.0859375,0.609375,0.9921875,0.599853515625,0.0625,0.625,1.0 | |
| 73,model.layers.10.self_attn.o_proj,tx.10.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.2880859375,0.6279296875,0.84130859375,0.599853515625,0.015625,0.625,0.96875,0.599853515625,0.0,0.625,1.0 | |
| 74,model.layers.10.mlp.gate_proj,tx.10.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.52001953125,0.6015625,0.65625,0.599853515625,0.03125,0.609375,0.8359375,0.599853515625,0.0,0.625,1.0 | |
| 75,model.layers.10.mlp.up_proj,tx.10.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.521484375,0.6005859375,0.65576171875,0.599853515625,0.1171875,0.6015625,0.828125,0.599853515625,0.0,0.625,1.0 | |
| 76,model.layers.10.mlp.down_proj,tx.10.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.47412109375,0.6005859375,0.69921875,0.5999581813812256,0.0078125,0.6171875,0.953125,0.5999581813812256,0.0,0.625,1.0 | |
| 77,model.layers.11.self_attn.q_proj,tx.11.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.48046875,0.6005859375,0.673828125,0.599853515625,0.015625,0.609375,0.953125,0.599853515625,0.0625,0.625,1.0 | |
| 78,model.layers.11.self_attn.k_proj,tx.11.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4892578125,0.60107421875,0.67529296875,0.599853515625,0.015625,0.6171875,0.9765625,0.599853515625,0.0625,0.625,1.0 | |
| 79,model.layers.11.self_attn.v_proj,tx.11.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.50439453125,0.60009765625,0.67578125,0.599853515625,0.1171875,0.609375,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 80,model.layers.11.self_attn.o_proj,tx.11.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.3662109375,0.59521484375,0.88818359375,0.599853515625,0.03125,0.6171875,0.984375,0.599853515625,0.0,0.625,1.0 | |
| 81,model.layers.11.mlp.gate_proj,tx.11.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.517578125,0.60107421875,0.658203125,0.599853515625,0.015625,0.609375,0.828125,0.599853515625,0.0,0.625,1.0 | |
| 82,model.layers.11.mlp.up_proj,tx.11.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.52880859375,0.6005859375,0.65576171875,0.599853515625,0.109375,0.609375,0.8515625,0.599853515625,0.0,0.625,1.0 | |
| 83,model.layers.11.mlp.down_proj,tx.11.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.47607421875,0.60009765625,0.72412109375,0.5999581813812256,0.0,0.6171875,0.96875,0.5999581813812256,0.0,0.625,1.0 | |
| 84,model.layers.12.self_attn.q_proj,tx.12.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.4853515625,0.6015625,0.68701171875,0.599853515625,0.015625,0.609375,0.953125,0.599853515625,0.0,0.625,1.0 | |
| 85,model.layers.12.self_attn.k_proj,tx.12.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.47900390625,0.6015625,0.6767578125,0.599853515625,0.0234375,0.6171875,0.984375,0.599853515625,0.0625,0.625,1.0 | |
| 86,model.layers.12.self_attn.v_proj,tx.12.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.49609375,0.60009765625,0.66796875,0.599853515625,0.0703125,0.6015625,0.9921875,0.599853515625,0.0625,0.625,1.0 | |
| 87,model.layers.12.self_attn.o_proj,tx.12.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.35009765625,0.60498046875,0.8291015625,0.599853515625,0.046875,0.6171875,0.9765625,0.599853515625,0.0,0.625,1.0 | |
| 88,model.layers.12.mlp.gate_proj,tx.12.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.5107421875,0.6015625,0.65869140625,0.599853515625,0.0390625,0.609375,0.828125,0.599853515625,0.0625,0.625,1.0 | |
| 89,model.layers.12.mlp.up_proj,tx.12.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51708984375,0.6005859375,0.65576171875,0.599853515625,0.09375,0.609375,0.8203125,0.599853515625,0.0,0.625,1.0 | |
| 90,model.layers.12.mlp.down_proj,tx.12.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.4736328125,0.6015625,0.728515625,0.5999581813812256,0.015625,0.6171875,0.984375,0.5999581813812256,0.0625,0.625,1.0 | |
| 91,model.layers.13.self_attn.q_proj,tx.13.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.50244140625,0.60009765625,0.68115234375,0.599853515625,0.0078125,0.609375,0.984375,0.599853515625,0.0625,0.625,1.0 | |
| 92,model.layers.13.self_attn.k_proj,tx.13.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.49755859375,0.6015625,0.6748046875,0.599853515625,0.0234375,0.6171875,0.9921875,0.599853515625,0.0625,0.625,1.0 | |
| 93,model.layers.13.self_attn.v_proj,tx.13.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.52587890625,0.60009765625,0.6708984375,0.599853515625,0.1015625,0.609375,0.9921875,0.599853515625,0.0,0.625,1.0 | |
| 94,model.layers.13.self_attn.o_proj,tx.13.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.33837890625,0.6083984375,0.80615234375,0.599853515625,0.046875,0.625,0.9296875,0.599853515625,0.0,0.625,1.0 | |
| 95,model.layers.13.mlp.gate_proj,tx.13.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.52001953125,0.6015625,0.6650390625,0.599853515625,0.03125,0.609375,0.8125,0.599853515625,0.0625,0.625,1.0 | |
| 96,model.layers.13.mlp.up_proj,tx.13.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.53076171875,0.6005859375,0.65478515625,0.599853515625,0.109375,0.609375,0.8203125,0.599853515625,0.0625,0.625,1.0 | |
| 97,model.layers.13.mlp.down_proj,tx.13.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.474609375,0.60107421875,0.72216796875,0.5999581813812256,0.0,0.609375,0.9765625,0.5999581813812256,0.0625,0.625,1.0 | |
| 98,model.layers.14.self_attn.q_proj,tx.14.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.49169921875,0.60107421875,0.68505859375,0.599853515625,0.015625,0.6171875,0.984375,0.599853515625,0.0625,0.625,1.0 | |
| 99,model.layers.14.self_attn.k_proj,tx.14.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4931640625,0.6005859375,0.67333984375,0.599853515625,0.015625,0.6171875,0.984375,0.599853515625,0.0625,0.625,1.0 | |
| 100,model.layers.14.self_attn.v_proj,tx.14.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4736328125,0.60107421875,0.69775390625,0.599853515625,0.03125,0.6171875,0.9921875,0.599853515625,0.0625,0.625,1.0 | |
| 101,model.layers.14.self_attn.o_proj,tx.14.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.36328125,0.61328125,0.8388671875,0.599853515625,0.0234375,0.609375,0.96875,0.599853515625,0.0,0.625,1.0 | |
| 102,model.layers.14.mlp.gate_proj,tx.14.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.5146484375,0.60107421875,0.66650390625,0.599853515625,0.03125,0.6171875,0.8359375,0.599853515625,0.0,0.625,1.0 | |
| 103,model.layers.14.mlp.up_proj,tx.14.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.525390625,0.6005859375,0.6591796875,0.599853515625,0.1015625,0.609375,0.8359375,0.599853515625,0.0,0.625,1.0 | |
| 104,model.layers.14.mlp.down_proj,tx.14.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.4267578125,0.6015625,0.71484375,0.5999581813812256,0.0390625,0.6171875,0.9765625,0.5999581813812256,0.0,0.625,1.0 | |
| 105,model.layers.15.self_attn.q_proj,tx.15.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.49853515625,0.6015625,0.6767578125,0.599853515625,0.0078125,0.6171875,0.984375,0.599853515625,0.0625,0.625,1.0 | |
| 106,model.layers.15.self_attn.k_proj,tx.15.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.5009765625,0.6005859375,0.6708984375,0.599853515625,0.015625,0.625,0.984375,0.599853515625,0.0625,0.625,1.0 | |
| 107,model.layers.15.self_attn.v_proj,tx.15.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.48291015625,0.6015625,0.6767578125,0.599853515625,0.03125,0.6171875,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 108,model.layers.15.self_attn.o_proj,tx.15.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.32373046875,0.58447265625,0.9599609375,0.599853515625,0.046875,0.6015625,1.0,0.599853515625,0.0,0.625,1.0 | |
| 109,model.layers.15.mlp.gate_proj,tx.15.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51171875,0.60107421875,0.6640625,0.599853515625,0.015625,0.6171875,0.84375,0.599853515625,0.0,0.625,1.0 | |
| 110,model.layers.15.mlp.up_proj,tx.15.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.5244140625,0.6005859375,0.66259765625,0.599853515625,0.0703125,0.609375,0.84375,0.599853515625,0.0,0.625,1.0 | |
| 111,model.layers.15.mlp.down_proj,tx.15.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.4580078125,0.60009765625,0.72412109375,0.5999581813812256,0.0390625,0.609375,0.9921875,0.5999581813812256,0.0,0.625,1.0 | |
| 112,model.layers.16.self_attn.q_proj,tx.16.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.4853515625,0.60205078125,0.6884765625,0.599853515625,0.015625,0.625,0.9609375,0.599853515625,0.0,0.625,1.0 | |
| 113,model.layers.16.self_attn.k_proj,tx.16.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.490234375,0.60205078125,0.68212890625,0.599853515625,0.0,0.625,0.96875,0.599853515625,0.0625,0.625,1.0 | |
| 114,model.layers.16.self_attn.v_proj,tx.16.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.46240234375,0.6015625,0.6962890625,0.599853515625,0.0234375,0.625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 115,model.layers.16.self_attn.o_proj,tx.16.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.2744140625,0.5703125,0.99462890625,0.599853515625,0.0390625,0.5859375,1.0,0.599853515625,0.0,0.5625,1.0 | |
| 116,model.layers.16.mlp.gate_proj,tx.16.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51416015625,0.60107421875,0.669921875,0.599853515625,0.0,0.6171875,0.8671875,0.599853515625,0.0,0.625,1.0 | |
| 117,model.layers.16.mlp.up_proj,tx.16.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51953125,0.6005859375,0.65869140625,0.599853515625,0.046875,0.609375,0.875,0.599853515625,0.0,0.625,1.0 | |
| 118,model.layers.16.mlp.down_proj,tx.16.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.4609375,0.60107421875,0.72607421875,0.5999581813812256,0.015625,0.609375,0.9765625,0.5999581813812256,0.0625,0.625,1.0 | |
| 119,model.layers.17.self_attn.q_proj,tx.17.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.49853515625,0.6015625,0.67626953125,0.599853515625,0.015625,0.625,0.890625,0.599853515625,0.0625,0.625,1.0 | |
| 120,model.layers.17.self_attn.k_proj,tx.17.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.49365234375,0.6015625,0.68115234375,0.599853515625,0.015625,0.6328125,0.953125,0.599853515625,0.0625,0.625,1.0 | |
| 121,model.layers.17.self_attn.v_proj,tx.17.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.49560546875,0.6005859375,0.6728515625,0.599853515625,0.0859375,0.6171875,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 122,model.layers.17.self_attn.o_proj,tx.17.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.31689453125,0.564453125,0.99560546875,0.599853515625,0.0625,0.578125,1.0,0.599853515625,0.0,0.5625,1.0 | |
| 123,model.layers.17.mlp.gate_proj,tx.17.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.5126953125,0.6005859375,0.662109375,0.599853515625,0.0234375,0.6171875,0.875,0.599853515625,0.0,0.625,1.0 | |
| 124,model.layers.17.mlp.up_proj,tx.17.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51904296875,0.6005859375,0.66455078125,0.599853515625,0.046875,0.609375,0.8828125,0.599853515625,0.0,0.625,1.0 | |
| 125,model.layers.17.mlp.down_proj,tx.17.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.48291015625,0.60009765625,0.70849609375,0.5999581813812256,0.03125,0.6171875,0.9921875,0.5999581813812256,0.0625,0.625,1.0 | |
| 126,model.layers.18.self_attn.q_proj,tx.18.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.49853515625,0.60107421875,0.68798828125,0.599853515625,0.0234375,0.625,0.96875,0.599853515625,0.0625,0.625,1.0 | |
| 127,model.layers.18.self_attn.k_proj,tx.18.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4912109375,0.60107421875,0.70361328125,0.599853515625,0.0078125,0.6328125,0.96875,0.599853515625,0.0625,0.625,1.0 | |
| 128,model.layers.18.self_attn.v_proj,tx.18.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.46630859375,0.6015625,0.69775390625,0.599853515625,0.03125,0.625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 129,model.layers.18.self_attn.o_proj,tx.18.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.23779296875,0.55126953125,1.0,0.599853515625,0.0234375,0.5703125,1.0,0.599853515625,0.0,0.5625,1.0 | |
| 130,model.layers.18.mlp.gate_proj,tx.18.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51123046875,0.60107421875,0.6640625,0.599853515625,0.015625,0.6171875,0.859375,0.599853515625,0.0,0.625,1.0 | |
| 131,model.layers.18.mlp.up_proj,tx.18.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.52099609375,0.6005859375,0.66552734375,0.599853515625,0.0625,0.609375,0.9296875,0.599853515625,0.0,0.625,1.0 | |
| 132,model.layers.18.mlp.down_proj,tx.18.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.46923828125,0.60009765625,0.70703125,0.5999581813812256,0.015625,0.6171875,0.984375,0.5999581813812256,0.0,0.625,1.0 | |
| 133,model.layers.19.self_attn.q_proj,tx.19.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.48486328125,0.60205078125,0.6884765625,0.599853515625,0.015625,0.625,0.8671875,0.599853515625,0.0,0.625,1.0 | |
| 134,model.layers.19.self_attn.k_proj,tx.19.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4775390625,0.60205078125,0.71533203125,0.599853515625,0.0078125,0.6328125,0.9609375,0.599853515625,0.0625,0.625,1.0 | |
| 135,model.layers.19.self_attn.v_proj,tx.19.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4912109375,0.6015625,0.697265625,0.599853515625,0.03125,0.625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 136,model.layers.19.self_attn.o_proj,tx.19.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.27978515625,0.4970703125,1.0,0.599853515625,0.0234375,0.5390625,1.0,0.599853515625,0.0,0.5625,1.0 | |
| 137,model.layers.19.mlp.gate_proj,tx.19.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51171875,0.60107421875,0.66162109375,0.599853515625,0.015625,0.609375,0.875,0.599853515625,0.0,0.625,1.0 | |
| 138,model.layers.19.mlp.up_proj,tx.19.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.5244140625,0.6005859375,0.66064453125,0.599853515625,0.0703125,0.609375,0.9140625,0.599853515625,0.0,0.625,1.0 | |
| 139,model.layers.19.mlp.down_proj,tx.19.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.46533203125,0.6005859375,0.71630859375,0.5999581813812256,0.015625,0.625,0.96875,0.5999581813812256,0.0,0.625,1.0 | |
| 140,model.layers.20.self_attn.q_proj,tx.20.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.478515625,0.6015625,0.69287109375,0.599853515625,0.015625,0.625,0.875,0.599853515625,0.0625,0.625,1.0 | |
| 141,model.layers.20.self_attn.k_proj,tx.20.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.48095703125,0.60205078125,0.693359375,0.599853515625,0.0078125,0.6328125,0.9609375,0.599853515625,0.0625,0.625,1.0 | |
| 142,model.layers.20.self_attn.v_proj,tx.20.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.46875,0.60107421875,0.705078125,0.599853515625,0.0234375,0.625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 143,model.layers.20.self_attn.o_proj,tx.20.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.28125,0.56591796875,0.99365234375,0.599853515625,0.03125,0.5625,1.0,0.599853515625,0.0,0.5625,1.0 | |
| 144,model.layers.20.mlp.gate_proj,tx.20.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.50927734375,0.60107421875,0.6630859375,0.599853515625,0.015625,0.6171875,0.875,0.599853515625,0.0,0.625,1.0 | |
| 145,model.layers.20.mlp.up_proj,tx.20.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51953125,0.6005859375,0.66064453125,0.599853515625,0.0625,0.609375,0.9296875,0.599853515625,0.0,0.625,1.0 | |
| 146,model.layers.20.mlp.down_proj,tx.20.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.48193359375,0.6005859375,0.71826171875,0.5999581813812256,0.015625,0.625,0.9609375,0.5999581813812256,0.0625,0.625,1.0 | |
| 147,model.layers.21.self_attn.q_proj,tx.21.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.48388671875,0.60205078125,0.685546875,0.599853515625,0.0078125,0.625,0.8671875,0.599853515625,0.0625,0.625,1.0 | |
| 148,model.layers.21.self_attn.k_proj,tx.21.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.48583984375,0.60205078125,0.68994140625,0.599853515625,0.0,0.6328125,0.8984375,0.599853515625,0.0625,0.625,1.0 | |
| 149,model.layers.21.self_attn.v_proj,tx.21.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.47705078125,0.60107421875,0.68701171875,0.599853515625,0.0234375,0.625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 150,model.layers.21.self_attn.o_proj,tx.21.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.3125,0.564453125,1.0,0.599853515625,0.0078125,0.578125,1.0,0.599853515625,0.0,0.5625,1.0 | |
| 151,model.layers.21.mlp.gate_proj,tx.21.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.513671875,0.6005859375,0.66748046875,0.599853515625,0.0,0.6171875,0.8984375,0.599853515625,0.0,0.625,1.0 | |
| 152,model.layers.21.mlp.up_proj,tx.21.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51171875,0.6005859375,0.65966796875,0.599853515625,0.0625,0.609375,0.9453125,0.599853515625,0.0,0.625,1.0 | |
| 153,model.layers.21.mlp.down_proj,tx.21.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.44482421875,0.60107421875,0.708984375,0.5999581813812256,0.0234375,0.625,0.953125,0.5999581813812256,0.0625,0.625,1.0 | |
| 154,model.layers.22.self_attn.q_proj,tx.22.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.47705078125,0.60107421875,0.6943359375,0.599853515625,0.0078125,0.625,0.859375,0.599853515625,0.0,0.625,1.0 | |
| 155,model.layers.22.self_attn.k_proj,tx.22.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4873046875,0.6015625,0.69482421875,0.599853515625,0.0078125,0.6328125,0.90625,0.599853515625,0.0625,0.625,1.0 | |
| 156,model.layers.22.self_attn.v_proj,tx.22.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.48681640625,0.60107421875,0.70263671875,0.599853515625,0.0234375,0.625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 157,model.layers.22.self_attn.o_proj,tx.22.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.263671875,0.525390625,1.0,0.599853515625,0.03125,0.546875,1.0,0.599853515625,0.0,0.5625,1.0 | |
| 158,model.layers.22.mlp.gate_proj,tx.22.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.50927734375,0.60107421875,0.66064453125,0.599853515625,0.015625,0.609375,0.8671875,0.599853515625,0.0,0.625,1.0 | |
| 159,model.layers.22.mlp.up_proj,tx.22.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.50927734375,0.6005859375,0.66015625,0.599853515625,0.046875,0.609375,0.9375,0.599853515625,0.0,0.625,1.0 | |
| 160,model.layers.22.mlp.down_proj,tx.22.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.462890625,0.60107421875,0.71484375,0.5999581813812256,0.015625,0.6328125,0.9375,0.5999581813812256,0.0,0.625,1.0 | |
| 161,model.layers.23.self_attn.q_proj,tx.23.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.47705078125,0.60205078125,0.69140625,0.599853515625,0.0,0.625,0.8671875,0.599853515625,0.0,0.625,1.0 | |
| 162,model.layers.23.self_attn.k_proj,tx.23.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4775390625,0.6015625,0.69287109375,0.599853515625,0.0,0.6328125,0.9140625,0.599853515625,0.0625,0.625,1.0 | |
| 163,model.layers.23.self_attn.v_proj,tx.23.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4912109375,0.60107421875,0.69140625,0.599853515625,0.0234375,0.625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 164,model.layers.23.self_attn.o_proj,tx.23.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.3017578125,0.56884765625,0.99951171875,0.599853515625,0.015625,0.578125,1.0,0.599853515625,0.0,0.5625,1.0 | |
| 165,model.layers.23.mlp.gate_proj,tx.23.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.513671875,0.60107421875,0.66015625,0.599853515625,0.0078125,0.609375,0.875,0.599853515625,0.0,0.625,1.0 | |
| 166,model.layers.23.mlp.up_proj,tx.23.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.5146484375,0.60107421875,0.662109375,0.599853515625,0.046875,0.609375,0.921875,0.599853515625,0.0,0.625,1.0 | |
| 167,model.layers.23.mlp.down_proj,tx.23.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.46044921875,0.60205078125,0.708984375,0.5999581813812256,0.0078125,0.6328125,0.9140625,0.5999581813812256,0.0,0.625,1.0 | |
| 168,model.layers.24.self_attn.q_proj,tx.24.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.4755859375,0.6015625,0.685546875,0.599853515625,0.0078125,0.6328125,0.859375,0.599853515625,0.0625,0.625,1.0 | |
| 169,model.layers.24.self_attn.k_proj,tx.24.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4912109375,0.60205078125,0.7080078125,0.599853515625,0.0,0.6328125,0.890625,0.599853515625,0.0625,0.625,1.0 | |
| 170,model.layers.24.self_attn.v_proj,tx.24.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.474609375,0.6015625,0.705078125,0.599853515625,0.0234375,0.625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 171,model.layers.24.self_attn.o_proj,tx.24.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.26171875,0.55029296875,1.0,0.599853515625,0.0234375,0.5625,1.0,0.599853515625,0.0,0.5625,1.0 | |
| 172,model.layers.24.mlp.gate_proj,tx.24.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.515625,0.6005859375,0.6640625,0.599853515625,0.0078125,0.609375,0.890625,0.599853515625,0.0,0.625,1.0 | |
| 173,model.layers.24.mlp.up_proj,tx.24.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.52392578125,0.60107421875,0.65966796875,0.599853515625,0.046875,0.609375,0.9296875,0.599853515625,0.0625,0.625,1.0 | |
| 174,model.layers.24.mlp.down_proj,tx.24.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.4423828125,0.60205078125,0.71435546875,0.5999581813812256,0.0,0.640625,0.9140625,0.5999581813812256,0.0625,0.625,1.0 | |
| 175,model.layers.25.self_attn.q_proj,tx.25.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.49365234375,0.60107421875,0.69921875,0.599853515625,0.0,0.6328125,0.8671875,0.599853515625,0.0625,0.625,1.0 | |
| 176,model.layers.25.self_attn.k_proj,tx.25.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4794921875,0.60205078125,0.72607421875,0.599853515625,0.0,0.6328125,0.8984375,0.599853515625,0.0,0.625,1.0 | |
| 177,model.layers.25.self_attn.v_proj,tx.25.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.46337890625,0.6005859375,0.69482421875,0.599853515625,0.0234375,0.625,1.0,0.599853515625,0.125,0.625,1.0 | |
| 178,model.layers.25.self_attn.o_proj,tx.25.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.2294921875,0.62451171875,1.0,0.599853515625,0.0234375,0.609375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 179,model.layers.25.mlp.gate_proj,tx.25.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.52197265625,0.60107421875,0.6591796875,0.599853515625,0.015625,0.609375,0.8984375,0.599853515625,0.0,0.625,1.0 | |
| 180,model.layers.25.mlp.up_proj,tx.25.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.52099609375,0.60107421875,0.658203125,0.599853515625,0.046875,0.609375,0.90625,0.599853515625,0.0625,0.625,1.0 | |
| 181,model.layers.25.mlp.down_proj,tx.25.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.46484375,0.60205078125,0.72265625,0.5999581813812256,0.0078125,0.640625,0.90625,0.5999581813812256,0.0,0.625,1.0 | |
| 182,model.layers.26.self_attn.q_proj,tx.26.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.4833984375,0.6015625,0.68408203125,0.599853515625,0.0,0.6328125,0.890625,0.599853515625,0.0625,0.625,1.0 | |
| 183,model.layers.26.self_attn.k_proj,tx.26.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.46728515625,0.60205078125,0.71142578125,0.599853515625,0.0,0.6328125,0.90625,0.599853515625,0.0625,0.625,1.0 | |
| 184,model.layers.26.self_attn.v_proj,tx.26.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.45947265625,0.6015625,0.6875,0.599853515625,0.015625,0.6171875,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 185,model.layers.26.self_attn.o_proj,tx.26.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.2255859375,0.541015625,1.0,0.599853515625,0.015625,0.5390625,1.0,0.599853515625,0.0,0.5625,1.0 | |
| 186,model.layers.26.mlp.gate_proj,tx.26.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.5185546875,0.60107421875,0.66650390625,0.599853515625,0.0078125,0.609375,0.8984375,0.599853515625,0.0,0.625,1.0 | |
| 187,model.layers.26.mlp.up_proj,tx.26.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51806640625,0.60107421875,0.6591796875,0.599853515625,0.03125,0.609375,0.921875,0.599853515625,0.0,0.625,1.0 | |
| 188,model.layers.26.mlp.down_proj,tx.26.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.4052734375,0.60107421875,0.7265625,0.5999581813812256,0.0078125,0.6484375,0.921875,0.5999581813812256,0.0,0.625,1.0 | |
| 189,model.layers.27.self_attn.q_proj,tx.27.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.48291015625,0.6015625,0.69677734375,0.599853515625,0.0,0.6328125,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 190,model.layers.27.self_attn.k_proj,tx.27.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.48486328125,0.60205078125,0.71630859375,0.599853515625,0.0,0.6328125,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 191,model.layers.27.self_attn.v_proj,tx.27.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.47705078125,0.60205078125,0.70458984375,0.599853515625,0.0234375,0.625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 192,model.layers.27.self_attn.o_proj,tx.27.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.18310546875,0.5771484375,1.0,0.599853515625,0.0078125,0.578125,1.0,0.599853515625,0.0,0.5625,1.0 | |
| 193,model.layers.27.mlp.gate_proj,tx.27.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.5126953125,0.60107421875,0.6630859375,0.599853515625,0.015625,0.609375,0.8828125,0.599853515625,0.0,0.625,1.0 | |
| 194,model.layers.27.mlp.up_proj,tx.27.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51708984375,0.6015625,0.66552734375,0.599853515625,0.046875,0.609375,0.8984375,0.599853515625,0.0,0.625,1.0 | |
| 195,model.layers.27.mlp.down_proj,tx.27.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.44091796875,0.60107421875,0.72412109375,0.5999581813812256,0.0,0.65625,0.921875,0.5999581813812256,0.0,0.625,1.0 | |
| 196,model.layers.28.self_attn.q_proj,tx.28.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.4794921875,0.6015625,0.693359375,0.599853515625,0.0,0.6328125,0.875,0.599853515625,0.0,0.625,1.0 | |
| 197,model.layers.28.self_attn.k_proj,tx.28.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.47607421875,0.6015625,0.6923828125,0.599853515625,0.0,0.6328125,0.90625,0.599853515625,0.0625,0.625,1.0 | |
| 198,model.layers.28.self_attn.v_proj,tx.28.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.49365234375,0.6005859375,0.6943359375,0.599853515625,0.0234375,0.6171875,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 199,model.layers.28.self_attn.o_proj,tx.28.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.2392578125,0.587890625,0.9990234375,0.599853515625,0.03125,0.6015625,1.0,0.599853515625,0.0,0.625,1.0 | |
| 200,model.layers.28.mlp.gate_proj,tx.28.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51318359375,0.6015625,0.6767578125,0.599853515625,0.015625,0.6171875,0.859375,0.599853515625,0.0625,0.625,1.0 | |
| 201,model.layers.28.mlp.up_proj,tx.28.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51318359375,0.6015625,0.66259765625,0.599853515625,0.03125,0.609375,0.859375,0.599853515625,0.0625,0.625,1.0 | |
| 202,model.layers.28.mlp.down_proj,tx.28.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.41943359375,0.60302734375,0.74169921875,0.5999581813812256,0.0078125,0.6640625,0.9296875,0.5999581813812256,0.0625,0.625,1.0 | |
| 203,model.layers.29.self_attn.q_proj,tx.29.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.48681640625,0.6015625,0.69775390625,0.599853515625,0.0,0.625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 204,model.layers.29.self_attn.k_proj,tx.29.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4970703125,0.60205078125,0.69384765625,0.599853515625,0.0,0.625,1.0,0.599853515625,0.125,0.625,1.0 | |
| 205,model.layers.29.self_attn.v_proj,tx.29.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.5,0.60107421875,0.68310546875,0.599853515625,0.015625,0.6171875,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 206,model.layers.29.self_attn.o_proj,tx.29.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.134765625,0.57470703125,0.9921875,0.599853515625,0.015625,0.5703125,1.0,0.599853515625,0.0,0.625,1.0 | |
| 207,model.layers.29.mlp.gate_proj,tx.29.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51025390625,0.6015625,0.66943359375,0.599853515625,0.0078125,0.6171875,0.8828125,0.599853515625,0.0,0.625,1.0 | |
| 208,model.layers.29.mlp.up_proj,tx.29.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.5029296875,0.6015625,0.66748046875,0.599853515625,0.015625,0.6171875,0.8671875,0.599853515625,0.0625,0.625,1.0 | |
| 209,model.layers.29.mlp.down_proj,tx.29.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.44189453125,0.603515625,0.73583984375,0.5999581813812256,0.0,0.6640625,0.921875,0.5999581813812256,0.0,0.625,1.0 | |
| 210,model.layers.30.self_attn.q_proj,tx.30.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.45263671875,0.60107421875,0.69091796875,0.599853515625,0.0,0.6328125,0.8984375,0.599853515625,0.0625,0.625,1.0 | |
| 211,model.layers.30.self_attn.k_proj,tx.30.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.47412109375,0.60107421875,0.69677734375,0.599853515625,0.0,0.640625,0.921875,0.599853515625,0.0625,0.625,1.0 | |
| 212,model.layers.30.self_attn.v_proj,tx.30.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4755859375,0.6015625,0.69384765625,0.599853515625,0.0234375,0.6171875,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 213,model.layers.30.self_attn.o_proj,tx.30.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.23876953125,0.5595703125,0.99462890625,0.599853515625,0.0546875,0.5546875,1.0,0.599853515625,0.0,0.5625,1.0 | |
| 214,model.layers.30.mlp.gate_proj,tx.30.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51318359375,0.60107421875,0.67041015625,0.599853515625,0.015625,0.6171875,0.9609375,0.599853515625,0.0,0.625,1.0 | |
| 215,model.layers.30.mlp.up_proj,tx.30.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51318359375,0.60107421875,0.67138671875,0.599853515625,0.03125,0.6171875,0.953125,0.599853515625,0.0625,0.625,1.0 | |
| 216,model.layers.30.mlp.down_proj,tx.30.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.408203125,0.6025390625,0.7666015625,0.5999581813812256,0.0,0.671875,0.96875,0.5999581813812256,0.0,0.625,1.0 | |
| 217,model.layers.31.self_attn.q_proj,tx.31.attn.q,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.47607421875,0.60009765625,0.6943359375,0.599853515625,0.0,0.625,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 218,model.layers.31.self_attn.k_proj,tx.31.attn.k,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.4775390625,0.60107421875,0.6884765625,0.599853515625,0.0078125,0.625,0.9765625,0.599853515625,0.0625,0.625,1.0 | |
| 219,model.layers.31.self_attn.v_proj,tx.31.attn.v,Linear,weight,"[1024, 4096]",4194304,1678336,0.599853515625,"(128, 16)",8 x 256,2048,0.599853515625,0.47412109375,0.60009765625,0.69873046875,0.599853515625,0.0078125,0.6171875,1.0,0.599853515625,0.0625,0.625,1.0 | |
| 220,model.layers.31.self_attn.o_proj,tx.31.attn.o,Linear,weight,"[4096, 4096]",16777216,6713344,0.599853515625,"(128, 16)",32 x 256,8192,0.599853515625,0.10791015625,0.603515625,1.0,0.599853515625,0.0,0.609375,1.0,0.599853515625,0.0,0.625,1.0 | |
| 221,model.layers.31.mlp.gate_proj,tx.31.mlp.gate,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51220703125,0.60009765625,0.67138671875,0.599853515625,0.0703125,0.6171875,0.9609375,0.599853515625,0.0625,0.625,1.0 | |
| 222,model.layers.31.mlp.up_proj,tx.31.mlp.up,Linear,weight,"[14336, 4096]",58720256,23496704,0.599853515625,"(128, 16)",112 x 256,28672,0.599853515625,0.51513671875,0.60009765625,0.67529296875,0.599853515625,0.0703125,0.6171875,0.9296875,0.599853515625,0.0,0.625,1.0 | |
| 223,model.layers.31.mlp.down_proj,tx.31.mlp.down,Linear,weight,"[4096, 14336]",58720256,23490560,0.5999581217765808,"(128, 16)",32 x 896,28672,0.5999581813812256,0.37353515625,0.59912109375,0.77978515625,0.5999581813812256,0.0,0.6953125,1.0,0.5999581813812256,0.0,0.625,1.0 | |
| 224,lm_head,lm_head,Linear,weight,"[128256, 4096]",525336576,525336576,0.0,"(128, 16)",1002 x 256,256512,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0 | |