| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 3.0, |
| "eval_steps": 500, |
| "global_step": 150, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.02, |
| "grad_norm": 0.01196271863296122, |
| "learning_rate": 5.333333333333334e-06, |
| "loss": 0.1729, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 0.012495676529254032, |
| "learning_rate": 1.0666666666666667e-05, |
| "loss": 0.1737, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 0.011667181840412507, |
| "learning_rate": 1.6000000000000003e-05, |
| "loss": 0.1694, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 0.01135971709343269, |
| "learning_rate": 2.1333333333333335e-05, |
| "loss": 0.1695, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 0.011980162405667093, |
| "learning_rate": 2.6666666666666667e-05, |
| "loss": 0.1718, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.12, |
| "grad_norm": 0.012718313075719344, |
| "learning_rate": 3.2000000000000005e-05, |
| "loss": 0.171, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.14, |
| "grad_norm": 0.013386177887216561, |
| "learning_rate": 3.733333333333334e-05, |
| "loss": 0.1688, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.16, |
| "grad_norm": 0.015529179806394781, |
| "learning_rate": 4.266666666666667e-05, |
| "loss": 0.171, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.18, |
| "grad_norm": 0.017127729559861825, |
| "learning_rate": 4.8e-05, |
| "loss": 0.1679, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.2, |
| "grad_norm": 0.017643415841480662, |
| "learning_rate": 5.333333333333333e-05, |
| "loss": 0.1651, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.22, |
| "grad_norm": 0.01854355719667723, |
| "learning_rate": 5.8666666666666665e-05, |
| "loss": 0.1606, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.24, |
| "grad_norm": 0.01693555621022761, |
| "learning_rate": 6.400000000000001e-05, |
| "loss": 0.1564, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.26, |
| "grad_norm": 0.017659872824557738, |
| "learning_rate": 6.933333333333334e-05, |
| "loss": 0.1527, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.28, |
| "grad_norm": 0.019459518002822133, |
| "learning_rate": 7.466666666666667e-05, |
| "loss": 0.1478, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.3, |
| "grad_norm": 0.019901019610822045, |
| "learning_rate": 8e-05, |
| "loss": 0.1463, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.32, |
| "grad_norm": 0.018446252316564236, |
| "learning_rate": 7.998916964717848e-05, |
| "loss": 0.1393, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.34, |
| "grad_norm": 0.017867571731070883, |
| "learning_rate": 7.9956684453541e-05, |
| "loss": 0.1367, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.36, |
| "grad_norm": 0.02225189447914238, |
| "learning_rate": 7.990256201039297e-05, |
| "loss": 0.132, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.38, |
| "grad_norm": 0.024032477518503167, |
| "learning_rate": 7.982683162599218e-05, |
| "loss": 0.132, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.4, |
| "grad_norm": 0.022379629226120565, |
| "learning_rate": 7.972953430967773e-05, |
| "loss": 0.1235, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.42, |
| "grad_norm": 0.014679560897470098, |
| "learning_rate": 7.961072274966282e-05, |
| "loss": 0.1192, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.44, |
| "grad_norm": 0.012348486806842536, |
| "learning_rate": 7.947046128450319e-05, |
| "loss": 0.1142, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.46, |
| "grad_norm": 0.011718867067469581, |
| "learning_rate": 7.930882586825653e-05, |
| "loss": 0.1141, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.48, |
| "grad_norm": 0.011090541605219488, |
| "learning_rate": 7.912590402935223e-05, |
| "loss": 0.1069, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.5, |
| "grad_norm": 0.011263401354320204, |
| "learning_rate": 7.892179482319297e-05, |
| "loss": 0.1025, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.52, |
| "grad_norm": 0.010805318974544694, |
| "learning_rate": 7.869660877851456e-05, |
| "loss": 0.1015, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.54, |
| "grad_norm": 0.01136088979963782, |
| "learning_rate": 7.845046783753276e-05, |
| "loss": 0.1004, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.56, |
| "grad_norm": 0.010483311106243808, |
| "learning_rate": 7.818350528990929e-05, |
| "loss": 0.0912, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.58, |
| "grad_norm": 0.011051913804904468, |
| "learning_rate": 7.789586570057317e-05, |
| "loss": 0.0871, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.6, |
| "grad_norm": 0.011308977275479132, |
| "learning_rate": 7.758770483143634e-05, |
| "loss": 0.0828, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.62, |
| "grad_norm": 0.011604739819033429, |
| "learning_rate": 7.72591895570457e-05, |
| "loss": 0.0806, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.64, |
| "grad_norm": 0.011548436429341773, |
| "learning_rate": 7.69104977742177e-05, |
| "loss": 0.0764, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.66, |
| "grad_norm": 0.009950507093629006, |
| "learning_rate": 7.654181830570404e-05, |
| "loss": 0.0763, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.68, |
| "grad_norm": 0.010058021232671382, |
| "learning_rate": 7.615335079794083e-05, |
| "loss": 0.0738, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.7, |
| "grad_norm": 0.010723017492533498, |
| "learning_rate": 7.57453056129365e-05, |
| "loss": 0.0687, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.72, |
| "grad_norm": 0.009183653470914266, |
| "learning_rate": 7.531790371435709e-05, |
| "loss": 0.0694, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.74, |
| "grad_norm": 0.009580743833606489, |
| "learning_rate": 7.48713765478705e-05, |
| "loss": 0.0689, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.76, |
| "grad_norm": 0.010503324369714725, |
| "learning_rate": 7.440596591581463e-05, |
| "loss": 0.0642, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.78, |
| "grad_norm": 0.01038134566129547, |
| "learning_rate": 7.392192384625704e-05, |
| "loss": 0.0634, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.8, |
| "grad_norm": 0.011275300914278943, |
| "learning_rate": 7.341951245651747e-05, |
| "loss": 0.0645, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.82, |
| "grad_norm": 0.010114318132651946, |
| "learning_rate": 7.28990038112265e-05, |
| "loss": 0.0587, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.84, |
| "grad_norm": 0.011577914829038317, |
| "learning_rate": 7.236067977499791e-05, |
| "loss": 0.0587, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.86, |
| "grad_norm": 0.01285048907312673, |
| "learning_rate": 7.180483185979392e-05, |
| "loss": 0.0564, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.88, |
| "grad_norm": 0.011208959430733204, |
| "learning_rate": 7.123176106706638e-05, |
| "loss": 0.0561, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.9, |
| "grad_norm": 0.014928383275488958, |
| "learning_rate": 7.064177772475912e-05, |
| "loss": 0.0568, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.92, |
| "grad_norm": 0.012915599822153272, |
| "learning_rate": 7.003520131925997e-05, |
| "loss": 0.0548, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.94, |
| "grad_norm": 0.012522403535772119, |
| "learning_rate": 6.941236032239316e-05, |
| "loss": 0.0517, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.96, |
| "grad_norm": 0.011060690333286749, |
| "learning_rate": 6.877359201354606e-05, |
| "loss": 0.0507, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.98, |
| "grad_norm": 0.010717596749382448, |
| "learning_rate": 6.811924229702648e-05, |
| "loss": 0.051, |
| "step": 49 |
| }, |
| { |
| "epoch": 1.0, |
| "grad_norm": 0.01168606880435161, |
| "learning_rate": 6.744966551474936e-05, |
| "loss": 0.0494, |
| "step": 50 |
| }, |
| { |
| "epoch": 1.02, |
| "grad_norm": 0.010234880530364953, |
| "learning_rate": 6.676522425435433e-05, |
| "loss": 0.0494, |
| "step": 51 |
| }, |
| { |
| "epoch": 1.04, |
| "grad_norm": 0.010118289243122447, |
| "learning_rate": 6.606628915285822e-05, |
| "loss": 0.0484, |
| "step": 52 |
| }, |
| { |
| "epoch": 1.06, |
| "grad_norm": 0.009213382658341502, |
| "learning_rate": 6.53532386959484e-05, |
| "loss": 0.0437, |
| "step": 53 |
| }, |
| { |
| "epoch": 1.08, |
| "grad_norm": 0.012844445869941712, |
| "learning_rate": 6.462645901302633e-05, |
| "loss": 0.0456, |
| "step": 54 |
| }, |
| { |
| "epoch": 1.1, |
| "grad_norm": 0.009518428990903768, |
| "learning_rate": 6.388634366811146e-05, |
| "loss": 0.0451, |
| "step": 55 |
| }, |
| { |
| "epoch": 1.12, |
| "grad_norm": 0.009734225068738799, |
| "learning_rate": 6.313329344671946e-05, |
| "loss": 0.0455, |
| "step": 56 |
| }, |
| { |
| "epoch": 1.1400000000000001, |
| "grad_norm": 0.012155363298661265, |
| "learning_rate": 6.236771613882987e-05, |
| "loss": 0.0464, |
| "step": 57 |
| }, |
| { |
| "epoch": 1.16, |
| "grad_norm": 0.011597817236694865, |
| "learning_rate": 6.159002631806052e-05, |
| "loss": 0.0475, |
| "step": 58 |
| }, |
| { |
| "epoch": 1.18, |
| "grad_norm": 0.01030399528687484, |
| "learning_rate": 6.0800645117168616e-05, |
| "loss": 0.0449, |
| "step": 59 |
| }, |
| { |
| "epoch": 1.2, |
| "grad_norm": 0.008733695109158793, |
| "learning_rate": 6.000000000000001e-05, |
| "loss": 0.0414, |
| "step": 60 |
| }, |
| { |
| "epoch": 1.22, |
| "grad_norm": 0.008433218506851681, |
| "learning_rate": 5.918852453000986e-05, |
| "loss": 0.0414, |
| "step": 61 |
| }, |
| { |
| "epoch": 1.24, |
| "grad_norm": 0.009845710768130733, |
| "learning_rate": 5.836665813548047e-05, |
| "loss": 0.0418, |
| "step": 62 |
| }, |
| { |
| "epoch": 1.26, |
| "grad_norm": 0.009546673206550652, |
| "learning_rate": 5.75348458715631e-05, |
| "loss": 0.0435, |
| "step": 63 |
| }, |
| { |
| "epoch": 1.28, |
| "grad_norm": 0.010315661407117281, |
| "learning_rate": 5.669353817927272e-05, |
| "loss": 0.0429, |
| "step": 64 |
| }, |
| { |
| "epoch": 1.3, |
| "grad_norm": 0.007168202098824094, |
| "learning_rate": 5.584319064156628e-05, |
| "loss": 0.0424, |
| "step": 65 |
| }, |
| { |
| "epoch": 1.32, |
| "grad_norm": 0.008967213164891482, |
| "learning_rate": 5.4984263736636494e-05, |
| "loss": 0.0429, |
| "step": 66 |
| }, |
| { |
| "epoch": 1.34, |
| "grad_norm": 0.008431841927346365, |
| "learning_rate": 5.4117222588554756e-05, |
| "loss": 0.041, |
| "step": 67 |
| }, |
| { |
| "epoch": 1.3599999999999999, |
| "grad_norm": 0.008649261047415096, |
| "learning_rate": 5.324253671539833e-05, |
| "loss": 0.0429, |
| "step": 68 |
| }, |
| { |
| "epoch": 1.38, |
| "grad_norm": 0.006959029177449907, |
| "learning_rate": 5.23606797749979e-05, |
| "loss": 0.0424, |
| "step": 69 |
| }, |
| { |
| "epoch": 1.4, |
| "grad_norm": 0.006913063489881499, |
| "learning_rate": 5.1472129308443616e-05, |
| "loss": 0.0438, |
| "step": 70 |
| }, |
| { |
| "epoch": 1.42, |
| "grad_norm": 0.007774662572569623, |
| "learning_rate": 5.05773664814881e-05, |
| "loss": 0.0434, |
| "step": 71 |
| }, |
| { |
| "epoch": 1.44, |
| "grad_norm": 0.009010151992295618, |
| "learning_rate": 4.967687582398671e-05, |
| "loss": 0.0414, |
| "step": 72 |
| }, |
| { |
| "epoch": 1.46, |
| "grad_norm": 0.008168394295204415, |
| "learning_rate": 4.877114496751613e-05, |
| "loss": 0.0418, |
| "step": 73 |
| }, |
| { |
| "epoch": 1.48, |
| "grad_norm": 0.00690105803863704, |
| "learning_rate": 4.786066438131321e-05, |
| "loss": 0.0402, |
| "step": 74 |
| }, |
| { |
| "epoch": 1.5, |
| "grad_norm": 0.006977685638593046, |
| "learning_rate": 4.694592710667723e-05, |
| "loss": 0.0421, |
| "step": 75 |
| }, |
| { |
| "epoch": 1.52, |
| "grad_norm": 0.006958637351491174, |
| "learning_rate": 4.602742848997933e-05, |
| "loss": 0.0407, |
| "step": 76 |
| }, |
| { |
| "epoch": 1.54, |
| "grad_norm": 0.007598629736577876, |
| "learning_rate": 4.51056659144238e-05, |
| "loss": 0.0411, |
| "step": 77 |
| }, |
| { |
| "epoch": 1.56, |
| "grad_norm": 0.007588692623112087, |
| "learning_rate": 4.418113853070614e-05, |
| "loss": 0.0422, |
| "step": 78 |
| }, |
| { |
| "epoch": 1.58, |
| "grad_norm": 0.006489717572033393, |
| "learning_rate": 4.3254346986714334e-05, |
| "loss": 0.0403, |
| "step": 79 |
| }, |
| { |
| "epoch": 1.6, |
| "grad_norm": 0.006565260886424596, |
| "learning_rate": 4.2325793156419035e-05, |
| "loss": 0.0409, |
| "step": 80 |
| }, |
| { |
| "epoch": 1.62, |
| "grad_norm": 0.0062335119032897734, |
| "learning_rate": 4.139597986810005e-05, |
| "loss": 0.039, |
| "step": 81 |
| }, |
| { |
| "epoch": 1.6400000000000001, |
| "grad_norm": 0.007321115299500603, |
| "learning_rate": 4.046541063205589e-05, |
| "loss": 0.0407, |
| "step": 82 |
| }, |
| { |
| "epoch": 1.6600000000000001, |
| "grad_norm": 0.006145604116162107, |
| "learning_rate": 3.953458936794413e-05, |
| "loss": 0.0405, |
| "step": 83 |
| }, |
| { |
| "epoch": 1.6800000000000002, |
| "grad_norm": 0.006395161706327024, |
| "learning_rate": 3.860402013189998e-05, |
| "loss": 0.0412, |
| "step": 84 |
| }, |
| { |
| "epoch": 1.7, |
| "grad_norm": 0.0065954269271041865, |
| "learning_rate": 3.767420684358097e-05, |
| "loss": 0.0429, |
| "step": 85 |
| }, |
| { |
| "epoch": 1.72, |
| "grad_norm": 0.0066335654461026445, |
| "learning_rate": 3.674565301328568e-05, |
| "loss": 0.0399, |
| "step": 86 |
| }, |
| { |
| "epoch": 1.74, |
| "grad_norm": 0.006300324840837113, |
| "learning_rate": 3.581886146929387e-05, |
| "loss": 0.0419, |
| "step": 87 |
| }, |
| { |
| "epoch": 1.76, |
| "grad_norm": 0.006722496244893889, |
| "learning_rate": 3.4894334085576215e-05, |
| "loss": 0.0414, |
| "step": 88 |
| }, |
| { |
| "epoch": 1.78, |
| "grad_norm": 0.006642194562792848, |
| "learning_rate": 3.397257151002068e-05, |
| "loss": 0.0397, |
| "step": 89 |
| }, |
| { |
| "epoch": 1.8, |
| "grad_norm": 0.00781252853806199, |
| "learning_rate": 3.305407289332279e-05, |
| "loss": 0.0391, |
| "step": 90 |
| }, |
| { |
| "epoch": 1.8199999999999998, |
| "grad_norm": 0.006340215025020239, |
| "learning_rate": 3.213933561868679e-05, |
| "loss": 0.0415, |
| "step": 91 |
| }, |
| { |
| "epoch": 1.8399999999999999, |
| "grad_norm": 0.0071317608069426395, |
| "learning_rate": 3.122885503248386e-05, |
| "loss": 0.0404, |
| "step": 92 |
| }, |
| { |
| "epoch": 1.8599999999999999, |
| "grad_norm": 0.007420667599117525, |
| "learning_rate": 3.0323124176013297e-05, |
| "loss": 0.0427, |
| "step": 93 |
| }, |
| { |
| "epoch": 1.88, |
| "grad_norm": 0.006451786602759296, |
| "learning_rate": 2.9422633518511926e-05, |
| "loss": 0.0397, |
| "step": 94 |
| }, |
| { |
| "epoch": 1.9, |
| "grad_norm": 0.006161926335129036, |
| "learning_rate": 2.8527870691556404e-05, |
| "loss": 0.0397, |
| "step": 95 |
| }, |
| { |
| "epoch": 1.92, |
| "grad_norm": 0.006307087356260699, |
| "learning_rate": 2.7639320225002108e-05, |
| "loss": 0.0404, |
| "step": 96 |
| }, |
| { |
| "epoch": 1.94, |
| "grad_norm": 0.0068390532674039465, |
| "learning_rate": 2.6757463284601682e-05, |
| "loss": 0.0398, |
| "step": 97 |
| }, |
| { |
| "epoch": 1.96, |
| "grad_norm": 0.006399946995758622, |
| "learning_rate": 2.5882777411445254e-05, |
| "loss": 0.0394, |
| "step": 98 |
| }, |
| { |
| "epoch": 1.98, |
| "grad_norm": 0.006614547136094999, |
| "learning_rate": 2.501573626336352e-05, |
| "loss": 0.0408, |
| "step": 99 |
| }, |
| { |
| "epoch": 2.0, |
| "grad_norm": 0.0065020014974483085, |
| "learning_rate": 2.4156809358433728e-05, |
| "loss": 0.0401, |
| "step": 100 |
| }, |
| { |
| "epoch": 2.02, |
| "grad_norm": 0.006058599455897061, |
| "learning_rate": 2.330646182072729e-05, |
| "loss": 0.0377, |
| "step": 101 |
| }, |
| { |
| "epoch": 2.04, |
| "grad_norm": 0.006208162243890031, |
| "learning_rate": 2.24651541284369e-05, |
| "loss": 0.0401, |
| "step": 102 |
| }, |
| { |
| "epoch": 2.06, |
| "grad_norm": 0.008426734711046678, |
| "learning_rate": 2.1633341864519526e-05, |
| "loss": 0.038, |
| "step": 103 |
| }, |
| { |
| "epoch": 2.08, |
| "grad_norm": 0.006098567809952458, |
| "learning_rate": 2.0811475469990167e-05, |
| "loss": 0.0386, |
| "step": 104 |
| }, |
| { |
| "epoch": 2.1, |
| "grad_norm": 0.0059914787934257, |
| "learning_rate": 2.0000000000000012e-05, |
| "loss": 0.0383, |
| "step": 105 |
| }, |
| { |
| "epoch": 2.12, |
| "grad_norm": 0.006085187265121628, |
| "learning_rate": 1.9199354882831387e-05, |
| "loss": 0.0413, |
| "step": 106 |
| }, |
| { |
| "epoch": 2.14, |
| "grad_norm": 0.006150971173906847, |
| "learning_rate": 1.8409973681939498e-05, |
| "loss": 0.0397, |
| "step": 107 |
| }, |
| { |
| "epoch": 2.16, |
| "grad_norm": 0.006130643029450737, |
| "learning_rate": 1.7632283861170135e-05, |
| "loss": 0.0381, |
| "step": 108 |
| }, |
| { |
| "epoch": 2.18, |
| "grad_norm": 0.006584786677044574, |
| "learning_rate": 1.686670655328054e-05, |
| "loss": 0.04, |
| "step": 109 |
| }, |
| { |
| "epoch": 2.2, |
| "grad_norm": 0.005962359743132183, |
| "learning_rate": 1.6113656331888563e-05, |
| "loss": 0.038, |
| "step": 110 |
| }, |
| { |
| "epoch": 2.22, |
| "grad_norm": 0.0068903701990081214, |
| "learning_rate": 1.537354098697367e-05, |
| "loss": 0.0388, |
| "step": 111 |
| }, |
| { |
| "epoch": 2.24, |
| "grad_norm": 0.006496322271615654, |
| "learning_rate": 1.4646761304051587e-05, |
| "loss": 0.0363, |
| "step": 112 |
| }, |
| { |
| "epoch": 2.26, |
| "grad_norm": 0.006178955905560712, |
| "learning_rate": 1.3933710847141795e-05, |
| "loss": 0.0393, |
| "step": 113 |
| }, |
| { |
| "epoch": 2.2800000000000002, |
| "grad_norm": 0.006163335120142197, |
| "learning_rate": 1.3234775745645684e-05, |
| "loss": 0.0385, |
| "step": 114 |
| }, |
| { |
| "epoch": 2.3, |
| "grad_norm": 0.00662720459008428, |
| "learning_rate": 1.2550334485250661e-05, |
| "loss": 0.0358, |
| "step": 115 |
| }, |
| { |
| "epoch": 2.32, |
| "grad_norm": 0.006425841666268113, |
| "learning_rate": 1.1880757702973531e-05, |
| "loss": 0.0384, |
| "step": 116 |
| }, |
| { |
| "epoch": 2.34, |
| "grad_norm": 0.006656256363181535, |
| "learning_rate": 1.1226407986453963e-05, |
| "loss": 0.0396, |
| "step": 117 |
| }, |
| { |
| "epoch": 2.36, |
| "grad_norm": 0.006199253939773614, |
| "learning_rate": 1.0587639677606857e-05, |
| "loss": 0.0366, |
| "step": 118 |
| }, |
| { |
| "epoch": 2.38, |
| "grad_norm": 0.006579884364152935, |
| "learning_rate": 9.964798680740033e-06, |
| "loss": 0.0408, |
| "step": 119 |
| }, |
| { |
| "epoch": 2.4, |
| "grad_norm": 0.006726423214027467, |
| "learning_rate": 9.358222275240884e-06, |
| "loss": 0.0396, |
| "step": 120 |
| }, |
| { |
| "epoch": 2.42, |
| "grad_norm": 0.006603359310147371, |
| "learning_rate": 8.768238932933632e-06, |
| "loss": 0.0385, |
| "step": 121 |
| }, |
| { |
| "epoch": 2.44, |
| "grad_norm": 0.006377855347682361, |
| "learning_rate": 8.195168140206084e-06, |
| "loss": 0.037, |
| "step": 122 |
| }, |
| { |
| "epoch": 2.46, |
| "grad_norm": 0.006382748830046442, |
| "learning_rate": 7.639320225002106e-06, |
| "loss": 0.0399, |
| "step": 123 |
| }, |
| { |
| "epoch": 2.48, |
| "grad_norm": 0.006293817904237255, |
| "learning_rate": 7.1009961887735075e-06, |
| "loss": 0.0379, |
| "step": 124 |
| }, |
| { |
| "epoch": 2.5, |
| "grad_norm": 0.0062574577969606885, |
| "learning_rate": 6.58048754348255e-06, |
| "loss": 0.0383, |
| "step": 125 |
| }, |
| { |
| "epoch": 2.52, |
| "grad_norm": 0.006281219686477559, |
| "learning_rate": 6.078076153742962e-06, |
| "loss": 0.0382, |
| "step": 126 |
| }, |
| { |
| "epoch": 2.54, |
| "grad_norm": 0.006162925544148044, |
| "learning_rate": 5.5940340841853915e-06, |
| "loss": 0.0382, |
| "step": 127 |
| }, |
| { |
| "epoch": 2.56, |
| "grad_norm": 0.006353507662667686, |
| "learning_rate": 5.128623452129508e-06, |
| "loss": 0.04, |
| "step": 128 |
| }, |
| { |
| "epoch": 2.58, |
| "grad_norm": 0.007192196263511861, |
| "learning_rate": 4.6820962856429205e-06, |
| "loss": 0.0396, |
| "step": 129 |
| }, |
| { |
| "epoch": 2.6, |
| "grad_norm": 0.006545353556526849, |
| "learning_rate": 4.254694387063514e-06, |
| "loss": 0.0377, |
| "step": 130 |
| }, |
| { |
| "epoch": 2.62, |
| "grad_norm": 0.006278438500415658, |
| "learning_rate": 3.846649202059181e-06, |
| "loss": 0.0387, |
| "step": 131 |
| }, |
| { |
| "epoch": 2.64, |
| "grad_norm": 0.006785630463381635, |
| "learning_rate": 3.458181694295961e-06, |
| "loss": 0.0385, |
| "step": 132 |
| }, |
| { |
| "epoch": 2.66, |
| "grad_norm": 0.006019048036617288, |
| "learning_rate": 3.0895022257823083e-06, |
| "loss": 0.0372, |
| "step": 133 |
| }, |
| { |
| "epoch": 2.68, |
| "grad_norm": 0.006292600245502946, |
| "learning_rate": 2.7408104429543025e-06, |
| "loss": 0.0366, |
| "step": 134 |
| }, |
| { |
| "epoch": 2.7, |
| "grad_norm": 0.0068058681051229, |
| "learning_rate": 2.4122951685636674e-06, |
| "loss": 0.0401, |
| "step": 135 |
| }, |
| { |
| "epoch": 2.7199999999999998, |
| "grad_norm": 0.006704436014402007, |
| "learning_rate": 2.104134299426832e-06, |
| "loss": 0.0393, |
| "step": 136 |
| }, |
| { |
| "epoch": 2.74, |
| "grad_norm": 0.0068475746559251965, |
| "learning_rate": 1.8164947100907238e-06, |
| "loss": 0.0397, |
| "step": 137 |
| }, |
| { |
| "epoch": 2.76, |
| "grad_norm": 0.006598968805029075, |
| "learning_rate": 1.5495321624672443e-06, |
| "loss": 0.0376, |
| "step": 138 |
| }, |
| { |
| "epoch": 2.7800000000000002, |
| "grad_norm": 0.00618210034494269, |
| "learning_rate": 1.3033912214854482e-06, |
| "loss": 0.0394, |
| "step": 139 |
| }, |
| { |
| "epoch": 2.8, |
| "grad_norm": 0.006358151817184969, |
| "learning_rate": 1.0782051768070477e-06, |
| "loss": 0.0394, |
| "step": 140 |
| }, |
| { |
| "epoch": 2.82, |
| "grad_norm": 0.006493263460417818, |
| "learning_rate": 8.740959706477725e-07, |
| "loss": 0.0394, |
| "step": 141 |
| }, |
| { |
| "epoch": 2.84, |
| "grad_norm": 0.006345978006140842, |
| "learning_rate": 6.911741317434706e-07, |
| "loss": 0.0393, |
| "step": 142 |
| }, |
| { |
| "epoch": 2.86, |
| "grad_norm": 0.006567195230458572, |
| "learning_rate": 5.295387154968312e-07, |
| "loss": 0.0383, |
| "step": 143 |
| }, |
| { |
| "epoch": 2.88, |
| "grad_norm": 0.006436382305239059, |
| "learning_rate": 3.8927725033718553e-07, |
| "loss": 0.0386, |
| "step": 144 |
| }, |
| { |
| "epoch": 2.9, |
| "grad_norm": 0.0070820634086422, |
| "learning_rate": 2.704656903222791e-07, |
| "loss": 0.0403, |
| "step": 145 |
| }, |
| { |
| "epoch": 2.92, |
| "grad_norm": 0.006889363917979449, |
| "learning_rate": 1.7316837400782604e-07, |
| "loss": 0.041, |
| "step": 146 |
| }, |
| { |
| "epoch": 2.94, |
| "grad_norm": 0.005953146361827321, |
| "learning_rate": 9.74379896070321e-08, |
| "loss": 0.0373, |
| "step": 147 |
| }, |
| { |
| "epoch": 2.96, |
| "grad_norm": 0.006429647637374188, |
| "learning_rate": 4.331554645901737e-08, |
| "loss": 0.038, |
| "step": 148 |
| }, |
| { |
| "epoch": 2.98, |
| "grad_norm": 0.0065416859799214844, |
| "learning_rate": 1.0830352821531442e-08, |
| "loss": 0.0401, |
| "step": 149 |
| }, |
| { |
| "epoch": 3.0, |
| "grad_norm": 0.006133326618093241, |
| "learning_rate": 0.0, |
| "loss": 0.0386, |
| "step": 150 |
| }, |
| { |
| "epoch": 3.0, |
| "step": 150, |
| "total_flos": 112494170603520.0, |
| "train_loss": 0.06295452485481898, |
| "train_runtime": 1232.2386, |
| "train_samples_per_second": 0.974, |
| "train_steps_per_second": 0.122 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 150, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": false, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 112494170603520.0, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
