| { | |
| "best_metric": 1.726232647895813, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 1.1611030478955007, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.005805515239477504, | |
| "grad_norm": 1.2802343368530273, | |
| "learning_rate": 1e-05, | |
| "loss": 2.1148, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.005805515239477504, | |
| "eval_loss": 2.0874240398406982, | |
| "eval_runtime": 24.197, | |
| "eval_samples_per_second": 11.985, | |
| "eval_steps_per_second": 3.017, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.011611030478955007, | |
| "grad_norm": 1.4416449069976807, | |
| "learning_rate": 2e-05, | |
| "loss": 2.0287, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01741654571843251, | |
| "grad_norm": 1.375106692314148, | |
| "learning_rate": 3e-05, | |
| "loss": 2.0014, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.023222060957910014, | |
| "grad_norm": 1.1383010149002075, | |
| "learning_rate": 4e-05, | |
| "loss": 2.0238, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02902757619738752, | |
| "grad_norm": 0.9523646235466003, | |
| "learning_rate": 5e-05, | |
| "loss": 2.0582, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03483309143686502, | |
| "grad_norm": 0.573900043964386, | |
| "learning_rate": 6e-05, | |
| "loss": 1.9721, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.040638606676342524, | |
| "grad_norm": 0.5881373286247253, | |
| "learning_rate": 7e-05, | |
| "loss": 2.0104, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.04644412191582003, | |
| "grad_norm": 0.7135363221168518, | |
| "learning_rate": 8e-05, | |
| "loss": 2.0332, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.05224963715529753, | |
| "grad_norm": 0.8216105699539185, | |
| "learning_rate": 9e-05, | |
| "loss": 1.9973, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.05805515239477504, | |
| "grad_norm": 0.766675591468811, | |
| "learning_rate": 0.0001, | |
| "loss": 1.7773, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.06386066763425254, | |
| "grad_norm": 0.5826238393783569, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.9543, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.06966618287373004, | |
| "grad_norm": 0.5086792707443237, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.9092, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.07547169811320754, | |
| "grad_norm": 0.5587817430496216, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.9346, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.08127721335268505, | |
| "grad_norm": 0.5718170404434204, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.8512, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.08708272859216255, | |
| "grad_norm": 0.5616905093193054, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.957, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.09288824383164006, | |
| "grad_norm": 0.540808379650116, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.775, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.09869375907111756, | |
| "grad_norm": 0.521541178226471, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.9074, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.10449927431059507, | |
| "grad_norm": 0.48680904507637024, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.8549, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.11030478955007257, | |
| "grad_norm": 0.5210248231887817, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.7519, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.11611030478955008, | |
| "grad_norm": 0.4791758954524994, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.9111, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.12191582002902758, | |
| "grad_norm": 0.5570107698440552, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.8326, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.12772133526850507, | |
| "grad_norm": 0.5348185896873474, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.8603, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.13352685050798258, | |
| "grad_norm": 0.5128710865974426, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.7787, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.13933236574746008, | |
| "grad_norm": 0.6001898050308228, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.8758, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.14513788098693758, | |
| "grad_norm": 0.5341706871986389, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.8062, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.1509433962264151, | |
| "grad_norm": 0.5209715366363525, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.8484, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.1567489114658926, | |
| "grad_norm": 0.5500786304473877, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.7907, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.1625544267053701, | |
| "grad_norm": 0.5753782391548157, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.7892, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.1683599419448476, | |
| "grad_norm": 0.5808932781219482, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.8597, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.1741654571843251, | |
| "grad_norm": 0.5610922574996948, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.8381, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.1799709724238026, | |
| "grad_norm": 0.6095215082168579, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.7755, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.18577648766328012, | |
| "grad_norm": 0.6006529927253723, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 1.8907, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.19158200290275762, | |
| "grad_norm": 0.5902464389801025, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.8253, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.19738751814223512, | |
| "grad_norm": 0.5767065286636353, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 1.8035, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.20319303338171263, | |
| "grad_norm": 0.6569252610206604, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.7737, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.20899854862119013, | |
| "grad_norm": 0.6383949518203735, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 1.7617, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.21480406386066764, | |
| "grad_norm": 0.6573511958122253, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.7464, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.22060957910014514, | |
| "grad_norm": 0.9087986350059509, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.61, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.22641509433962265, | |
| "grad_norm": 0.6875400543212891, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 1.8118, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.23222060957910015, | |
| "grad_norm": 0.6848728060722351, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 1.7437, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.23802612481857766, | |
| "grad_norm": 0.7457948923110962, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 1.5457, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.24383164005805516, | |
| "grad_norm": 1.0072450637817383, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 1.4557, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.24963715529753266, | |
| "grad_norm": 0.9280003309249878, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.4485, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.25544267053701014, | |
| "grad_norm": 0.5074917078018188, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 2.0211, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.2612481857764877, | |
| "grad_norm": 0.5715458989143372, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 1.927, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.26705370101596515, | |
| "grad_norm": 0.5026673078536987, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 1.8472, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.2728592162554427, | |
| "grad_norm": 0.47914066910743713, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.7791, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.27866473149492016, | |
| "grad_norm": 0.43209248781204224, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.8447, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.2844702467343977, | |
| "grad_norm": 0.40648382902145386, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 1.8144, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.29027576197387517, | |
| "grad_norm": 0.3830343782901764, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 1.8797, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.29027576197387517, | |
| "eval_loss": 1.7835716009140015, | |
| "eval_runtime": 24.7179, | |
| "eval_samples_per_second": 11.732, | |
| "eval_steps_per_second": 2.953, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2960812772133527, | |
| "grad_norm": 0.40847522020339966, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.8509, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.3018867924528302, | |
| "grad_norm": 0.41111356019973755, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.8034, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.3076923076923077, | |
| "grad_norm": 0.39087432622909546, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 1.8401, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.3134978229317852, | |
| "grad_norm": 0.39595600962638855, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.8924, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.3193033381712627, | |
| "grad_norm": 0.4029504060745239, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.7423, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.3251088534107402, | |
| "grad_norm": 0.3971380591392517, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.8423, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.3309143686502177, | |
| "grad_norm": 0.4342693090438843, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.7788, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.3367198838896952, | |
| "grad_norm": 0.38115909695625305, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.8777, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.34252539912917274, | |
| "grad_norm": 0.526332676410675, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.8461, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.3483309143686502, | |
| "grad_norm": 0.45587947964668274, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.6886, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.35413642960812775, | |
| "grad_norm": 0.40341612696647644, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.8618, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.3599419448476052, | |
| "grad_norm": 0.5133787393569946, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.6871, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.36574746008708275, | |
| "grad_norm": 0.45418137311935425, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.7365, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.37155297532656023, | |
| "grad_norm": 0.4577997028827667, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.7544, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.37735849056603776, | |
| "grad_norm": 0.44454094767570496, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.7263, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.38316400580551524, | |
| "grad_norm": 0.43912273645401, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.7176, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.3889695210449927, | |
| "grad_norm": 0.4508310854434967, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.8291, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.39477503628447025, | |
| "grad_norm": 0.46372470259666443, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.7316, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.4005805515239477, | |
| "grad_norm": 0.4595024585723877, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.7626, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.40638606676342526, | |
| "grad_norm": 0.47752857208251953, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.7949, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.41219158200290273, | |
| "grad_norm": 0.4701814353466034, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.884, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.41799709724238027, | |
| "grad_norm": 0.49096331000328064, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.8549, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.42380261248185774, | |
| "grad_norm": 0.5183019638061523, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.7396, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.4296081277213353, | |
| "grad_norm": 0.6562801599502563, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.8548, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.43541364296081275, | |
| "grad_norm": 0.5333163142204285, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.7933, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.4412191582002903, | |
| "grad_norm": 0.5598500370979309, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.764, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.44702467343976776, | |
| "grad_norm": 0.616254985332489, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.8519, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.4528301886792453, | |
| "grad_norm": 0.5128965973854065, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.6843, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.45863570391872277, | |
| "grad_norm": 0.5723729133605957, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.8211, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.4644412191582003, | |
| "grad_norm": 0.5582553148269653, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.7159, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.4702467343976778, | |
| "grad_norm": 0.6017863750457764, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.8328, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.4760522496371553, | |
| "grad_norm": 0.646784245967865, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.7488, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.4818577648766328, | |
| "grad_norm": 0.621480405330658, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 1.7754, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.4876632801161103, | |
| "grad_norm": 0.7075855731964111, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.6894, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.4934687953555878, | |
| "grad_norm": 0.8735288977622986, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.5269, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.49927431059506533, | |
| "grad_norm": 2.9787540435791016, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.7141, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.5050798258345428, | |
| "grad_norm": 0.8107753992080688, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 2.018, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.5108853410740203, | |
| "grad_norm": 0.42176181077957153, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.9651, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.5166908563134979, | |
| "grad_norm": 0.4151810109615326, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 1.789, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.5224963715529753, | |
| "grad_norm": 0.386679470539093, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 1.8299, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.5283018867924528, | |
| "grad_norm": 0.4106220602989197, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.8651, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.5341074020319303, | |
| "grad_norm": 0.39643245935440063, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.9113, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.5399129172714079, | |
| "grad_norm": 0.3862471580505371, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.8414, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.5457184325108854, | |
| "grad_norm": 0.3962131142616272, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 1.9122, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.5515239477503628, | |
| "grad_norm": 0.4000537395477295, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.723, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.5573294629898403, | |
| "grad_norm": 0.38008323311805725, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.8512, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.5631349782293179, | |
| "grad_norm": 0.38584694266319275, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.8828, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.5689404934687954, | |
| "grad_norm": 0.38245323300361633, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 1.6953, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.5747460087082729, | |
| "grad_norm": 0.4093793034553528, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 1.8376, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.5805515239477503, | |
| "grad_norm": 0.38503092527389526, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 1.8735, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.5805515239477503, | |
| "eval_loss": 1.7677788734436035, | |
| "eval_runtime": 24.699, | |
| "eval_samples_per_second": 11.741, | |
| "eval_steps_per_second": 2.956, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.5863570391872278, | |
| "grad_norm": 0.44562649726867676, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.9916, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.5921625544267054, | |
| "grad_norm": 0.39542460441589355, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.8145, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.5979680696661829, | |
| "grad_norm": 0.3893499970436096, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 1.8439, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.6037735849056604, | |
| "grad_norm": 0.44246092438697815, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 1.7739, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.6095791001451378, | |
| "grad_norm": 0.41057920455932617, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8047, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.6153846153846154, | |
| "grad_norm": 0.44987499713897705, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.9288, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.6211901306240929, | |
| "grad_norm": 0.42091771960258484, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.7894, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.6269956458635704, | |
| "grad_norm": 0.4402920603752136, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.8346, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6328011611030478, | |
| "grad_norm": 0.4101702570915222, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.9121, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.6386066763425254, | |
| "grad_norm": 0.4471023678779602, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.7271, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.6444121915820029, | |
| "grad_norm": 0.4522208273410797, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.7367, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.6502177068214804, | |
| "grad_norm": 0.4687577486038208, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.7813, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.6560232220609579, | |
| "grad_norm": 0.5011056065559387, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.7069, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.6618287373004355, | |
| "grad_norm": 0.48258066177368164, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.7788, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.6676342525399129, | |
| "grad_norm": 0.7686281800270081, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.7729, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.6734397677793904, | |
| "grad_norm": 0.5120784044265747, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.7578, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.6792452830188679, | |
| "grad_norm": 0.4679645895957947, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 1.7349, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.6850507982583455, | |
| "grad_norm": 0.4701582193374634, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.7461, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.690856313497823, | |
| "grad_norm": 0.5181167721748352, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.6972, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.6966618287373004, | |
| "grad_norm": 0.5515949130058289, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.6896, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.7024673439767779, | |
| "grad_norm": 0.5267184972763062, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.6509, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.7082728592162555, | |
| "grad_norm": 0.5740118622779846, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 1.6835, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.714078374455733, | |
| "grad_norm": 0.5620703101158142, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.7505, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.7198838896952104, | |
| "grad_norm": 0.5882965922355652, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.8095, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.7256894049346879, | |
| "grad_norm": 0.6772586107254028, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.7775, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.7314949201741655, | |
| "grad_norm": 0.7016092538833618, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.5983, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.737300435413643, | |
| "grad_norm": 0.6854375004768372, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.7959, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.7431059506531205, | |
| "grad_norm": 0.7054009437561035, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.362, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.7489114658925979, | |
| "grad_norm": 1.2267603874206543, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.7349, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.7547169811320755, | |
| "grad_norm": 0.3392610549926758, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.8782, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.760522496371553, | |
| "grad_norm": 0.3541852533817291, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.8358, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.7663280116110305, | |
| "grad_norm": 0.3445144593715668, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.8241, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.772133526850508, | |
| "grad_norm": 0.36159756779670715, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.8097, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.7779390420899854, | |
| "grad_norm": 0.3605585992336273, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 1.8716, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.783744557329463, | |
| "grad_norm": 0.3616204261779785, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.7814, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.7895500725689405, | |
| "grad_norm": 0.37260672450065613, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.7747, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.795355587808418, | |
| "grad_norm": 0.36629241704940796, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.8003, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.8011611030478955, | |
| "grad_norm": 0.3812524080276489, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.8456, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.806966618287373, | |
| "grad_norm": 0.3807627856731415, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 1.9383, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.8127721335268505, | |
| "grad_norm": 0.3906376361846924, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 1.7923, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.818577648766328, | |
| "grad_norm": 0.382329523563385, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.7776, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.8243831640058055, | |
| "grad_norm": 0.398005872964859, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 1.7775, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.8301886792452831, | |
| "grad_norm": 0.3928670287132263, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 1.6754, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.8359941944847605, | |
| "grad_norm": 0.40780770778656006, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.826, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.841799709724238, | |
| "grad_norm": 0.41190868616104126, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 1.76, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.8476052249637155, | |
| "grad_norm": 0.4169665277004242, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 1.6916, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.8534107402031931, | |
| "grad_norm": 0.410805881023407, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.7968, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.8592162554426706, | |
| "grad_norm": 0.4383578300476074, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 1.8428, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.865021770682148, | |
| "grad_norm": 0.4339222013950348, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 1.8092, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.8708272859216255, | |
| "grad_norm": 0.4114333987236023, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.7835, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.8708272859216255, | |
| "eval_loss": 1.733366847038269, | |
| "eval_runtime": 24.7087, | |
| "eval_samples_per_second": 11.737, | |
| "eval_steps_per_second": 2.954, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.8766328011611031, | |
| "grad_norm": 0.4979633688926697, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.6657, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.8824383164005806, | |
| "grad_norm": 0.43685823678970337, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 1.7748, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.888243831640058, | |
| "grad_norm": 0.4467308223247528, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 1.7467, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.8940493468795355, | |
| "grad_norm": 0.4414559304714203, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 1.7874, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.8998548621190131, | |
| "grad_norm": 0.44535455107688904, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 1.8155, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.9056603773584906, | |
| "grad_norm": 0.4941321909427643, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 1.8024, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.9114658925979681, | |
| "grad_norm": 0.4703381061553955, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 1.6794, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.9172714078374455, | |
| "grad_norm": 0.4796736240386963, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 1.7112, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.9230769230769231, | |
| "grad_norm": 0.5125783681869507, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 1.8101, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.9288824383164006, | |
| "grad_norm": 0.4842379689216614, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 1.6402, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.9346879535558781, | |
| "grad_norm": 0.5073548555374146, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.7246, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.9404934687953556, | |
| "grad_norm": 0.5602807402610779, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.7355, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9462989840348331, | |
| "grad_norm": 0.5214800238609314, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 1.6661, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.9521044992743106, | |
| "grad_norm": 0.5030867457389832, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 1.7319, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.9579100145137881, | |
| "grad_norm": 0.5263693332672119, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 1.6441, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.9637155297532656, | |
| "grad_norm": 0.6761137843132019, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.7177, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.969521044992743, | |
| "grad_norm": 0.5821870565414429, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 1.7148, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.9753265602322206, | |
| "grad_norm": 0.5921502709388733, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.5284, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.9811320754716981, | |
| "grad_norm": 0.6768378615379333, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.7401, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.9869375907111756, | |
| "grad_norm": 0.6789736747741699, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.5939, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.9927431059506531, | |
| "grad_norm": 0.7853531241416931, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 1.6542, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.9985486211901307, | |
| "grad_norm": 1.2762771844863892, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 1.6363, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.004354136429608, | |
| "grad_norm": 1.114426612854004, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 3.1026, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.0101596516690856, | |
| "grad_norm": 0.31035736203193665, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.7538, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.0159651669085632, | |
| "grad_norm": 0.31741762161254883, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 1.6421, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.0217706821480406, | |
| "grad_norm": 0.33394038677215576, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 1.7997, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.0275761973875182, | |
| "grad_norm": 0.35024937987327576, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 1.7062, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.0333817126269957, | |
| "grad_norm": 0.31366899609565735, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 1.767, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.039187227866473, | |
| "grad_norm": 0.33037716150283813, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.7555, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.0449927431059507, | |
| "grad_norm": 0.3353957533836365, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.7856, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.050798258345428, | |
| "grad_norm": 0.3467152416706085, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.6602, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.0566037735849056, | |
| "grad_norm": 0.33460795879364014, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 1.771, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.0624092888243832, | |
| "grad_norm": 0.3509085476398468, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 1.7165, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.0682148040638606, | |
| "grad_norm": 0.3500897288322449, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.742, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.0740203193033382, | |
| "grad_norm": 0.35934382677078247, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.706, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.0798258345428158, | |
| "grad_norm": 0.3761908710002899, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.7417, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.0856313497822931, | |
| "grad_norm": 0.353887677192688, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 1.728, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.0914368650217707, | |
| "grad_norm": 0.3830077052116394, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 1.5503, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.097242380261248, | |
| "grad_norm": 0.3868183195590973, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 1.6244, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.1030478955007257, | |
| "grad_norm": 0.38480788469314575, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 1.6885, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.1088534107402033, | |
| "grad_norm": 0.40707969665527344, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 1.6366, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.1146589259796806, | |
| "grad_norm": 0.40027621388435364, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 1.7272, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.1204644412191582, | |
| "grad_norm": 0.38935405015945435, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 1.6621, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.1262699564586356, | |
| "grad_norm": 0.4096282720565796, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 1.6818, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.1320754716981132, | |
| "grad_norm": 0.40856990218162537, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 1.5515, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.1378809869375908, | |
| "grad_norm": 0.4047314524650574, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.6674, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.1436865021770681, | |
| "grad_norm": 0.415522962808609, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 1.6384, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.1494920174165457, | |
| "grad_norm": 0.4089977443218231, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 1.72, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.1552975326560233, | |
| "grad_norm": 0.4907004237174988, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 1.667, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.1611030478955007, | |
| "grad_norm": 0.4212486445903778, | |
| "learning_rate": 0.0, | |
| "loss": 1.649, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.1611030478955007, | |
| "eval_loss": 1.726232647895813, | |
| "eval_runtime": 24.7335, | |
| "eval_samples_per_second": 11.725, | |
| "eval_steps_per_second": 2.951, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.288221468628091e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |