| { | |
| "best_metric": 0.8745942711830139, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-150", | |
| "epoch": 2.6143790849673203, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.013071895424836602, | |
| "grad_norm": 2.132277488708496, | |
| "learning_rate": 1e-05, | |
| "loss": 2.5243, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.013071895424836602, | |
| "eval_loss": 3.2979736328125, | |
| "eval_runtime": 5.4494, | |
| "eval_samples_per_second": 23.672, | |
| "eval_steps_per_second": 6.056, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.026143790849673203, | |
| "grad_norm": 2.422544479370117, | |
| "learning_rate": 2e-05, | |
| "loss": 2.7257, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0392156862745098, | |
| "grad_norm": 2.7282228469848633, | |
| "learning_rate": 3e-05, | |
| "loss": 2.9835, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.05228758169934641, | |
| "grad_norm": 2.7331345081329346, | |
| "learning_rate": 4e-05, | |
| "loss": 2.8142, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.06535947712418301, | |
| "grad_norm": 2.851044178009033, | |
| "learning_rate": 5e-05, | |
| "loss": 2.793, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0784313725490196, | |
| "grad_norm": 2.955557107925415, | |
| "learning_rate": 6e-05, | |
| "loss": 2.5726, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0915032679738562, | |
| "grad_norm": 2.9412848949432373, | |
| "learning_rate": 7e-05, | |
| "loss": 2.3244, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.10457516339869281, | |
| "grad_norm": 2.4464271068573, | |
| "learning_rate": 8e-05, | |
| "loss": 1.9284, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.11764705882352941, | |
| "grad_norm": 2.2438085079193115, | |
| "learning_rate": 9e-05, | |
| "loss": 1.9129, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.13071895424836602, | |
| "grad_norm": 2.632805347442627, | |
| "learning_rate": 0.0001, | |
| "loss": 1.5819, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.1437908496732026, | |
| "grad_norm": 2.4615933895111084, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.3983, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.1568627450980392, | |
| "grad_norm": 2.2855052947998047, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.2679, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.16993464052287582, | |
| "grad_norm": 1.9483047723770142, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.1113, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.1830065359477124, | |
| "grad_norm": 1.898349404335022, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.0255, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.19607843137254902, | |
| "grad_norm": 1.7717444896697998, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.7884, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.20915032679738563, | |
| "grad_norm": 1.4485807418823242, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.7521, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.2222222222222222, | |
| "grad_norm": 1.6411559581756592, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.8263, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.23529411764705882, | |
| "grad_norm": 1.5858994722366333, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.842, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.24836601307189543, | |
| "grad_norm": 2.436020612716675, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.8108, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.26143790849673204, | |
| "grad_norm": 2.779942035675049, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.7328, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.27450980392156865, | |
| "grad_norm": 2.2072298526763916, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.6658, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.2875816993464052, | |
| "grad_norm": 1.6562902927398682, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.4948, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.3006535947712418, | |
| "grad_norm": 1.3172097206115723, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.44, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.3137254901960784, | |
| "grad_norm": 1.4236687421798706, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.3705, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.32679738562091504, | |
| "grad_norm": 1.2323154211044312, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.1964, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.33986928104575165, | |
| "grad_norm": 1.1312896013259888, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.2, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.35294117647058826, | |
| "grad_norm": 1.179223895072937, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.3134, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.3660130718954248, | |
| "grad_norm": 1.1927425861358643, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.1121, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.3790849673202614, | |
| "grad_norm": 1.1144378185272217, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.0084, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.39215686274509803, | |
| "grad_norm": 1.1265126466751099, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.9776, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.40522875816993464, | |
| "grad_norm": 1.2972180843353271, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.8398, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.41830065359477125, | |
| "grad_norm": 1.1511213779449463, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.9028, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.43137254901960786, | |
| "grad_norm": 1.4088971614837646, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.8215, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.4444444444444444, | |
| "grad_norm": 1.2136602401733398, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.7638, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.45751633986928103, | |
| "grad_norm": 1.492048978805542, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.7516, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.47058823529411764, | |
| "grad_norm": 1.7785533666610718, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.6431, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.48366013071895425, | |
| "grad_norm": 1.3489888906478882, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.5841, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.49673202614379086, | |
| "grad_norm": 1.3530248403549194, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.6004, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.5098039215686274, | |
| "grad_norm": 2.185941696166992, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 1.5148, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.5228758169934641, | |
| "grad_norm": 1.8576021194458008, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 1.3991, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.5359477124183006, | |
| "grad_norm": 1.5531032085418701, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 1.3601, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.5490196078431373, | |
| "grad_norm": 1.2389577627182007, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 1.2707, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.5620915032679739, | |
| "grad_norm": 1.0573772192001343, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.3461, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.5751633986928104, | |
| "grad_norm": 1.0560119152069092, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 1.2343, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.5882352941176471, | |
| "grad_norm": 1.1088478565216064, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 1.2519, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.6013071895424836, | |
| "grad_norm": 1.0878572463989258, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 1.1998, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.6143790849673203, | |
| "grad_norm": 1.063195824623108, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.0202, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.6274509803921569, | |
| "grad_norm": 1.0787616968154907, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.9743, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.6405228758169934, | |
| "grad_norm": 1.0731960535049438, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.9882, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.6535947712418301, | |
| "grad_norm": 0.968326210975647, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.8914, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.6535947712418301, | |
| "eval_loss": 0.9767971634864807, | |
| "eval_runtime": 5.52, | |
| "eval_samples_per_second": 23.37, | |
| "eval_steps_per_second": 5.978, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 1.006164312362671, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.0184, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.6797385620915033, | |
| "grad_norm": 1.0974806547164917, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.8728, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.6928104575163399, | |
| "grad_norm": 0.8787504434585571, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.8526, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.7058823529411765, | |
| "grad_norm": 1.078235387802124, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.6676, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.7189542483660131, | |
| "grad_norm": 0.9935086965560913, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.7242, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.7320261437908496, | |
| "grad_norm": 0.9902070760726929, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.6749, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.7450980392156863, | |
| "grad_norm": 1.241457462310791, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.581, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.7581699346405228, | |
| "grad_norm": 1.074371337890625, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.3114, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.7712418300653595, | |
| "grad_norm": 1.2651773691177368, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.2693, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.7843137254901961, | |
| "grad_norm": 1.2351809740066528, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.3778, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.7973856209150327, | |
| "grad_norm": 1.084375262260437, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.2305, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.8104575163398693, | |
| "grad_norm": 1.0903037786483765, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.3489, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.8235294117647058, | |
| "grad_norm": 0.9965918064117432, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.2081, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.8366013071895425, | |
| "grad_norm": 0.9407563209533691, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.0894, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.8496732026143791, | |
| "grad_norm": 0.9775397181510925, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.1545, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.8627450980392157, | |
| "grad_norm": 1.056516408920288, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.0848, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.8758169934640523, | |
| "grad_norm": 0.9769263863563538, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.0375, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.8888888888888888, | |
| "grad_norm": 1.0836280584335327, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.0754, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.9019607843137255, | |
| "grad_norm": 0.9505405426025391, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.8716, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.9150326797385621, | |
| "grad_norm": 0.8555734753608704, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.7938, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.9281045751633987, | |
| "grad_norm": 0.9190294146537781, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.6422, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.9411764705882353, | |
| "grad_norm": 0.9099729657173157, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.7491, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.954248366013072, | |
| "grad_norm": 0.8842048645019531, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.6227, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.9673202614379085, | |
| "grad_norm": 0.9345638155937195, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.6593, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.9803921568627451, | |
| "grad_norm": 1.0071007013320923, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.706, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.9934640522875817, | |
| "grad_norm": 0.9857985973358154, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.595, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.0065359477124183, | |
| "grad_norm": 1.5075626373291016, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.6987, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.0196078431372548, | |
| "grad_norm": 0.7958778142929077, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.143, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.0326797385620916, | |
| "grad_norm": 0.9403725266456604, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.2158, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.0457516339869282, | |
| "grad_norm": 0.8732569813728333, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.0675, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.0588235294117647, | |
| "grad_norm": 0.8857873678207397, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.0507, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.0718954248366013, | |
| "grad_norm": 0.9464306831359863, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.1929, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.0849673202614378, | |
| "grad_norm": 0.902678906917572, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.9533, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.0980392156862746, | |
| "grad_norm": 0.8275942802429199, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.8214, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.1111111111111112, | |
| "grad_norm": 0.9533475041389465, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.9551, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.1241830065359477, | |
| "grad_norm": 1.024800181388855, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.925, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.1372549019607843, | |
| "grad_norm": 0.9481696486473083, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.7826, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.1503267973856208, | |
| "grad_norm": 0.9391154646873474, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.7744, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.1633986928104576, | |
| "grad_norm": 0.9326615333557129, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.6893, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.1764705882352942, | |
| "grad_norm": 1.0131608247756958, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.653, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.1895424836601307, | |
| "grad_norm": 1.0302122831344604, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.5878, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.2026143790849673, | |
| "grad_norm": 0.9927499890327454, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.5668, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.215686274509804, | |
| "grad_norm": 0.7958215475082397, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.3341, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.2287581699346406, | |
| "grad_norm": 0.9926536679267883, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.5082, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.2418300653594772, | |
| "grad_norm": 1.0847405195236206, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.541, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.2549019607843137, | |
| "grad_norm": 1.1613997220993042, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.8567, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.2679738562091503, | |
| "grad_norm": 1.0054422616958618, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.0562, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.2810457516339868, | |
| "grad_norm": 1.1442482471466064, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 1.1478, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.2941176470588236, | |
| "grad_norm": 0.9902859926223755, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.8904, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.3071895424836601, | |
| "grad_norm": 1.0396746397018433, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.9889, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.3071895424836601, | |
| "eval_loss": 0.9468737840652466, | |
| "eval_runtime": 5.5099, | |
| "eval_samples_per_second": 23.413, | |
| "eval_steps_per_second": 5.989, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.3202614379084967, | |
| "grad_norm": 1.001630425453186, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.9051, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 0.9949597120285034, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.8841, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.34640522875817, | |
| "grad_norm": 1.2167212963104248, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.9706, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.3594771241830066, | |
| "grad_norm": 1.032914400100708, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.8012, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.3725490196078431, | |
| "grad_norm": 0.9574472308158875, | |
| "learning_rate": 5e-05, | |
| "loss": 0.7671, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.3856209150326797, | |
| "grad_norm": 0.9467063546180725, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.7085, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.3986928104575163, | |
| "grad_norm": 0.9491695165634155, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.6107, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.4117647058823528, | |
| "grad_norm": 1.0215766429901123, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.751, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.4248366013071896, | |
| "grad_norm": 1.01005220413208, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.6102, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.4379084967320261, | |
| "grad_norm": 1.1116946935653687, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.6036, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.4509803921568627, | |
| "grad_norm": 1.002540946006775, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.5764, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.4640522875816995, | |
| "grad_norm": 0.976798415184021, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.4618, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.477124183006536, | |
| "grad_norm": 1.2129331827163696, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.639, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.4901960784313726, | |
| "grad_norm": 1.0513018369674683, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.4686, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.5032679738562091, | |
| "grad_norm": 1.0926587581634521, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.7651, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.5163398692810457, | |
| "grad_norm": 1.1153149604797363, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.0239, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.5294117647058822, | |
| "grad_norm": 1.051416039466858, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.9919, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.5424836601307188, | |
| "grad_norm": 1.070836067199707, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.0555, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.5555555555555556, | |
| "grad_norm": 1.0305131673812866, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.0289, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.5686274509803921, | |
| "grad_norm": 1.0160155296325684, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.9316, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.581699346405229, | |
| "grad_norm": 1.0957988500595093, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.9029, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.5947712418300655, | |
| "grad_norm": 1.0488003492355347, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.9264, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.607843137254902, | |
| "grad_norm": 1.1128772497177124, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.9138, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.6209150326797386, | |
| "grad_norm": 1.0316942930221558, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.7689, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.6339869281045751, | |
| "grad_norm": 1.0211790800094604, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.7791, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.6470588235294117, | |
| "grad_norm": 0.9485324025154114, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.6934, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.6601307189542482, | |
| "grad_norm": 0.8742186427116394, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.6121, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.673202614379085, | |
| "grad_norm": 0.8446812033653259, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.4883, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.6862745098039216, | |
| "grad_norm": 0.8964700102806091, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.5028, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.6993464052287581, | |
| "grad_norm": 0.9149512648582458, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.5843, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.712418300653595, | |
| "grad_norm": 0.9009532928466797, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.4608, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.7254901960784315, | |
| "grad_norm": 1.0163938999176025, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.478, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.738562091503268, | |
| "grad_norm": 1.0101041793823242, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.5673, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.7516339869281046, | |
| "grad_norm": 1.1123989820480347, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.8224, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.7647058823529411, | |
| "grad_norm": 0.9570160508155823, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.9761, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.7777777777777777, | |
| "grad_norm": 0.9643335342407227, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.0833, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.7908496732026142, | |
| "grad_norm": 0.9873813390731812, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.0124, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.803921568627451, | |
| "grad_norm": 1.0618641376495361, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.0116, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.8169934640522876, | |
| "grad_norm": 1.0005154609680176, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.9121, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.8300653594771243, | |
| "grad_norm": 0.9893289804458618, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.9013, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.843137254901961, | |
| "grad_norm": 0.990556001663208, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.8494, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.8562091503267975, | |
| "grad_norm": 1.0399258136749268, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.929, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.869281045751634, | |
| "grad_norm": 0.9311930537223816, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.7265, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.8823529411764706, | |
| "grad_norm": 1.0030659437179565, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.7628, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.8954248366013071, | |
| "grad_norm": 0.9615251421928406, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.6266, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.9084967320261437, | |
| "grad_norm": 1.0245639085769653, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.6527, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.9215686274509802, | |
| "grad_norm": 0.8971703052520752, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.5798, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.934640522875817, | |
| "grad_norm": 1.0201623439788818, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.6534, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.9477124183006536, | |
| "grad_norm": 0.9810142517089844, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.588, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.9607843137254903, | |
| "grad_norm": 1.0093626976013184, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.565, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.9607843137254903, | |
| "eval_loss": 0.8745942711830139, | |
| "eval_runtime": 5.5113, | |
| "eval_samples_per_second": 23.406, | |
| "eval_steps_per_second": 5.988, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.973856209150327, | |
| "grad_norm": 0.929472029209137, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.4916, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.9869281045751634, | |
| "grad_norm": 1.2074962854385376, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.5232, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 1.5716825723648071, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.9962, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 2.0130718954248366, | |
| "grad_norm": 0.8483561873435974, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.8884, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 2.026143790849673, | |
| "grad_norm": 0.8857851624488831, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.8928, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 2.0392156862745097, | |
| "grad_norm": 0.916907548904419, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.8515, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 2.052287581699346, | |
| "grad_norm": 0.9618643522262573, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.8297, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 2.065359477124183, | |
| "grad_norm": 0.9240990281105042, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.768, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 2.0784313725490198, | |
| "grad_norm": 0.9330710768699646, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.7756, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 2.0915032679738563, | |
| "grad_norm": 0.8368220329284668, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.6443, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.104575163398693, | |
| "grad_norm": 0.9179477691650391, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.7124, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 2.1176470588235294, | |
| "grad_norm": 0.8481573462486267, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.6228, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 2.130718954248366, | |
| "grad_norm": 0.9518793821334839, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.6303, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 2.1437908496732025, | |
| "grad_norm": 0.9530749320983887, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.6341, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 2.156862745098039, | |
| "grad_norm": 0.9007125496864319, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.5463, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 2.1699346405228757, | |
| "grad_norm": 0.9253951907157898, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.4965, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 2.183006535947712, | |
| "grad_norm": 0.8218400478363037, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.3962, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 2.196078431372549, | |
| "grad_norm": 0.8689912557601929, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.3911, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 2.2091503267973858, | |
| "grad_norm": 0.9473594427108765, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.4522, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 2.2222222222222223, | |
| "grad_norm": 0.8523597121238708, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.3947, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.235294117647059, | |
| "grad_norm": 0.9270200133323669, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.3792, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 2.2483660130718954, | |
| "grad_norm": 1.0883715152740479, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.3685, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 2.261437908496732, | |
| "grad_norm": 1.0210227966308594, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.9483, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 2.2745098039215685, | |
| "grad_norm": 1.0105195045471191, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.8802, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 2.287581699346405, | |
| "grad_norm": 1.010636329650879, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.8808, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 2.3006535947712417, | |
| "grad_norm": 0.9809849858283997, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.7888, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 2.313725490196078, | |
| "grad_norm": 1.0462572574615479, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.8731, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 2.326797385620915, | |
| "grad_norm": 0.9693484306335449, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.6398, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 2.3398692810457518, | |
| "grad_norm": 1.0878243446350098, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.783, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 2.3529411764705883, | |
| "grad_norm": 1.0398106575012207, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.72, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.366013071895425, | |
| "grad_norm": 0.9807730317115784, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.6037, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 2.3790849673202614, | |
| "grad_norm": 0.9589698314666748, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.5099, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 2.392156862745098, | |
| "grad_norm": 0.9539539217948914, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.5172, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 2.4052287581699345, | |
| "grad_norm": 0.9103758335113525, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.5113, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 2.418300653594771, | |
| "grad_norm": 0.9845137000083923, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.5019, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 2.431372549019608, | |
| "grad_norm": 0.9078724980354309, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.4375, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 2.4444444444444446, | |
| "grad_norm": 0.8656890392303467, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.432, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 2.457516339869281, | |
| "grad_norm": 0.869166910648346, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.3807, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 2.4705882352941178, | |
| "grad_norm": 0.8666725754737854, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.3571, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 2.4836601307189543, | |
| "grad_norm": 0.8940958380699158, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.4027, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.496732026143791, | |
| "grad_norm": 0.9951069951057434, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.4314, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 2.5098039215686274, | |
| "grad_norm": 1.0286827087402344, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.9509, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 2.522875816993464, | |
| "grad_norm": 0.9736284017562866, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.7855, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 2.5359477124183005, | |
| "grad_norm": 1.0003867149353027, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.8858, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 2.549019607843137, | |
| "grad_norm": 1.0242561101913452, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.8086, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 2.5620915032679736, | |
| "grad_norm": 1.0654770135879517, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.8227, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 2.57516339869281, | |
| "grad_norm": 1.0485032796859741, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.7859, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 2.588235294117647, | |
| "grad_norm": 1.0082951784133911, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.7204, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 2.6013071895424837, | |
| "grad_norm": 0.9806328415870667, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.6713, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 2.6143790849673203, | |
| "grad_norm": 1.0059866905212402, | |
| "learning_rate": 0.0, | |
| "loss": 0.6682, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.6143790849673203, | |
| "eval_loss": 0.8808222413063049, | |
| "eval_runtime": 5.5162, | |
| "eval_samples_per_second": 23.386, | |
| "eval_steps_per_second": 5.982, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 1 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.146618963296256e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |