{ "best_metric": 4.33704948425293, "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/full/lstm/3/checkpoints/checkpoint-228957", "epoch": 0.025000278439663435, "eval_steps": 10, "global_step": 228957, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.999998362119627e-05, "loss": 10.8183, "step": 1 }, { "epoch": 0.0, "learning_rate": 4.999161405248948e-05, "loss": 7.5661, "step": 512 }, { "epoch": 0.0, "learning_rate": 4.998322810497896e-05, "loss": 7.0563, "step": 1024 }, { "epoch": 0.0, "learning_rate": 4.997484215746844e-05, "loss": 6.985, "step": 1536 }, { "epoch": 0.0, "learning_rate": 4.996645620995792e-05, "loss": 6.956, "step": 2048 }, { "epoch": 0.0, "learning_rate": 4.99580702624474e-05, "loss": 6.8543, "step": 2560 }, { "epoch": 0.0, "learning_rate": 4.994970069374061e-05, "loss": 6.7253, "step": 3072 }, { "epoch": 0.0, "learning_rate": 4.994131474623009e-05, "loss": 6.6189, "step": 3584 }, { "epoch": 0.0, "learning_rate": 4.993292879871958e-05, "loss": 6.5339, "step": 4096 }, { "epoch": 0.0, "learning_rate": 4.992454285120906e-05, "loss": 6.4422, "step": 4608 }, { "epoch": 0.0, "learning_rate": 4.991615690369854e-05, "loss": 6.3916, "step": 5120 }, { "epoch": 0.0, "learning_rate": 4.990777095618801e-05, "loss": 6.3209, "step": 5632 }, { "epoch": 0.0, "learning_rate": 4.989938500867749e-05, "loss": 6.253, "step": 6144 }, { "epoch": 0.0, "learning_rate": 4.989099906116697e-05, "loss": 6.1866, "step": 6656 }, { "epoch": 0.0, "learning_rate": 4.988261311365645e-05, "loss": 6.1383, "step": 7168 }, { "epoch": 0.0, "learning_rate": 4.987422716614593e-05, "loss": 6.0974, "step": 7680 }, { "epoch": 0.0, "learning_rate": 4.986584121863541e-05, "loss": 6.0514, "step": 8192 }, { "epoch": 0.0, "learning_rate": 4.985745527112489e-05, "loss": 5.9997, "step": 8704 }, { "epoch": 0.0, "learning_rate": 4.984906932361437e-05, "loss": 5.9645, "step": 9216 }, { "epoch": 0.0, "learning_rate": 4.984068337610385e-05, "loss": 5.9268, "step": 9728 }, { "epoch": 0.0, "learning_rate": 4.983229742859333e-05, "loss": 5.8802, "step": 10240 }, { "epoch": 0.0, "learning_rate": 4.982391148108281e-05, "loss": 5.8432, "step": 10752 }, { "epoch": 0.0, "learning_rate": 4.981552553357229e-05, "loss": 5.8191, "step": 11264 }, { "epoch": 0.0, "learning_rate": 4.980713958606178e-05, "loss": 5.7811, "step": 11776 }, { "epoch": 0.0, "learning_rate": 4.9798770017354986e-05, "loss": 5.7597, "step": 12288 }, { "epoch": 0.0, "learning_rate": 4.9790384069844466e-05, "loss": 5.7314, "step": 12800 }, { "epoch": 0.0, "learning_rate": 4.9781998122333946e-05, "loss": 5.7082, "step": 13312 }, { "epoch": 0.0, "learning_rate": 4.9773612174823426e-05, "loss": 5.6786, "step": 13824 }, { "epoch": 0.0, "learning_rate": 4.97652262273129e-05, "loss": 5.6449, "step": 14336 }, { "epoch": 0.0, "learning_rate": 4.9756856658606115e-05, "loss": 5.638, "step": 14848 }, { "epoch": 0.01, "learning_rate": 4.9748470711095595e-05, "loss": 5.6037, "step": 15360 }, { "epoch": 0.01, "learning_rate": 4.974008476358507e-05, "loss": 5.5886, "step": 15872 }, { "epoch": 0.01, "learning_rate": 4.973169881607455e-05, "loss": 5.5626, "step": 16384 }, { "epoch": 0.01, "learning_rate": 4.972331286856403e-05, "loss": 5.5594, "step": 16896 }, { "epoch": 0.01, "learning_rate": 4.9714943299857244e-05, "loss": 5.5281, "step": 17408 }, { "epoch": 0.01, "learning_rate": 4.9706557352346724e-05, "loss": 5.5185, "step": 17920 }, { "epoch": 0.01, "learning_rate": 4.969818778363994e-05, "loss": 5.5051, "step": 18432 }, { "epoch": 0.01, "learning_rate": 4.968980183612942e-05, "loss": 5.4855, "step": 18944 }, { "epoch": 0.01, "learning_rate": 4.96814158886189e-05, "loss": 5.475, "step": 19456 }, { "epoch": 0.01, "learning_rate": 4.967302994110837e-05, "loss": 5.4524, "step": 19968 }, { "epoch": 0.01, "learning_rate": 4.966464399359785e-05, "loss": 5.4292, "step": 20480 }, { "epoch": 0.01, "learning_rate": 4.965625804608733e-05, "loss": 5.4232, "step": 20992 }, { "epoch": 0.01, "learning_rate": 4.964787209857681e-05, "loss": 5.3952, "step": 21504 }, { "epoch": 0.01, "learning_rate": 4.963948615106629e-05, "loss": 5.408, "step": 22016 }, { "epoch": 0.01, "learning_rate": 4.96311165823595e-05, "loss": 5.3798, "step": 22528 }, { "epoch": 0.01, "learning_rate": 4.962273063484898e-05, "loss": 5.3763, "step": 23040 }, { "epoch": 0.01, "learning_rate": 4.961434468733847e-05, "loss": 5.3602, "step": 23552 }, { "epoch": 0.01, "learning_rate": 4.960597511863168e-05, "loss": 5.358, "step": 24064 }, { "epoch": 0.01, "learning_rate": 4.959758917112116e-05, "loss": 5.3509, "step": 24576 }, { "epoch": 0.01, "learning_rate": 4.958920322361064e-05, "loss": 5.3221, "step": 25088 }, { "epoch": 0.01, "learning_rate": 4.958081727610012e-05, "loss": 5.3108, "step": 25600 }, { "epoch": 0.01, "learning_rate": 4.95724313285896e-05, "loss": 5.3022, "step": 26112 }, { "epoch": 0.01, "learning_rate": 4.956404538107908e-05, "loss": 5.3086, "step": 26624 }, { "epoch": 0.01, "learning_rate": 4.955565943356856e-05, "loss": 5.2857, "step": 27136 }, { "epoch": 0.01, "learning_rate": 4.954727348605804e-05, "loss": 5.2924, "step": 27648 }, { "epoch": 0.01, "learning_rate": 4.953890391735125e-05, "loss": 5.263, "step": 28160 }, { "epoch": 0.01, "learning_rate": 4.9530517969840727e-05, "loss": 5.26, "step": 28672 }, { "epoch": 0.01, "learning_rate": 4.9522132022330207e-05, "loss": 5.2459, "step": 29184 }, { "epoch": 0.01, "learning_rate": 4.9513746074819686e-05, "loss": 5.241, "step": 29696 }, { "epoch": 0.01, "learning_rate": 4.9505360127309166e-05, "loss": 5.2379, "step": 30208 }, { "epoch": 0.01, "learning_rate": 4.9496974179798646e-05, "loss": 5.2178, "step": 30720 }, { "epoch": 0.01, "learning_rate": 4.948858823228813e-05, "loss": 5.2112, "step": 31232 }, { "epoch": 0.01, "learning_rate": 4.948021866358134e-05, "loss": 5.2158, "step": 31744 }, { "epoch": 0.01, "learning_rate": 4.947183271607082e-05, "loss": 5.1943, "step": 32256 }, { "epoch": 0.01, "learning_rate": 4.94634467685603e-05, "loss": 5.1926, "step": 32768 }, { "epoch": 0.01, "learning_rate": 4.945506082104978e-05, "loss": 5.1823, "step": 33280 }, { "epoch": 0.01, "learning_rate": 4.9446674873539255e-05, "loss": 5.1801, "step": 33792 }, { "epoch": 0.01, "learning_rate": 4.9438288926028735e-05, "loss": 5.1659, "step": 34304 }, { "epoch": 0.01, "learning_rate": 4.9429902978518215e-05, "loss": 5.1684, "step": 34816 }, { "epoch": 0.01, "learning_rate": 4.942153340981143e-05, "loss": 5.1552, "step": 35328 }, { "epoch": 0.01, "learning_rate": 4.9413147462300904e-05, "loss": 5.146, "step": 35840 }, { "epoch": 0.01, "learning_rate": 4.9404761514790384e-05, "loss": 5.148, "step": 36352 }, { "epoch": 0.01, "learning_rate": 4.939637556727987e-05, "loss": 5.134, "step": 36864 }, { "epoch": 0.01, "learning_rate": 4.938800599857309e-05, "loss": 5.1416, "step": 37376 }, { "epoch": 0.01, "learning_rate": 4.937962005106256e-05, "loss": 5.1388, "step": 37888 }, { "epoch": 0.01, "learning_rate": 4.937123410355204e-05, "loss": 5.129, "step": 38400 }, { "epoch": 0.01, "learning_rate": 4.9362864534845256e-05, "loss": 5.1101, "step": 38912 }, { "epoch": 0.01, "learning_rate": 4.935447858733473e-05, "loss": 5.0947, "step": 39424 }, { "epoch": 0.01, "learning_rate": 4.934609263982421e-05, "loss": 5.0881, "step": 39936 }, { "epoch": 0.01, "learning_rate": 4.9337723071117425e-05, "loss": 5.0927, "step": 40448 }, { "epoch": 0.01, "learning_rate": 4.9329337123606905e-05, "loss": 5.0953, "step": 40960 }, { "epoch": 0.01, "learning_rate": 4.932095117609638e-05, "loss": 5.0905, "step": 41472 }, { "epoch": 0.01, "learning_rate": 4.931256522858586e-05, "loss": 5.0644, "step": 41984 }, { "epoch": 0.01, "learning_rate": 4.930417928107534e-05, "loss": 5.0589, "step": 42496 }, { "epoch": 0.01, "learning_rate": 4.9295793333564825e-05, "loss": 5.0672, "step": 43008 }, { "epoch": 0.01, "learning_rate": 4.9287407386054305e-05, "loss": 5.0535, "step": 43520 }, { "epoch": 0.01, "learning_rate": 4.9279021438543785e-05, "loss": 5.0529, "step": 44032 }, { "epoch": 0.01, "learning_rate": 4.9270651869836994e-05, "loss": 5.0443, "step": 44544 }, { "epoch": 0.01, "learning_rate": 4.9262265922326474e-05, "loss": 5.0457, "step": 45056 }, { "epoch": 0.01, "learning_rate": 4.9253879974815954e-05, "loss": 5.035, "step": 45568 }, { "epoch": 0.02, "learning_rate": 4.9245494027305433e-05, "loss": 5.0266, "step": 46080 }, { "epoch": 0.02, "learning_rate": 4.9237108079794913e-05, "loss": 5.0134, "step": 46592 }, { "epoch": 0.02, "learning_rate": 4.922872213228439e-05, "loss": 5.0257, "step": 47104 }, { "epoch": 0.02, "learning_rate": 4.922033618477387e-05, "loss": 5.0113, "step": 47616 }, { "epoch": 0.02, "learning_rate": 4.921195023726335e-05, "loss": 4.9951, "step": 48128 }, { "epoch": 0.02, "learning_rate": 4.920356428975283e-05, "loss": 4.995, "step": 48640 }, { "epoch": 0.02, "learning_rate": 4.919517834224231e-05, "loss": 5.007, "step": 49152 }, { "epoch": 0.02, "learning_rate": 4.918680877353552e-05, "loss": 4.9967, "step": 49664 }, { "epoch": 0.02, "learning_rate": 4.917842282602501e-05, "loss": 4.9875, "step": 50176 }, { "epoch": 0.02, "learning_rate": 4.917003687851449e-05, "loss": 4.9758, "step": 50688 }, { "epoch": 0.02, "learning_rate": 4.916165093100397e-05, "loss": 4.9797, "step": 51200 }, { "epoch": 0.02, "learning_rate": 4.915328136229718e-05, "loss": 4.9595, "step": 51712 }, { "epoch": 0.02, "learning_rate": 4.914489541478666e-05, "loss": 4.9742, "step": 52224 }, { "epoch": 0.02, "learning_rate": 4.913650946727614e-05, "loss": 4.9756, "step": 52736 }, { "epoch": 0.02, "learning_rate": 4.912812351976562e-05, "loss": 4.9512, "step": 53248 }, { "epoch": 0.02, "learning_rate": 4.911973757225509e-05, "loss": 4.9452, "step": 53760 }, { "epoch": 0.02, "learning_rate": 4.911136800354831e-05, "loss": 4.922, "step": 54272 }, { "epoch": 0.02, "learning_rate": 4.910298205603779e-05, "loss": 4.9337, "step": 54784 }, { "epoch": 0.02, "learning_rate": 4.9094612487330996e-05, "loss": 4.9349, "step": 55296 }, { "epoch": 0.02, "learning_rate": 4.908624291862421e-05, "loss": 4.9289, "step": 55808 }, { "epoch": 0.02, "learning_rate": 4.907785697111369e-05, "loss": 4.9306, "step": 56320 }, { "epoch": 0.02, "learning_rate": 4.906947102360317e-05, "loss": 4.9136, "step": 56832 }, { "epoch": 0.02, "learning_rate": 4.906108507609265e-05, "loss": 4.9153, "step": 57344 }, { "epoch": 0.02, "learning_rate": 4.905269912858213e-05, "loss": 4.917, "step": 57856 }, { "epoch": 0.02, "learning_rate": 4.904431318107161e-05, "loss": 4.9185, "step": 58368 }, { "epoch": 0.02, "learning_rate": 4.903592723356109e-05, "loss": 4.9047, "step": 58880 }, { "epoch": 0.02, "learning_rate": 4.9027541286050565e-05, "loss": 4.8931, "step": 59392 }, { "epoch": 0.02, "learning_rate": 4.9019155338540045e-05, "loss": 4.8959, "step": 59904 }, { "epoch": 0.02, "learning_rate": 4.9010769391029525e-05, "loss": 4.8983, "step": 60416 }, { "epoch": 0.02, "learning_rate": 4.9002383443519005e-05, "loss": 4.8944, "step": 60928 }, { "epoch": 0.02, "learning_rate": 4.8994013874812214e-05, "loss": 4.8881, "step": 61440 }, { "epoch": 0.02, "learning_rate": 4.89856279273017e-05, "loss": 4.8778, "step": 61952 }, { "epoch": 0.02, "learning_rate": 4.897724197979118e-05, "loss": 4.8712, "step": 62464 }, { "epoch": 0.02, "learning_rate": 4.896885603228066e-05, "loss": 4.8642, "step": 62976 }, { "epoch": 0.02, "learning_rate": 4.896047008477014e-05, "loss": 4.8712, "step": 63488 }, { "epoch": 0.02, "learning_rate": 4.895208413725962e-05, "loss": 4.8685, "step": 64000 }, { "epoch": 0.02, "learning_rate": 4.89436981897491e-05, "loss": 4.8707, "step": 64512 }, { "epoch": 0.02, "learning_rate": 4.893531224223858e-05, "loss": 4.8451, "step": 65024 }, { "epoch": 0.02, "learning_rate": 4.892692629472806e-05, "loss": 4.8582, "step": 65536 }, { "epoch": 0.02, "learning_rate": 4.891854034721754e-05, "loss": 4.8587, "step": 66048 }, { "epoch": 0.02, "learning_rate": 4.891017077851075e-05, "loss": 4.8442, "step": 66560 }, { "epoch": 0.02, "learning_rate": 4.890178483100023e-05, "loss": 4.8594, "step": 67072 }, { "epoch": 0.02, "learning_rate": 4.889339888348971e-05, "loss": 4.8397, "step": 67584 }, { "epoch": 0.02, "learning_rate": 4.888502931478292e-05, "loss": 4.8539, "step": 68096 }, { "epoch": 0.02, "learning_rate": 4.88766433672724e-05, "loss": 4.8418, "step": 68608 }, { "epoch": 0.02, "learning_rate": 4.886825741976188e-05, "loss": 4.8217, "step": 69120 }, { "epoch": 0.02, "learning_rate": 4.8859871472251365e-05, "loss": 4.8207, "step": 69632 }, { "epoch": 0.02, "learning_rate": 4.8851485524740845e-05, "loss": 4.8202, "step": 70144 }, { "epoch": 0.02, "learning_rate": 4.8843099577230325e-05, "loss": 4.823, "step": 70656 }, { "epoch": 0.02, "learning_rate": 4.8834713629719805e-05, "loss": 4.8275, "step": 71168 }, { "epoch": 0.02, "learning_rate": 4.8826344061013014e-05, "loss": 4.8246, "step": 71680 }, { "epoch": 0.02, "learning_rate": 4.8817958113502494e-05, "loss": 4.8196, "step": 72192 }, { "epoch": 0.02, "learning_rate": 4.8809572165991974e-05, "loss": 4.8071, "step": 72704 }, { "epoch": 0.02, "learning_rate": 4.8801186218481454e-05, "loss": 4.7964, "step": 73216 }, { "epoch": 0.02, "learning_rate": 4.879280027097093e-05, "loss": 4.7952, "step": 73728 }, { "epoch": 0.02, "learning_rate": 4.878441432346041e-05, "loss": 4.805, "step": 74240 }, { "epoch": 0.02, "learning_rate": 4.877604475475362e-05, "loss": 4.8102, "step": 74752 }, { "epoch": 0.02, "learning_rate": 4.87676588072431e-05, "loss": 4.8008, "step": 75264 }, { "epoch": 0.02, "learning_rate": 4.875927285973258e-05, "loss": 4.7837, "step": 75776 }, { "epoch": 0.02, "learning_rate": 4.875088691222206e-05, "loss": 4.7832, "step": 76288 }, { "epoch": 0.03, "eval_loss": 4.748494625091553, "eval_runtime": 552.6156, "eval_samples_per_second": 690.518, "eval_steps_per_second": 21.579, "step": 76319 }, { "epoch": 0.0, "learning_rate": 4.874250096471154e-05, "loss": 4.7924, "step": 76800 }, { "epoch": 0.0, "learning_rate": 4.873411501720102e-05, "loss": 4.7847, "step": 77312 }, { "epoch": 0.0, "learning_rate": 4.87257290696905e-05, "loss": 4.7762, "step": 77824 }, { "epoch": 0.0, "learning_rate": 4.871734312217998e-05, "loss": 4.7744, "step": 78336 }, { "epoch": 0.0, "learning_rate": 4.870895717466946e-05, "loss": 4.7644, "step": 78848 }, { "epoch": 0.0, "learning_rate": 4.870057122715894e-05, "loss": 4.7505, "step": 79360 }, { "epoch": 0.0, "learning_rate": 4.869218527964842e-05, "loss": 4.7583, "step": 79872 }, { "epoch": 0.0, "learning_rate": 4.86837993321379e-05, "loss": 4.7679, "step": 80384 }, { "epoch": 0.0, "learning_rate": 4.867541338462738e-05, "loss": 4.75, "step": 80896 }, { "epoch": 0.0, "learning_rate": 4.866702743711686e-05, "loss": 4.7578, "step": 81408 }, { "epoch": 0.0, "learning_rate": 4.865864148960634e-05, "loss": 4.7612, "step": 81920 }, { "epoch": 0.0, "learning_rate": 4.8650255542095815e-05, "loss": 4.761, "step": 82432 }, { "epoch": 0.0, "learning_rate": 4.86418695945853e-05, "loss": 4.7497, "step": 82944 }, { "epoch": 0.0, "learning_rate": 4.863348364707478e-05, "loss": 4.7469, "step": 83456 }, { "epoch": 0.0, "learning_rate": 4.862509769956426e-05, "loss": 4.7446, "step": 83968 }, { "epoch": 0.0, "learning_rate": 4.861671175205374e-05, "loss": 4.7335, "step": 84480 }, { "epoch": 0.0, "learning_rate": 4.860832580454322e-05, "loss": 4.7369, "step": 84992 }, { "epoch": 0.0, "learning_rate": 4.85999398570327e-05, "loss": 4.7371, "step": 85504 }, { "epoch": 0.0, "learning_rate": 4.859155390952218e-05, "loss": 4.7361, "step": 86016 }, { "epoch": 0.0, "learning_rate": 4.858320071961912e-05, "loss": 4.7337, "step": 86528 }, { "epoch": 0.0, "learning_rate": 4.85748147721086e-05, "loss": 4.7221, "step": 87040 }, { "epoch": 0.0, "learning_rate": 4.856642882459808e-05, "loss": 4.7341, "step": 87552 }, { "epoch": 0.0, "learning_rate": 4.855804287708756e-05, "loss": 4.7204, "step": 88064 }, { "epoch": 0.0, "learning_rate": 4.854967330838077e-05, "loss": 4.7285, "step": 88576 }, { "epoch": 0.0, "learning_rate": 4.8541287360870256e-05, "loss": 4.7185, "step": 89088 }, { "epoch": 0.0, "learning_rate": 4.8532901413359736e-05, "loss": 4.7171, "step": 89600 }, { "epoch": 0.0, "learning_rate": 4.8524515465849216e-05, "loss": 4.7203, "step": 90112 }, { "epoch": 0.0, "learning_rate": 4.8516145897142425e-05, "loss": 4.7019, "step": 90624 }, { "epoch": 0.0, "learning_rate": 4.850777632843564e-05, "loss": 4.7161, "step": 91136 }, { "epoch": 0.01, "learning_rate": 4.8499390380925114e-05, "loss": 4.6952, "step": 91648 }, { "epoch": 0.01, "learning_rate": 4.8491004433414594e-05, "loss": 4.7013, "step": 92160 }, { "epoch": 0.01, "learning_rate": 4.8482618485904074e-05, "loss": 4.698, "step": 92672 }, { "epoch": 0.01, "learning_rate": 4.8474232538393554e-05, "loss": 4.7058, "step": 93184 }, { "epoch": 0.01, "learning_rate": 4.8465846590883034e-05, "loss": 4.6932, "step": 93696 }, { "epoch": 0.01, "learning_rate": 4.8457460643372514e-05, "loss": 4.6863, "step": 94208 }, { "epoch": 0.01, "learning_rate": 4.8449074695861993e-05, "loss": 4.7029, "step": 94720 }, { "epoch": 0.01, "learning_rate": 4.8440688748351473e-05, "loss": 4.6903, "step": 95232 }, { "epoch": 0.01, "learning_rate": 4.843230280084095e-05, "loss": 4.6971, "step": 95744 }, { "epoch": 0.01, "learning_rate": 4.842393323213417e-05, "loss": 4.6849, "step": 96256 }, { "epoch": 0.01, "learning_rate": 4.841554728462365e-05, "loss": 4.6705, "step": 96768 }, { "epoch": 0.01, "learning_rate": 4.840716133711313e-05, "loss": 4.6734, "step": 97280 }, { "epoch": 0.01, "learning_rate": 4.839877538960261e-05, "loss": 4.6661, "step": 97792 }, { "epoch": 0.01, "learning_rate": 4.839038944209209e-05, "loss": 4.6811, "step": 98304 }, { "epoch": 0.01, "learning_rate": 4.83820198733853e-05, "loss": 4.6713, "step": 98816 }, { "epoch": 0.01, "learning_rate": 4.837363392587478e-05, "loss": 4.6767, "step": 99328 }, { "epoch": 0.01, "learning_rate": 4.836524797836426e-05, "loss": 4.6682, "step": 99840 }, { "epoch": 0.01, "learning_rate": 4.835686203085374e-05, "loss": 4.6741, "step": 100352 }, { "epoch": 0.01, "learning_rate": 4.834847608334322e-05, "loss": 4.6696, "step": 100864 }, { "epoch": 0.01, "learning_rate": 4.83400901358327e-05, "loss": 4.6476, "step": 101376 }, { "epoch": 0.01, "learning_rate": 4.833170418832218e-05, "loss": 4.651, "step": 101888 }, { "epoch": 0.01, "learning_rate": 4.832331824081166e-05, "loss": 4.6494, "step": 102400 }, { "epoch": 0.01, "learning_rate": 4.8314948672104874e-05, "loss": 4.6627, "step": 102912 }, { "epoch": 0.01, "learning_rate": 4.8306562724594354e-05, "loss": 4.6403, "step": 103424 }, { "epoch": 0.01, "learning_rate": 4.829819315588756e-05, "loss": 4.6644, "step": 103936 }, { "epoch": 0.01, "learning_rate": 4.828980720837704e-05, "loss": 4.6381, "step": 104448 }, { "epoch": 0.01, "learning_rate": 4.828142126086652e-05, "loss": 4.6386, "step": 104960 }, { "epoch": 0.01, "learning_rate": 4.8273035313356e-05, "loss": 4.6402, "step": 105472 }, { "epoch": 0.01, "learning_rate": 4.8264649365845476e-05, "loss": 4.6369, "step": 105984 }, { "epoch": 0.01, "learning_rate": 4.8256263418334956e-05, "loss": 4.639, "step": 106496 }, { "epoch": 0.01, "learning_rate": 4.8247877470824436e-05, "loss": 4.6302, "step": 107008 }, { "epoch": 0.01, "learning_rate": 4.8239491523313916e-05, "loss": 4.6264, "step": 107520 }, { "epoch": 0.01, "learning_rate": 4.8231121954607125e-05, "loss": 4.6386, "step": 108032 }, { "epoch": 0.01, "learning_rate": 4.822273600709661e-05, "loss": 4.624, "step": 108544 }, { "epoch": 0.01, "learning_rate": 4.821435005958609e-05, "loss": 4.6222, "step": 109056 }, { "epoch": 0.01, "learning_rate": 4.820596411207557e-05, "loss": 4.6226, "step": 109568 }, { "epoch": 0.01, "learning_rate": 4.819759454336878e-05, "loss": 4.6268, "step": 110080 }, { "epoch": 0.01, "learning_rate": 4.818920859585826e-05, "loss": 4.6113, "step": 110592 }, { "epoch": 0.01, "learning_rate": 4.818082264834774e-05, "loss": 4.6301, "step": 111104 }, { "epoch": 0.01, "learning_rate": 4.817243670083722e-05, "loss": 4.6165, "step": 111616 }, { "epoch": 0.01, "learning_rate": 4.816406713213043e-05, "loss": 4.6144, "step": 112128 }, { "epoch": 0.01, "learning_rate": 4.815568118461991e-05, "loss": 4.6209, "step": 112640 }, { "epoch": 0.01, "learning_rate": 4.814729523710939e-05, "loss": 4.6083, "step": 113152 }, { "epoch": 0.01, "learning_rate": 4.813890928959887e-05, "loss": 4.6265, "step": 113664 }, { "epoch": 0.01, "learning_rate": 4.813053972089208e-05, "loss": 4.6297, "step": 114176 }, { "epoch": 0.01, "learning_rate": 4.8122153773381565e-05, "loss": 4.623, "step": 114688 }, { "epoch": 0.01, "learning_rate": 4.8113767825871045e-05, "loss": 4.604, "step": 115200 }, { "epoch": 0.01, "learning_rate": 4.8105381878360525e-05, "loss": 4.5988, "step": 115712 }, { "epoch": 0.01, "learning_rate": 4.8096995930850005e-05, "loss": 4.5934, "step": 116224 }, { "epoch": 0.01, "learning_rate": 4.8088609983339485e-05, "loss": 4.6041, "step": 116736 }, { "epoch": 0.01, "learning_rate": 4.8080240414632694e-05, "loss": 4.6161, "step": 117248 }, { "epoch": 0.01, "learning_rate": 4.8071854467122174e-05, "loss": 4.6076, "step": 117760 }, { "epoch": 0.01, "learning_rate": 4.8063468519611654e-05, "loss": 4.5948, "step": 118272 }, { "epoch": 0.01, "learning_rate": 4.8055082572101134e-05, "loss": 4.5829, "step": 118784 }, { "epoch": 0.01, "learning_rate": 4.8046696624590614e-05, "loss": 4.6014, "step": 119296 }, { "epoch": 0.01, "learning_rate": 4.8038310677080094e-05, "loss": 4.5865, "step": 119808 }, { "epoch": 0.01, "learning_rate": 4.8029924729569574e-05, "loss": 4.5891, "step": 120320 }, { "epoch": 0.01, "learning_rate": 4.8021538782059054e-05, "loss": 4.5909, "step": 120832 }, { "epoch": 0.01, "learning_rate": 4.801316921335226e-05, "loss": 4.5991, "step": 121344 }, { "epoch": 0.01, "learning_rate": 4.800479964464548e-05, "loss": 4.587, "step": 121856 }, { "epoch": 0.02, "learning_rate": 4.799641369713496e-05, "loss": 4.5844, "step": 122368 }, { "epoch": 0.02, "learning_rate": 4.798802774962444e-05, "loss": 4.5752, "step": 122880 }, { "epoch": 0.02, "learning_rate": 4.797964180211392e-05, "loss": 4.5873, "step": 123392 }, { "epoch": 0.02, "learning_rate": 4.797127223340713e-05, "loss": 4.5863, "step": 123904 }, { "epoch": 0.02, "learning_rate": 4.796288628589661e-05, "loss": 4.5621, "step": 124416 }, { "epoch": 0.02, "learning_rate": 4.795450033838609e-05, "loss": 4.5711, "step": 124928 }, { "epoch": 0.02, "learning_rate": 4.794611439087557e-05, "loss": 4.5864, "step": 125440 }, { "epoch": 0.02, "learning_rate": 4.793772844336505e-05, "loss": 4.5867, "step": 125952 }, { "epoch": 0.02, "learning_rate": 4.792934249585453e-05, "loss": 4.5735, "step": 126464 }, { "epoch": 0.02, "learning_rate": 4.792095654834401e-05, "loss": 4.5675, "step": 126976 }, { "epoch": 0.02, "learning_rate": 4.791257060083349e-05, "loss": 4.5785, "step": 127488 }, { "epoch": 0.02, "learning_rate": 4.7904201032126697e-05, "loss": 4.5568, "step": 128000 }, { "epoch": 0.02, "learning_rate": 4.789581508461618e-05, "loss": 4.5744, "step": 128512 }, { "epoch": 0.02, "learning_rate": 4.788744551590939e-05, "loss": 4.5775, "step": 129024 }, { "epoch": 0.02, "learning_rate": 4.787905956839887e-05, "loss": 4.5649, "step": 129536 }, { "epoch": 0.02, "learning_rate": 4.787067362088835e-05, "loss": 4.5643, "step": 130048 }, { "epoch": 0.02, "learning_rate": 4.786228767337783e-05, "loss": 4.5412, "step": 130560 }, { "epoch": 0.02, "learning_rate": 4.785390172586731e-05, "loss": 4.5464, "step": 131072 }, { "epoch": 0.02, "learning_rate": 4.7845515778356785e-05, "loss": 4.5531, "step": 131584 }, { "epoch": 0.02, "learning_rate": 4.7837129830846265e-05, "loss": 4.5553, "step": 132096 }, { "epoch": 0.02, "learning_rate": 4.782876026213948e-05, "loss": 4.5584, "step": 132608 }, { "epoch": 0.02, "learning_rate": 4.782037431462896e-05, "loss": 4.5463, "step": 133120 }, { "epoch": 0.02, "learning_rate": 4.7811988367118434e-05, "loss": 4.5451, "step": 133632 }, { "epoch": 0.02, "learning_rate": 4.780360241960792e-05, "loss": 4.5519, "step": 134144 }, { "epoch": 0.02, "learning_rate": 4.77952164720974e-05, "loss": 4.558, "step": 134656 }, { "epoch": 0.02, "learning_rate": 4.778683052458688e-05, "loss": 4.546, "step": 135168 }, { "epoch": 0.02, "learning_rate": 4.777844457707636e-05, "loss": 4.5458, "step": 135680 }, { "epoch": 0.02, "learning_rate": 4.777005862956584e-05, "loss": 4.5433, "step": 136192 }, { "epoch": 0.02, "learning_rate": 4.776168906085905e-05, "loss": 4.5465, "step": 136704 }, { "epoch": 0.02, "learning_rate": 4.775330311334853e-05, "loss": 4.547, "step": 137216 }, { "epoch": 0.02, "learning_rate": 4.774491716583801e-05, "loss": 4.5395, "step": 137728 }, { "epoch": 0.02, "learning_rate": 4.773654759713122e-05, "loss": 4.5382, "step": 138240 }, { "epoch": 0.02, "learning_rate": 4.77281616496207e-05, "loss": 4.5322, "step": 138752 }, { "epoch": 0.02, "learning_rate": 4.771977570211018e-05, "loss": 4.5274, "step": 139264 }, { "epoch": 0.02, "learning_rate": 4.771138975459966e-05, "loss": 4.5335, "step": 139776 }, { "epoch": 0.02, "learning_rate": 4.770300380708914e-05, "loss": 4.5363, "step": 140288 }, { "epoch": 0.02, "learning_rate": 4.769461785957862e-05, "loss": 4.5385, "step": 140800 }, { "epoch": 0.02, "learning_rate": 4.7686231912068105e-05, "loss": 4.5224, "step": 141312 }, { "epoch": 0.02, "learning_rate": 4.7677862343361315e-05, "loss": 4.5285, "step": 141824 }, { "epoch": 0.02, "learning_rate": 4.7669476395850795e-05, "loss": 4.5403, "step": 142336 }, { "epoch": 0.02, "learning_rate": 4.7661090448340274e-05, "loss": 4.5251, "step": 142848 }, { "epoch": 0.02, "learning_rate": 4.7652704500829754e-05, "loss": 4.5384, "step": 143360 }, { "epoch": 0.02, "learning_rate": 4.7644318553319234e-05, "loss": 4.523, "step": 143872 }, { "epoch": 0.02, "learning_rate": 4.7635932605808714e-05, "loss": 4.541, "step": 144384 }, { "epoch": 0.02, "learning_rate": 4.7627546658298194e-05, "loss": 4.5275, "step": 144896 }, { "epoch": 0.02, "learning_rate": 4.7619160710787674e-05, "loss": 4.5116, "step": 145408 }, { "epoch": 0.02, "learning_rate": 4.761079114208088e-05, "loss": 4.5152, "step": 145920 }, { "epoch": 0.02, "learning_rate": 4.760242157337409e-05, "loss": 4.5125, "step": 146432 }, { "epoch": 0.02, "learning_rate": 4.759403562586357e-05, "loss": 4.5231, "step": 146944 }, { "epoch": 0.02, "learning_rate": 4.758566605715679e-05, "loss": 4.5229, "step": 147456 }, { "epoch": 0.02, "learning_rate": 4.757728010964627e-05, "loss": 4.5252, "step": 147968 }, { "epoch": 0.02, "learning_rate": 4.756889416213575e-05, "loss": 4.52, "step": 148480 }, { "epoch": 0.02, "learning_rate": 4.756050821462523e-05, "loss": 4.5111, "step": 148992 }, { "epoch": 0.02, "learning_rate": 4.755212226711471e-05, "loss": 4.506, "step": 149504 }, { "epoch": 0.02, "learning_rate": 4.754373631960419e-05, "loss": 4.5029, "step": 150016 }, { "epoch": 0.02, "learning_rate": 4.753535037209367e-05, "loss": 4.5134, "step": 150528 }, { "epoch": 0.02, "learning_rate": 4.752696442458315e-05, "loss": 4.5265, "step": 151040 }, { "epoch": 0.02, "learning_rate": 4.751857847707262e-05, "loss": 4.5096, "step": 151552 }, { "epoch": 0.02, "learning_rate": 4.751020890836584e-05, "loss": 4.5027, "step": 152064 }, { "epoch": 0.02, "learning_rate": 4.750182296085532e-05, "loss": 4.5027, "step": 152576 }, { "epoch": 0.03, "eval_loss": 4.47078800201416, "eval_runtime": 551.3077, "eval_samples_per_second": 692.156, "eval_steps_per_second": 21.63, "step": 152638 }, { "epoch": 0.0, "learning_rate": 4.74934370133448e-05, "loss": 4.5138, "step": 153088 }, { "epoch": 0.0, "learning_rate": 4.748505106583428e-05, "loss": 4.5086, "step": 153600 }, { "epoch": 0.0, "learning_rate": 4.747666511832376e-05, "loss": 4.4956, "step": 154112 }, { "epoch": 0.0, "learning_rate": 4.746827917081324e-05, "loss": 4.4975, "step": 154624 }, { "epoch": 0.0, "learning_rate": 4.745989322330272e-05, "loss": 4.4939, "step": 155136 }, { "epoch": 0.0, "learning_rate": 4.74515072757922e-05, "loss": 4.479, "step": 155648 }, { "epoch": 0.0, "learning_rate": 4.744312132828168e-05, "loss": 4.4885, "step": 156160 }, { "epoch": 0.0, "learning_rate": 4.743473538077116e-05, "loss": 4.5019, "step": 156672 }, { "epoch": 0.0, "learning_rate": 4.7426349433260637e-05, "loss": 4.4809, "step": 157184 }, { "epoch": 0.0, "learning_rate": 4.7417963485750117e-05, "loss": 4.4904, "step": 157696 }, { "epoch": 0.0, "learning_rate": 4.7409577538239597e-05, "loss": 4.5005, "step": 158208 }, { "epoch": 0.0, "learning_rate": 4.7401191590729076e-05, "loss": 4.4998, "step": 158720 }, { "epoch": 0.0, "learning_rate": 4.7392805643218556e-05, "loss": 4.4867, "step": 159232 }, { "epoch": 0.0, "learning_rate": 4.7384419695708036e-05, "loss": 4.4868, "step": 159744 }, { "epoch": 0.0, "learning_rate": 4.737603374819751e-05, "loss": 4.4857, "step": 160256 }, { "epoch": 0.0, "learning_rate": 4.736764780068699e-05, "loss": 4.4789, "step": 160768 }, { "epoch": 0.0, "learning_rate": 4.7359261853176476e-05, "loss": 4.4781, "step": 161280 }, { "epoch": 0.0, "learning_rate": 4.7350892284469685e-05, "loss": 4.4845, "step": 161792 }, { "epoch": 0.0, "learning_rate": 4.7342506336959165e-05, "loss": 4.4875, "step": 162304 }, { "epoch": 0.0, "learning_rate": 4.7334120389448645e-05, "loss": 4.4839, "step": 162816 }, { "epoch": 0.0, "learning_rate": 4.7325734441938125e-05, "loss": 4.4755, "step": 163328 }, { "epoch": 0.0, "learning_rate": 4.7317348494427605e-05, "loss": 4.4866, "step": 163840 }, { "epoch": 0.0, "learning_rate": 4.7308978925720814e-05, "loss": 4.4764, "step": 164352 }, { "epoch": 0.0, "learning_rate": 4.7300592978210294e-05, "loss": 4.4821, "step": 164864 }, { "epoch": 0.0, "learning_rate": 4.7292207030699774e-05, "loss": 4.4766, "step": 165376 }, { "epoch": 0.0, "learning_rate": 4.7283821083189254e-05, "loss": 4.4747, "step": 165888 }, { "epoch": 0.0, "learning_rate": 4.7275435135678734e-05, "loss": 4.4804, "step": 166400 }, { "epoch": 0.0, "learning_rate": 4.7267049188168214e-05, "loss": 4.4637, "step": 166912 }, { "epoch": 0.0, "learning_rate": 4.7258663240657694e-05, "loss": 4.4781, "step": 167424 }, { "epoch": 0.01, "learning_rate": 4.7250277293147174e-05, "loss": 4.4627, "step": 167936 }, { "epoch": 0.01, "learning_rate": 4.724190772444039e-05, "loss": 4.4659, "step": 168448 }, { "epoch": 0.01, "learning_rate": 4.72335381557336e-05, "loss": 4.4669, "step": 168960 }, { "epoch": 0.01, "learning_rate": 4.722515220822308e-05, "loss": 4.4697, "step": 169472 }, { "epoch": 0.01, "learning_rate": 4.721676626071256e-05, "loss": 4.4634, "step": 169984 }, { "epoch": 0.01, "learning_rate": 4.720838031320204e-05, "loss": 4.4557, "step": 170496 }, { "epoch": 0.01, "learning_rate": 4.719999436569152e-05, "loss": 4.4744, "step": 171008 }, { "epoch": 0.01, "learning_rate": 4.7191608418181e-05, "loss": 4.4625, "step": 171520 }, { "epoch": 0.01, "learning_rate": 4.718322247067048e-05, "loss": 4.4759, "step": 172032 }, { "epoch": 0.01, "learning_rate": 4.717483652315996e-05, "loss": 4.4619, "step": 172544 }, { "epoch": 0.01, "learning_rate": 4.716646695445317e-05, "loss": 4.4455, "step": 173056 }, { "epoch": 0.01, "learning_rate": 4.715808100694265e-05, "loss": 4.4461, "step": 173568 }, { "epoch": 0.01, "learning_rate": 4.714969505943213e-05, "loss": 4.4495, "step": 174080 }, { "epoch": 0.01, "learning_rate": 4.7141309111921614e-05, "loss": 4.4664, "step": 174592 }, { "epoch": 0.01, "learning_rate": 4.7132939543214823e-05, "loss": 4.4457, "step": 175104 }, { "epoch": 0.01, "learning_rate": 4.7124553595704303e-05, "loss": 4.459, "step": 175616 }, { "epoch": 0.01, "learning_rate": 4.711616764819378e-05, "loss": 4.4541, "step": 176128 }, { "epoch": 0.01, "learning_rate": 4.710778170068326e-05, "loss": 4.4534, "step": 176640 }, { "epoch": 0.01, "learning_rate": 4.709941213197647e-05, "loss": 4.4595, "step": 177152 }, { "epoch": 0.01, "learning_rate": 4.709104256326968e-05, "loss": 4.4342, "step": 177664 }, { "epoch": 0.01, "learning_rate": 4.708265661575916e-05, "loss": 4.4396, "step": 178176 }, { "epoch": 0.01, "learning_rate": 4.707427066824864e-05, "loss": 4.4334, "step": 178688 }, { "epoch": 0.01, "learning_rate": 4.706588472073812e-05, "loss": 4.4548, "step": 179200 }, { "epoch": 0.01, "learning_rate": 4.70574987732276e-05, "loss": 4.431, "step": 179712 }, { "epoch": 0.01, "learning_rate": 4.704911282571708e-05, "loss": 4.4573, "step": 180224 }, { "epoch": 0.01, "learning_rate": 4.704072687820657e-05, "loss": 4.4324, "step": 180736 }, { "epoch": 0.01, "learning_rate": 4.703234093069605e-05, "loss": 4.4326, "step": 181248 }, { "epoch": 0.01, "learning_rate": 4.702397136198926e-05, "loss": 4.4335, "step": 181760 }, { "epoch": 0.01, "learning_rate": 4.701558541447874e-05, "loss": 4.4301, "step": 182272 }, { "epoch": 0.01, "learning_rate": 4.700719946696822e-05, "loss": 4.4377, "step": 182784 }, { "epoch": 0.01, "learning_rate": 4.69988135194577e-05, "loss": 4.4316, "step": 183296 }, { "epoch": 0.01, "learning_rate": 4.6990443950750906e-05, "loss": 4.4208, "step": 183808 }, { "epoch": 0.01, "learning_rate": 4.6982058003240386e-05, "loss": 4.4367, "step": 184320 }, { "epoch": 0.01, "learning_rate": 4.6973672055729866e-05, "loss": 4.4247, "step": 184832 }, { "epoch": 0.01, "learning_rate": 4.696528610821934e-05, "loss": 4.4277, "step": 185344 }, { "epoch": 0.01, "learning_rate": 4.6956916539512555e-05, "loss": 4.4245, "step": 185856 }, { "epoch": 0.01, "learning_rate": 4.6948530592002035e-05, "loss": 4.432, "step": 186368 }, { "epoch": 0.01, "learning_rate": 4.6940144644491515e-05, "loss": 4.411, "step": 186880 }, { "epoch": 0.01, "learning_rate": 4.6931758696980995e-05, "loss": 4.437, "step": 187392 }, { "epoch": 0.01, "learning_rate": 4.6923372749470475e-05, "loss": 4.4218, "step": 187904 }, { "epoch": 0.01, "learning_rate": 4.6914986801959955e-05, "loss": 4.4249, "step": 188416 }, { "epoch": 0.01, "learning_rate": 4.6906600854449435e-05, "loss": 4.4279, "step": 188928 }, { "epoch": 0.01, "learning_rate": 4.6898231285742644e-05, "loss": 4.4174, "step": 189440 }, { "epoch": 0.01, "learning_rate": 4.6889845338232124e-05, "loss": 4.4357, "step": 189952 }, { "epoch": 0.01, "learning_rate": 4.6881459390721604e-05, "loss": 4.438, "step": 190464 }, { "epoch": 0.01, "learning_rate": 4.6873073443211084e-05, "loss": 4.4371, "step": 190976 }, { "epoch": 0.01, "learning_rate": 4.6864687495700564e-05, "loss": 4.4216, "step": 191488 }, { "epoch": 0.01, "learning_rate": 4.6856301548190044e-05, "loss": 4.4115, "step": 192000 }, { "epoch": 0.01, "learning_rate": 4.684793197948325e-05, "loss": 4.4046, "step": 192512 }, { "epoch": 0.01, "learning_rate": 4.683954603197274e-05, "loss": 4.4228, "step": 193024 }, { "epoch": 0.01, "learning_rate": 4.683116008446222e-05, "loss": 4.43, "step": 193536 }, { "epoch": 0.01, "learning_rate": 4.68227741369517e-05, "loss": 4.4261, "step": 194048 }, { "epoch": 0.01, "learning_rate": 4.681438818944118e-05, "loss": 4.4129, "step": 194560 }, { "epoch": 0.01, "learning_rate": 4.680600224193066e-05, "loss": 4.4032, "step": 195072 }, { "epoch": 0.01, "learning_rate": 4.679761629442014e-05, "loss": 4.4162, "step": 195584 }, { "epoch": 0.01, "learning_rate": 4.678923034690962e-05, "loss": 4.4031, "step": 196096 }, { "epoch": 0.01, "learning_rate": 4.678086077820283e-05, "loss": 4.4083, "step": 196608 }, { "epoch": 0.01, "learning_rate": 4.677247483069231e-05, "loss": 4.4168, "step": 197120 }, { "epoch": 0.01, "learning_rate": 4.676408888318179e-05, "loss": 4.4165, "step": 197632 }, { "epoch": 0.01, "learning_rate": 4.675570293567127e-05, "loss": 4.4094, "step": 198144 }, { "epoch": 0.02, "learning_rate": 4.674731698816075e-05, "loss": 4.4124, "step": 198656 }, { "epoch": 0.02, "learning_rate": 4.673893104065023e-05, "loss": 4.3982, "step": 199168 }, { "epoch": 0.02, "learning_rate": 4.673054509313971e-05, "loss": 4.4121, "step": 199680 }, { "epoch": 0.02, "learning_rate": 4.672215914562919e-05, "loss": 4.4082, "step": 200192 }, { "epoch": 0.02, "learning_rate": 4.671380595572613e-05, "loss": 4.3928, "step": 200704 }, { "epoch": 0.02, "learning_rate": 4.670542000821561e-05, "loss": 4.3989, "step": 201216 }, { "epoch": 0.02, "learning_rate": 4.669703406070509e-05, "loss": 4.4133, "step": 201728 }, { "epoch": 0.02, "learning_rate": 4.668864811319457e-05, "loss": 4.4144, "step": 202240 }, { "epoch": 0.02, "learning_rate": 4.668026216568405e-05, "loss": 4.4001, "step": 202752 }, { "epoch": 0.02, "learning_rate": 4.667187621817353e-05, "loss": 4.4019, "step": 203264 }, { "epoch": 0.02, "learning_rate": 4.6663490270663006e-05, "loss": 4.4075, "step": 203776 }, { "epoch": 0.02, "learning_rate": 4.6655104323152486e-05, "loss": 4.3892, "step": 204288 }, { "epoch": 0.02, "learning_rate": 4.66467347544457e-05, "loss": 4.4063, "step": 204800 }, { "epoch": 0.02, "learning_rate": 4.6638348806935175e-05, "loss": 4.4113, "step": 205312 }, { "epoch": 0.02, "learning_rate": 4.662996285942466e-05, "loss": 4.3958, "step": 205824 }, { "epoch": 0.02, "learning_rate": 4.662157691191414e-05, "loss": 4.4024, "step": 206336 }, { "epoch": 0.02, "learning_rate": 4.661320734320736e-05, "loss": 4.3777, "step": 206848 }, { "epoch": 0.02, "learning_rate": 4.660482139569683e-05, "loss": 4.3787, "step": 207360 }, { "epoch": 0.02, "learning_rate": 4.659643544818631e-05, "loss": 4.3912, "step": 207872 }, { "epoch": 0.02, "learning_rate": 4.658804950067579e-05, "loss": 4.3924, "step": 208384 }, { "epoch": 0.02, "learning_rate": 4.6579679931969006e-05, "loss": 4.3987, "step": 208896 }, { "epoch": 0.02, "learning_rate": 4.657129398445848e-05, "loss": 4.3834, "step": 209408 }, { "epoch": 0.02, "learning_rate": 4.6562924415751695e-05, "loss": 4.3815, "step": 209920 }, { "epoch": 0.02, "learning_rate": 4.6554538468241175e-05, "loss": 4.3877, "step": 210432 }, { "epoch": 0.02, "learning_rate": 4.654615252073065e-05, "loss": 4.3984, "step": 210944 }, { "epoch": 0.02, "learning_rate": 4.653776657322013e-05, "loss": 4.3905, "step": 211456 }, { "epoch": 0.02, "learning_rate": 4.6529380625709615e-05, "loss": 4.3934, "step": 211968 }, { "epoch": 0.02, "learning_rate": 4.6520994678199095e-05, "loss": 4.3828, "step": 212480 }, { "epoch": 0.02, "learning_rate": 4.6512608730688575e-05, "loss": 4.384, "step": 212992 }, { "epoch": 0.02, "learning_rate": 4.6504222783178055e-05, "loss": 4.3954, "step": 213504 }, { "epoch": 0.02, "learning_rate": 4.6495853214471264e-05, "loss": 4.3812, "step": 214016 }, { "epoch": 0.02, "learning_rate": 4.6487467266960744e-05, "loss": 4.382, "step": 214528 }, { "epoch": 0.02, "learning_rate": 4.6479081319450224e-05, "loss": 4.3794, "step": 215040 }, { "epoch": 0.02, "learning_rate": 4.647071175074343e-05, "loss": 4.3716, "step": 215552 }, { "epoch": 0.02, "learning_rate": 4.646232580323291e-05, "loss": 4.3803, "step": 216064 }, { "epoch": 0.02, "learning_rate": 4.645393985572239e-05, "loss": 4.38, "step": 216576 }, { "epoch": 0.02, "learning_rate": 4.644555390821187e-05, "loss": 4.3893, "step": 217088 }, { "epoch": 0.02, "learning_rate": 4.643718433950508e-05, "loss": 4.37, "step": 217600 }, { "epoch": 0.02, "learning_rate": 4.642879839199457e-05, "loss": 4.3743, "step": 218112 }, { "epoch": 0.02, "learning_rate": 4.642041244448405e-05, "loss": 4.3919, "step": 218624 }, { "epoch": 0.02, "learning_rate": 4.641202649697353e-05, "loss": 4.374, "step": 219136 }, { "epoch": 0.02, "learning_rate": 4.640365692826674e-05, "loss": 4.3887, "step": 219648 }, { "epoch": 0.02, "learning_rate": 4.639527098075622e-05, "loss": 4.3764, "step": 220160 }, { "epoch": 0.02, "learning_rate": 4.63868850332457e-05, "loss": 4.3921, "step": 220672 }, { "epoch": 0.02, "learning_rate": 4.637849908573518e-05, "loss": 4.3781, "step": 221184 }, { "epoch": 0.02, "learning_rate": 4.637011313822466e-05, "loss": 4.3683, "step": 221696 }, { "epoch": 0.02, "learning_rate": 4.636174356951787e-05, "loss": 4.3674, "step": 222208 }, { "epoch": 0.02, "learning_rate": 4.635335762200735e-05, "loss": 4.3647, "step": 222720 }, { "epoch": 0.02, "learning_rate": 4.634497167449683e-05, "loss": 4.376, "step": 223232 }, { "epoch": 0.02, "learning_rate": 4.633658572698631e-05, "loss": 4.3844, "step": 223744 }, { "epoch": 0.02, "learning_rate": 4.632819977947579e-05, "loss": 4.3763, "step": 224256 }, { "epoch": 0.02, "learning_rate": 4.631981383196527e-05, "loss": 4.3735, "step": 224768 }, { "epoch": 0.02, "learning_rate": 4.631144426325848e-05, "loss": 4.3668, "step": 225280 }, { "epoch": 0.02, "learning_rate": 4.630305831574796e-05, "loss": 4.3644, "step": 225792 }, { "epoch": 0.02, "learning_rate": 4.629467236823744e-05, "loss": 4.3598, "step": 226304 }, { "epoch": 0.02, "learning_rate": 4.628628642072692e-05, "loss": 4.3698, "step": 226816 }, { "epoch": 0.02, "learning_rate": 4.62779004732164e-05, "loss": 4.3816, "step": 227328 }, { "epoch": 0.02, "learning_rate": 4.626951452570588e-05, "loss": 4.3672, "step": 227840 }, { "epoch": 0.02, "learning_rate": 4.626114495699909e-05, "loss": 4.3635, "step": 228352 }, { "epoch": 0.02, "learning_rate": 4.625275900948857e-05, "loss": 4.3608, "step": 228864 }, { "epoch": 0.03, "eval_loss": 4.33704948425293, "eval_runtime": 541.6671, "eval_samples_per_second": 704.475, "eval_steps_per_second": 22.015, "step": 228957 } ], "logging_steps": 512, "max_steps": 3052726, "num_train_epochs": 9223372036854775807, "save_steps": 10, "total_flos": 9.449267462809522e+16, "trial_name": null, "trial_params": null }