| { | |
| "best_metric": 0.00864658784121275, | |
| "best_model_checkpoint": "sj-det-5/checkpoint-6104", | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 6104, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0081913499344692, | |
| "grad_norm": 10.024361610412598, | |
| "learning_rate": 1.3100436681222706e-06, | |
| "loss": 0.6701, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0163826998689384, | |
| "grad_norm": 5.330548286437988, | |
| "learning_rate": 2.6746724890829695e-06, | |
| "loss": 0.4647, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0245740498034076, | |
| "grad_norm": 2.015732765197754, | |
| "learning_rate": 4.039301310043668e-06, | |
| "loss": 0.1809, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0327653997378768, | |
| "grad_norm": 6.1893134117126465, | |
| "learning_rate": 5.4039301310043675e-06, | |
| "loss": 0.1308, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.040956749672346006, | |
| "grad_norm": 0.21893443167209625, | |
| "learning_rate": 6.768558951965067e-06, | |
| "loss": 0.0337, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.0491480996068152, | |
| "grad_norm": 5.707635402679443, | |
| "learning_rate": 8.133187772925763e-06, | |
| "loss": 0.0952, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.05733944954128441, | |
| "grad_norm": 0.042420562356710434, | |
| "learning_rate": 9.497816593886463e-06, | |
| "loss": 0.0225, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.0655307994757536, | |
| "grad_norm": 0.07790377736091614, | |
| "learning_rate": 1.0862445414847162e-05, | |
| "loss": 0.0961, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.0737221494102228, | |
| "grad_norm": 0.06831495463848114, | |
| "learning_rate": 1.222707423580786e-05, | |
| "loss": 0.0029, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.08191349934469201, | |
| "grad_norm": 0.3763171434402466, | |
| "learning_rate": 1.3591703056768559e-05, | |
| "loss": 0.0165, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.09010484927916121, | |
| "grad_norm": 0.042719755321741104, | |
| "learning_rate": 1.4956331877729258e-05, | |
| "loss": 0.0537, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.0982961992136304, | |
| "grad_norm": 1.7841322422027588, | |
| "learning_rate": 1.632096069868996e-05, | |
| "loss": 0.0304, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.1064875491480996, | |
| "grad_norm": 0.006917485501617193, | |
| "learning_rate": 1.7685589519650656e-05, | |
| "loss": 0.0061, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.11467889908256881, | |
| "grad_norm": 0.007079846691340208, | |
| "learning_rate": 1.9050218340611354e-05, | |
| "loss": 0.0234, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.12287024901703801, | |
| "grad_norm": 0.013362312689423561, | |
| "learning_rate": 2.0414847161572055e-05, | |
| "loss": 0.0214, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.1310615989515072, | |
| "grad_norm": 0.011783940717577934, | |
| "learning_rate": 2.1779475982532752e-05, | |
| "loss": 0.0543, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.13925294888597642, | |
| "grad_norm": 0.020665360614657402, | |
| "learning_rate": 2.3144104803493453e-05, | |
| "loss": 0.0316, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.1474442988204456, | |
| "grad_norm": 1.4373878240585327, | |
| "learning_rate": 2.450873362445415e-05, | |
| "loss": 0.06, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1556356487549148, | |
| "grad_norm": 3.8112599849700928, | |
| "learning_rate": 2.587336244541485e-05, | |
| "loss": 0.0182, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.16382699868938402, | |
| "grad_norm": 5.189245223999023, | |
| "learning_rate": 2.7237991266375546e-05, | |
| "loss": 0.154, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1720183486238532, | |
| "grad_norm": 0.015202702023088932, | |
| "learning_rate": 2.8602620087336247e-05, | |
| "loss": 0.0294, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.18020969855832242, | |
| "grad_norm": 0.015019269660115242, | |
| "learning_rate": 2.9967248908296948e-05, | |
| "loss": 0.0591, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.1884010484927916, | |
| "grad_norm": 0.0036466640885919333, | |
| "learning_rate": 3.133187772925764e-05, | |
| "loss": 0.0034, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.1965923984272608, | |
| "grad_norm": 0.011716458015143871, | |
| "learning_rate": 3.269650655021834e-05, | |
| "loss": 0.0015, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.20478374836173002, | |
| "grad_norm": 0.0010442484635859728, | |
| "learning_rate": 3.406113537117904e-05, | |
| "loss": 0.0002, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.2129750982961992, | |
| "grad_norm": 0.030393408611416817, | |
| "learning_rate": 3.542576419213974e-05, | |
| "loss": 0.0555, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.22116644823066842, | |
| "grad_norm": 0.004823782481253147, | |
| "learning_rate": 3.6790393013100435e-05, | |
| "loss": 0.0011, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.22935779816513763, | |
| "grad_norm": 0.02440587244927883, | |
| "learning_rate": 3.815502183406114e-05, | |
| "loss": 0.0504, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.2375491480996068, | |
| "grad_norm": 0.05391886085271835, | |
| "learning_rate": 3.951965065502183e-05, | |
| "loss": 0.0821, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.24574049803407602, | |
| "grad_norm": 0.006875240709632635, | |
| "learning_rate": 4.0884279475982535e-05, | |
| "loss": 0.0008, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.25393184796854523, | |
| "grad_norm": 0.0016325929900631309, | |
| "learning_rate": 4.224890829694323e-05, | |
| "loss": 0.0183, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.2621231979030144, | |
| "grad_norm": 0.029884610325098038, | |
| "learning_rate": 4.361353711790393e-05, | |
| "loss": 0.0525, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.2703145478374836, | |
| "grad_norm": 0.05142253637313843, | |
| "learning_rate": 4.497816593886463e-05, | |
| "loss": 0.0499, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.27850589777195284, | |
| "grad_norm": 0.0004886944079771638, | |
| "learning_rate": 4.634279475982533e-05, | |
| "loss": 0.0401, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.286697247706422, | |
| "grad_norm": 0.003223366802558303, | |
| "learning_rate": 4.770742358078603e-05, | |
| "loss": 0.0893, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.2948885976408912, | |
| "grad_norm": 0.02335241623222828, | |
| "learning_rate": 4.907205240174673e-05, | |
| "loss": 0.0003, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.30307994757536044, | |
| "grad_norm": 0.036217641085386276, | |
| "learning_rate": 4.9951456310679614e-05, | |
| "loss": 0.0387, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.3112712975098296, | |
| "grad_norm": 0.0005936770467087626, | |
| "learning_rate": 4.97997572815534e-05, | |
| "loss": 0.0231, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.3194626474442988, | |
| "grad_norm": 0.11047300696372986, | |
| "learning_rate": 4.964805825242719e-05, | |
| "loss": 0.0235, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.32765399737876805, | |
| "grad_norm": 0.0038526670541614294, | |
| "learning_rate": 4.9496359223300976e-05, | |
| "loss": 0.0804, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.33584534731323723, | |
| "grad_norm": 0.002965217223390937, | |
| "learning_rate": 4.934466019417476e-05, | |
| "loss": 0.0019, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.3440366972477064, | |
| "grad_norm": 1.2460094690322876, | |
| "learning_rate": 4.9192961165048545e-05, | |
| "loss": 0.0218, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.3522280471821756, | |
| "grad_norm": 0.00023055930796545, | |
| "learning_rate": 4.904126213592233e-05, | |
| "loss": 0.0824, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.36041939711664484, | |
| "grad_norm": 0.0008325474918819964, | |
| "learning_rate": 4.888956310679612e-05, | |
| "loss": 0.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.368610747051114, | |
| "grad_norm": 0.12137041985988617, | |
| "learning_rate": 4.87378640776699e-05, | |
| "loss": 0.0429, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.3768020969855832, | |
| "grad_norm": 0.12224938720464706, | |
| "learning_rate": 4.858616504854369e-05, | |
| "loss": 0.0287, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.38499344692005244, | |
| "grad_norm": 0.00017641298472881317, | |
| "learning_rate": 4.843446601941748e-05, | |
| "loss": 0.0937, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.3931847968545216, | |
| "grad_norm": 0.00043992913560941815, | |
| "learning_rate": 4.828276699029127e-05, | |
| "loss": 0.0001, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.4013761467889908, | |
| "grad_norm": 0.0067256479524075985, | |
| "learning_rate": 4.813106796116505e-05, | |
| "loss": 0.0403, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.40956749672346004, | |
| "grad_norm": 0.008025743998587132, | |
| "learning_rate": 4.797936893203884e-05, | |
| "loss": 0.1236, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.41775884665792923, | |
| "grad_norm": 0.0024881409481167793, | |
| "learning_rate": 4.7827669902912624e-05, | |
| "loss": 0.0065, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.4259501965923984, | |
| "grad_norm": 0.00034138892078772187, | |
| "learning_rate": 4.767597087378641e-05, | |
| "loss": 0.0007, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.43414154652686765, | |
| "grad_norm": 0.09658686816692352, | |
| "learning_rate": 4.7524271844660193e-05, | |
| "loss": 0.0557, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.44233289646133683, | |
| "grad_norm": 0.004737064708024263, | |
| "learning_rate": 4.7372572815533985e-05, | |
| "loss": 0.0658, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.450524246395806, | |
| "grad_norm": 0.014197874814271927, | |
| "learning_rate": 4.722087378640777e-05, | |
| "loss": 0.0263, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.45871559633027525, | |
| "grad_norm": 0.012036195956170559, | |
| "learning_rate": 4.7069174757281555e-05, | |
| "loss": 0.0672, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.46690694626474444, | |
| "grad_norm": 0.006545116659253836, | |
| "learning_rate": 4.691747572815534e-05, | |
| "loss": 0.0136, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.4750982961992136, | |
| "grad_norm": 0.003104938892647624, | |
| "learning_rate": 4.676577669902913e-05, | |
| "loss": 0.0412, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.4832896461336828, | |
| "grad_norm": 6.474489688873291, | |
| "learning_rate": 4.661407766990292e-05, | |
| "loss": 0.0564, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.49148099606815204, | |
| "grad_norm": 0.11798975616693497, | |
| "learning_rate": 4.64623786407767e-05, | |
| "loss": 0.0441, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.4996723460026212, | |
| "grad_norm": 0.004167527426034212, | |
| "learning_rate": 4.631067961165049e-05, | |
| "loss": 0.0049, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.5078636959370905, | |
| "grad_norm": 0.023102520033717155, | |
| "learning_rate": 4.615898058252427e-05, | |
| "loss": 0.0679, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.5160550458715596, | |
| "grad_norm": 0.12365124374628067, | |
| "learning_rate": 4.6007281553398063e-05, | |
| "loss": 0.0429, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.5242463958060288, | |
| "grad_norm": 0.02387346886098385, | |
| "learning_rate": 4.585558252427185e-05, | |
| "loss": 0.0706, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.5324377457404981, | |
| "grad_norm": 0.2509637773036957, | |
| "learning_rate": 4.570388349514563e-05, | |
| "loss": 0.0108, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.5406290956749672, | |
| "grad_norm": 0.006082233507186174, | |
| "learning_rate": 4.555218446601942e-05, | |
| "loss": 0.1004, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.5488204456094364, | |
| "grad_norm": 0.002660457044839859, | |
| "learning_rate": 4.540048543689321e-05, | |
| "loss": 0.0015, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.5570117955439057, | |
| "grad_norm": 0.029006479308009148, | |
| "learning_rate": 4.524878640776699e-05, | |
| "loss": 0.0698, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.5652031454783748, | |
| "grad_norm": 2.156792163848877, | |
| "learning_rate": 4.509708737864078e-05, | |
| "loss": 0.0054, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.573394495412844, | |
| "grad_norm": 0.004138014279305935, | |
| "learning_rate": 4.4945388349514565e-05, | |
| "loss": 0.0527, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.5815858453473133, | |
| "grad_norm": 3.908644199371338, | |
| "learning_rate": 4.479368932038835e-05, | |
| "loss": 0.0543, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.5897771952817824, | |
| "grad_norm": 0.4576684236526489, | |
| "learning_rate": 4.4641990291262135e-05, | |
| "loss": 0.0386, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.5979685452162516, | |
| "grad_norm": 0.00034501738264225423, | |
| "learning_rate": 4.4490291262135927e-05, | |
| "loss": 0.0479, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.6061598951507209, | |
| "grad_norm": 0.020500104874372482, | |
| "learning_rate": 4.433859223300971e-05, | |
| "loss": 0.0522, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.61435124508519, | |
| "grad_norm": 0.07448019832372665, | |
| "learning_rate": 4.4186893203883496e-05, | |
| "loss": 0.0044, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.6225425950196593, | |
| "grad_norm": 0.010664916597306728, | |
| "learning_rate": 4.403519417475728e-05, | |
| "loss": 0.0059, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.6307339449541285, | |
| "grad_norm": 0.0046129715628921986, | |
| "learning_rate": 4.388349514563107e-05, | |
| "loss": 0.0575, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.6389252948885976, | |
| "grad_norm": 0.0065656425431370735, | |
| "learning_rate": 4.373179611650486e-05, | |
| "loss": 0.0051, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.6471166448230669, | |
| "grad_norm": 0.045100320130586624, | |
| "learning_rate": 4.358009708737864e-05, | |
| "loss": 0.0765, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.6553079947575361, | |
| "grad_norm": 0.010744914412498474, | |
| "learning_rate": 4.342839805825243e-05, | |
| "loss": 0.0434, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.6634993446920052, | |
| "grad_norm": 0.006751019041985273, | |
| "learning_rate": 4.327669902912622e-05, | |
| "loss": 0.0112, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.6716906946264745, | |
| "grad_norm": 0.0010609191376715899, | |
| "learning_rate": 4.3125000000000005e-05, | |
| "loss": 0.0002, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.6798820445609436, | |
| "grad_norm": 0.0029373185243457556, | |
| "learning_rate": 4.297330097087379e-05, | |
| "loss": 0.0457, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.6880733944954128, | |
| "grad_norm": 0.03929394483566284, | |
| "learning_rate": 4.2821601941747575e-05, | |
| "loss": 0.0003, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.6962647444298821, | |
| "grad_norm": 0.0013411706313490868, | |
| "learning_rate": 4.266990291262136e-05, | |
| "loss": 0.0003, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.7044560943643512, | |
| "grad_norm": 0.15903156995773315, | |
| "learning_rate": 4.251820388349515e-05, | |
| "loss": 0.0713, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.7126474442988204, | |
| "grad_norm": 0.004102173261344433, | |
| "learning_rate": 4.236650485436893e-05, | |
| "loss": 0.0071, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.7208387942332897, | |
| "grad_norm": 0.0019201630493625998, | |
| "learning_rate": 4.221480582524272e-05, | |
| "loss": 0.0012, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.7290301441677588, | |
| "grad_norm": 0.08992388099431992, | |
| "learning_rate": 4.2063106796116506e-05, | |
| "loss": 0.0682, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.737221494102228, | |
| "grad_norm": 0.06924425065517426, | |
| "learning_rate": 4.19114077669903e-05, | |
| "loss": 0.0045, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.7454128440366973, | |
| "grad_norm": 0.006293242797255516, | |
| "learning_rate": 4.1759708737864076e-05, | |
| "loss": 0.0328, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.7536041939711664, | |
| "grad_norm": 0.0008105287561193109, | |
| "learning_rate": 4.160800970873787e-05, | |
| "loss": 0.0481, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.7617955439056356, | |
| "grad_norm": 0.01918267086148262, | |
| "learning_rate": 4.145631067961165e-05, | |
| "loss": 0.0529, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.7699868938401049, | |
| "grad_norm": 0.02567248046398163, | |
| "learning_rate": 4.130461165048544e-05, | |
| "loss": 0.0271, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.778178243774574, | |
| "grad_norm": 0.00797118991613388, | |
| "learning_rate": 4.115291262135922e-05, | |
| "loss": 0.0116, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.7863695937090432, | |
| "grad_norm": 0.034628257155418396, | |
| "learning_rate": 4.1001213592233014e-05, | |
| "loss": 0.0389, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.7945609436435125, | |
| "grad_norm": 0.004856760613620281, | |
| "learning_rate": 4.08495145631068e-05, | |
| "loss": 0.0076, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.8027522935779816, | |
| "grad_norm": 0.0018171003321185708, | |
| "learning_rate": 4.0697815533980584e-05, | |
| "loss": 0.0001, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.8109436435124509, | |
| "grad_norm": 0.002607023809105158, | |
| "learning_rate": 4.054611650485437e-05, | |
| "loss": 0.0529, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.8191349934469201, | |
| "grad_norm": 0.00494106812402606, | |
| "learning_rate": 4.039441747572816e-05, | |
| "loss": 0.0025, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.8273263433813892, | |
| "grad_norm": 7.6454033851623535, | |
| "learning_rate": 4.0242718446601946e-05, | |
| "loss": 0.0506, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.8355176933158585, | |
| "grad_norm": 0.012740246020257473, | |
| "learning_rate": 4.009101941747573e-05, | |
| "loss": 0.0207, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.8437090432503277, | |
| "grad_norm": 0.19343392550945282, | |
| "learning_rate": 3.9939320388349516e-05, | |
| "loss": 0.0018, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.8519003931847968, | |
| "grad_norm": 0.0015372405759990215, | |
| "learning_rate": 3.97876213592233e-05, | |
| "loss": 0.0475, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.8600917431192661, | |
| "grad_norm": 0.001495638396590948, | |
| "learning_rate": 3.963592233009709e-05, | |
| "loss": 0.0204, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.8682830930537353, | |
| "grad_norm": 0.024810276925563812, | |
| "learning_rate": 3.948422330097087e-05, | |
| "loss": 0.0377, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.8764744429882044, | |
| "grad_norm": 0.001583193545229733, | |
| "learning_rate": 3.933252427184466e-05, | |
| "loss": 0.0392, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.8846657929226737, | |
| "grad_norm": 0.04295048862695694, | |
| "learning_rate": 3.918082524271845e-05, | |
| "loss": 0.0349, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.8928571428571429, | |
| "grad_norm": 0.006643366534262896, | |
| "learning_rate": 3.902912621359224e-05, | |
| "loss": 0.0214, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.901048492791612, | |
| "grad_norm": 0.03927929326891899, | |
| "learning_rate": 3.887742718446602e-05, | |
| "loss": 0.0313, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.9092398427260813, | |
| "grad_norm": 0.0005574201932176948, | |
| "learning_rate": 3.872572815533981e-05, | |
| "loss": 0.0167, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.9174311926605505, | |
| "grad_norm": 0.0002800333604682237, | |
| "learning_rate": 3.8574029126213594e-05, | |
| "loss": 0.0038, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.9256225425950196, | |
| "grad_norm": 0.24904371798038483, | |
| "learning_rate": 3.842233009708738e-05, | |
| "loss": 0.0557, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.9338138925294889, | |
| "grad_norm": 0.009384050033986568, | |
| "learning_rate": 3.8270631067961164e-05, | |
| "loss": 0.0349, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.9420052424639581, | |
| "grad_norm": 0.0032450663857162, | |
| "learning_rate": 3.8118932038834956e-05, | |
| "loss": 0.0061, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.9501965923984272, | |
| "grad_norm": 0.002643973333761096, | |
| "learning_rate": 3.796723300970874e-05, | |
| "loss": 0.0011, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.9583879423328965, | |
| "grad_norm": 0.005450396798551083, | |
| "learning_rate": 3.7815533980582526e-05, | |
| "loss": 0.0472, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.9665792922673656, | |
| "grad_norm": 0.009381350129842758, | |
| "learning_rate": 3.766383495145631e-05, | |
| "loss": 0.0221, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.9747706422018348, | |
| "grad_norm": 0.002026522997766733, | |
| "learning_rate": 3.75121359223301e-05, | |
| "loss": 0.0015, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.9829619921363041, | |
| "grad_norm": 0.0021547062788158655, | |
| "learning_rate": 3.736043689320389e-05, | |
| "loss": 0.0346, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.9911533420707732, | |
| "grad_norm": 0.01624121516942978, | |
| "learning_rate": 3.720873786407767e-05, | |
| "loss": 0.0446, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.9993446920052425, | |
| "grad_norm": 0.0049235946498811245, | |
| "learning_rate": 3.705703883495146e-05, | |
| "loss": 0.0198, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9942660550458715, | |
| "eval_auc": 0.9860887006409395, | |
| "eval_f1": 0.7058823529411765, | |
| "eval_loss": 0.02599955163896084, | |
| "eval_precision": 0.9333333333333333, | |
| "eval_recall": 0.5675675675675675, | |
| "eval_runtime": 71.6069, | |
| "eval_samples_per_second": 85.243, | |
| "eval_steps_per_second": 5.335, | |
| "step": 3052 | |
| }, | |
| { | |
| "epoch": 1.0075360419397117, | |
| "grad_norm": 0.008600810542702675, | |
| "learning_rate": 3.690533980582524e-05, | |
| "loss": 0.0216, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.015727391874181, | |
| "grad_norm": 0.730075478553772, | |
| "learning_rate": 3.6753640776699034e-05, | |
| "loss": 0.0736, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.0239187418086502, | |
| "grad_norm": 0.0917787104845047, | |
| "learning_rate": 3.660194174757282e-05, | |
| "loss": 0.0145, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.0321100917431192, | |
| "grad_norm": 0.012463403865695, | |
| "learning_rate": 3.6450242718446604e-05, | |
| "loss": 0.0127, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.0403014416775884, | |
| "grad_norm": 0.0015968201914802194, | |
| "learning_rate": 3.629854368932039e-05, | |
| "loss": 0.0014, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.0484927916120577, | |
| "grad_norm": 0.00465328199788928, | |
| "learning_rate": 3.614684466019418e-05, | |
| "loss": 0.0266, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.056684141546527, | |
| "grad_norm": 1.6281688213348389, | |
| "learning_rate": 3.599514563106796e-05, | |
| "loss": 0.0324, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.0648754914809961, | |
| "grad_norm": 0.006340661086142063, | |
| "learning_rate": 3.584344660194175e-05, | |
| "loss": 0.0197, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.0730668414154652, | |
| "grad_norm": 0.0019152916502207518, | |
| "learning_rate": 3.5691747572815535e-05, | |
| "loss": 0.0056, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.0812581913499344, | |
| "grad_norm": 0.0004100031510461122, | |
| "learning_rate": 3.554004854368932e-05, | |
| "loss": 0.0003, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.0894495412844036, | |
| "grad_norm": 0.0011120839044451714, | |
| "learning_rate": 3.5388349514563105e-05, | |
| "loss": 0.0084, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.0976408912188729, | |
| "grad_norm": 0.0005429446464404464, | |
| "learning_rate": 3.52366504854369e-05, | |
| "loss": 0.0048, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.105832241153342, | |
| "grad_norm": 0.0026772853452712297, | |
| "learning_rate": 3.508495145631068e-05, | |
| "loss": 0.0449, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.1140235910878113, | |
| "grad_norm": 0.09353246539831161, | |
| "learning_rate": 3.493325242718447e-05, | |
| "loss": 0.0587, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.1222149410222806, | |
| "grad_norm": 0.010271481238305569, | |
| "learning_rate": 3.478155339805825e-05, | |
| "loss": 0.0019, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.1304062909567496, | |
| "grad_norm": 0.008390023373067379, | |
| "learning_rate": 3.4629854368932044e-05, | |
| "loss": 0.0244, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.1385976408912188, | |
| "grad_norm": 0.0010115652112290263, | |
| "learning_rate": 3.447815533980583e-05, | |
| "loss": 0.0416, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.146788990825688, | |
| "grad_norm": 0.10442027449607849, | |
| "learning_rate": 3.4326456310679614e-05, | |
| "loss": 0.0777, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.1549803407601573, | |
| "grad_norm": 0.005371388513594866, | |
| "learning_rate": 3.41747572815534e-05, | |
| "loss": 0.0026, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.1631716906946266, | |
| "grad_norm": 0.00910027977079153, | |
| "learning_rate": 3.4023058252427183e-05, | |
| "loss": 0.0191, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.1713630406290956, | |
| "grad_norm": 0.001924333511851728, | |
| "learning_rate": 3.3871359223300975e-05, | |
| "loss": 0.0037, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.1795543905635648, | |
| "grad_norm": 0.006023438181728125, | |
| "learning_rate": 3.371966019417476e-05, | |
| "loss": 0.061, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.187745740498034, | |
| "grad_norm": 0.07021211832761765, | |
| "learning_rate": 3.3567961165048545e-05, | |
| "loss": 0.0494, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.1959370904325033, | |
| "grad_norm": 0.007707479875534773, | |
| "learning_rate": 3.341626213592233e-05, | |
| "loss": 0.009, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.2041284403669725, | |
| "grad_norm": 0.005671200342476368, | |
| "learning_rate": 3.326456310679612e-05, | |
| "loss": 0.0006, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.2123197903014418, | |
| "grad_norm": 0.030211837962269783, | |
| "learning_rate": 3.31128640776699e-05, | |
| "loss": 0.0139, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.2205111402359108, | |
| "grad_norm": 0.4061221778392792, | |
| "learning_rate": 3.296116504854369e-05, | |
| "loss": 0.004, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.22870249017038, | |
| "grad_norm": 0.001372786471620202, | |
| "learning_rate": 3.280946601941748e-05, | |
| "loss": 0.0237, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.2368938401048493, | |
| "grad_norm": 0.0021536455024033785, | |
| "learning_rate": 3.265776699029127e-05, | |
| "loss": 0.0001, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.2450851900393185, | |
| "grad_norm": 0.00104960345197469, | |
| "learning_rate": 3.2506067961165047e-05, | |
| "loss": 0.0089, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.2532765399737877, | |
| "grad_norm": 0.0027104991022497416, | |
| "learning_rate": 3.235436893203884e-05, | |
| "loss": 0.0007, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.261467889908257, | |
| "grad_norm": 0.012539150193333626, | |
| "learning_rate": 3.220266990291262e-05, | |
| "loss": 0.0253, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.269659239842726, | |
| "grad_norm": 0.00037872151006013155, | |
| "learning_rate": 3.205097087378641e-05, | |
| "loss": 0.0003, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.2778505897771952, | |
| "grad_norm": 0.00394083047285676, | |
| "learning_rate": 3.189927184466019e-05, | |
| "loss": 0.0005, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.2860419397116645, | |
| "grad_norm": 0.00036229859688319266, | |
| "learning_rate": 3.1747572815533985e-05, | |
| "loss": 0.0001, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.2942332896461337, | |
| "grad_norm": 0.0016652983613312244, | |
| "learning_rate": 3.159587378640777e-05, | |
| "loss": 0.013, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.302424639580603, | |
| "grad_norm": 0.018371159210801125, | |
| "learning_rate": 3.1444174757281555e-05, | |
| "loss": 0.0301, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.310615989515072, | |
| "grad_norm": 3.56237530708313, | |
| "learning_rate": 3.129247572815534e-05, | |
| "loss": 0.011, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.3188073394495412, | |
| "grad_norm": 0.0032360991463065147, | |
| "learning_rate": 3.114077669902913e-05, | |
| "loss": 0.0005, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.3269986893840104, | |
| "grad_norm": 0.031877271831035614, | |
| "learning_rate": 3.099514563106796e-05, | |
| "loss": 0.057, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.3351900393184797, | |
| "grad_norm": 0.004732233006507158, | |
| "learning_rate": 3.084344660194175e-05, | |
| "loss": 0.0015, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.343381389252949, | |
| "grad_norm": 0.00040538341272622347, | |
| "learning_rate": 3.0691747572815536e-05, | |
| "loss": 0.0092, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.3515727391874182, | |
| "grad_norm": 0.0015137548325583339, | |
| "learning_rate": 3.054004854368932e-05, | |
| "loss": 0.0033, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.3597640891218874, | |
| "grad_norm": 0.00044548933510668576, | |
| "learning_rate": 3.038834951456311e-05, | |
| "loss": 0.0221, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.3679554390563564, | |
| "grad_norm": 0.007979629561305046, | |
| "learning_rate": 3.0236650485436897e-05, | |
| "loss": 0.0546, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.3761467889908257, | |
| "grad_norm": 0.0093601755797863, | |
| "learning_rate": 3.0084951456310682e-05, | |
| "loss": 0.0057, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.384338138925295, | |
| "grad_norm": 0.004190749954432249, | |
| "learning_rate": 2.9933252427184467e-05, | |
| "loss": 0.023, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.3925294888597641, | |
| "grad_norm": 0.0023747687228024006, | |
| "learning_rate": 2.9781553398058252e-05, | |
| "loss": 0.0104, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.4007208387942334, | |
| "grad_norm": 0.003429379081353545, | |
| "learning_rate": 2.962985436893204e-05, | |
| "loss": 0.0403, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.4089121887287024, | |
| "grad_norm": 0.0027962957974523306, | |
| "learning_rate": 2.9478155339805826e-05, | |
| "loss": 0.0002, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.4171035386631716, | |
| "grad_norm": 0.0007880463381297886, | |
| "learning_rate": 2.932645631067961e-05, | |
| "loss": 0.0002, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.4252948885976409, | |
| "grad_norm": 0.0016833733534440398, | |
| "learning_rate": 2.91747572815534e-05, | |
| "loss": 0.0, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.43348623853211, | |
| "grad_norm": 0.001454943441785872, | |
| "learning_rate": 2.9023058252427187e-05, | |
| "loss": 0.0001, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.4416775884665793, | |
| "grad_norm": 0.0008237131405621767, | |
| "learning_rate": 2.887135922330097e-05, | |
| "loss": 0.0541, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.4498689384010484, | |
| "grad_norm": 0.004223650321364403, | |
| "learning_rate": 2.8719660194174757e-05, | |
| "loss": 0.0721, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.4580602883355178, | |
| "grad_norm": 0.0012133247219026089, | |
| "learning_rate": 2.8567961165048546e-05, | |
| "loss": 0.0007, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.4662516382699868, | |
| "grad_norm": 0.004204601515084505, | |
| "learning_rate": 2.8416262135922334e-05, | |
| "loss": 0.0001, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.474442988204456, | |
| "grad_norm": 0.22309060394763947, | |
| "learning_rate": 2.8264563106796115e-05, | |
| "loss": 0.0446, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.4826343381389253, | |
| "grad_norm": 0.0030369218438863754, | |
| "learning_rate": 2.8112864077669904e-05, | |
| "loss": 0.007, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.4908256880733946, | |
| "grad_norm": 0.0010259496048092842, | |
| "learning_rate": 2.7961165048543692e-05, | |
| "loss": 0.0053, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.4990170380078638, | |
| "grad_norm": 0.00154329021461308, | |
| "learning_rate": 2.780946601941748e-05, | |
| "loss": 0.0248, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.5072083879423328, | |
| "grad_norm": 0.0024350325111299753, | |
| "learning_rate": 2.7657766990291262e-05, | |
| "loss": 0.0108, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.515399737876802, | |
| "grad_norm": 0.003887260565534234, | |
| "learning_rate": 2.750606796116505e-05, | |
| "loss": 0.0017, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.5235910878112713, | |
| "grad_norm": 0.030795983970165253, | |
| "learning_rate": 2.735436893203884e-05, | |
| "loss": 0.038, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.5317824377457405, | |
| "grad_norm": 0.0022212048061192036, | |
| "learning_rate": 2.7202669902912624e-05, | |
| "loss": 0.0016, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.5399737876802098, | |
| "grad_norm": 0.0006829659687355161, | |
| "learning_rate": 2.705097087378641e-05, | |
| "loss": 0.0086, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.5481651376146788, | |
| "grad_norm": 0.0009496193379163742, | |
| "learning_rate": 2.6899271844660197e-05, | |
| "loss": 0.0035, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.5563564875491482, | |
| "grad_norm": 0.0016711537027731538, | |
| "learning_rate": 2.6747572815533982e-05, | |
| "loss": 0.004, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.5645478374836173, | |
| "grad_norm": 0.0007192002376541495, | |
| "learning_rate": 2.6595873786407767e-05, | |
| "loss": 0.0008, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 1.5727391874180865, | |
| "grad_norm": 0.001931645441800356, | |
| "learning_rate": 2.6444174757281552e-05, | |
| "loss": 0.0402, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.5809305373525557, | |
| "grad_norm": 0.0018184804357588291, | |
| "learning_rate": 2.629247572815534e-05, | |
| "loss": 0.0034, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 1.5891218872870247, | |
| "grad_norm": 0.0007973504834808409, | |
| "learning_rate": 2.614077669902913e-05, | |
| "loss": 0.0007, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.5973132372214942, | |
| "grad_norm": 0.0021144933998584747, | |
| "learning_rate": 2.598907766990291e-05, | |
| "loss": 0.0417, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 1.6055045871559632, | |
| "grad_norm": 0.0050345249474048615, | |
| "learning_rate": 2.58373786407767e-05, | |
| "loss": 0.0188, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.6136959370904325, | |
| "grad_norm": 0.0019546381663531065, | |
| "learning_rate": 2.5685679611650487e-05, | |
| "loss": 0.0026, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 1.6218872870249017, | |
| "grad_norm": 0.0003715524508152157, | |
| "learning_rate": 2.5533980582524275e-05, | |
| "loss": 0.0119, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.630078636959371, | |
| "grad_norm": 0.0003383459406904876, | |
| "learning_rate": 2.5382281553398057e-05, | |
| "loss": 0.0, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.6382699868938402, | |
| "grad_norm": 0.0004806627694051713, | |
| "learning_rate": 2.5230582524271845e-05, | |
| "loss": 0.0009, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.6464613368283092, | |
| "grad_norm": 0.0002839645021595061, | |
| "learning_rate": 2.5078883495145633e-05, | |
| "loss": 0.0001, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 1.6546526867627787, | |
| "grad_norm": 0.00042549462523311377, | |
| "learning_rate": 2.492718446601942e-05, | |
| "loss": 0.0042, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 1.6628440366972477, | |
| "grad_norm": 0.0005391003796830773, | |
| "learning_rate": 2.4775485436893207e-05, | |
| "loss": 0.0566, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 1.671035386631717, | |
| "grad_norm": 0.00017977178504224867, | |
| "learning_rate": 2.462378640776699e-05, | |
| "loss": 0.0591, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.6792267365661862, | |
| "grad_norm": 0.7689712047576904, | |
| "learning_rate": 2.447208737864078e-05, | |
| "loss": 0.0574, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 1.6874180865006552, | |
| "grad_norm": 0.5321087837219238, | |
| "learning_rate": 2.4320388349514565e-05, | |
| "loss": 0.0536, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 1.6956094364351246, | |
| "grad_norm": 0.00359991448931396, | |
| "learning_rate": 2.4168689320388353e-05, | |
| "loss": 0.0181, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 1.7038007863695936, | |
| "grad_norm": 0.0036572192329913378, | |
| "learning_rate": 2.4016990291262138e-05, | |
| "loss": 0.0074, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.7119921363040629, | |
| "grad_norm": 0.0010478459298610687, | |
| "learning_rate": 2.3865291262135923e-05, | |
| "loss": 0.0071, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 1.7201834862385321, | |
| "grad_norm": 0.001847580773755908, | |
| "learning_rate": 2.3713592233009708e-05, | |
| "loss": 0.0005, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 1.7283748361730014, | |
| "grad_norm": 0.00035062164533883333, | |
| "learning_rate": 2.3561893203883497e-05, | |
| "loss": 0.0005, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 1.7365661861074706, | |
| "grad_norm": 0.006821767892688513, | |
| "learning_rate": 2.341019417475728e-05, | |
| "loss": 0.0295, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.7447575360419396, | |
| "grad_norm": 0.02410910837352276, | |
| "learning_rate": 2.3258495145631066e-05, | |
| "loss": 0.0313, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 1.752948885976409, | |
| "grad_norm": 0.0028191821184009314, | |
| "learning_rate": 2.3106796116504855e-05, | |
| "loss": 0.0264, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 1.761140235910878, | |
| "grad_norm": 0.39254188537597656, | |
| "learning_rate": 2.295509708737864e-05, | |
| "loss": 0.0009, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 1.7693315858453473, | |
| "grad_norm": 0.001157455495558679, | |
| "learning_rate": 2.2803398058252428e-05, | |
| "loss": 0.0353, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.7775229357798166, | |
| "grad_norm": 0.0038004741072654724, | |
| "learning_rate": 2.2651699029126213e-05, | |
| "loss": 0.0085, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 1.7857142857142856, | |
| "grad_norm": 0.00042372927418909967, | |
| "learning_rate": 2.25e-05, | |
| "loss": 0.016, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 1.793905635648755, | |
| "grad_norm": 0.0004292895901016891, | |
| "learning_rate": 2.2348300970873786e-05, | |
| "loss": 0.0072, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 1.802096985583224, | |
| "grad_norm": 0.0011797266779467463, | |
| "learning_rate": 2.2196601941747575e-05, | |
| "loss": 0.0012, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.8102883355176933, | |
| "grad_norm": 0.0008703326457180083, | |
| "learning_rate": 2.204490291262136e-05, | |
| "loss": 0.0002, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 1.8184796854521625, | |
| "grad_norm": 0.00140041159465909, | |
| "learning_rate": 2.1893203883495148e-05, | |
| "loss": 0.0433, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 1.8266710353866316, | |
| "grad_norm": 3.4882562160491943, | |
| "learning_rate": 2.1741504854368933e-05, | |
| "loss": 0.0045, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 1.834862385321101, | |
| "grad_norm": 0.0005097444518469274, | |
| "learning_rate": 2.158980582524272e-05, | |
| "loss": 0.0001, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.84305373525557, | |
| "grad_norm": 0.0011758035980165005, | |
| "learning_rate": 2.1438106796116506e-05, | |
| "loss": 0.0636, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 1.8512450851900393, | |
| "grad_norm": 0.00332819577306509, | |
| "learning_rate": 2.1286407766990295e-05, | |
| "loss": 0.0264, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 1.8594364351245085, | |
| "grad_norm": 0.031183086335659027, | |
| "learning_rate": 2.113470873786408e-05, | |
| "loss": 0.0134, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 1.8676277850589778, | |
| "grad_norm": 0.0163226630538702, | |
| "learning_rate": 2.0983009708737868e-05, | |
| "loss": 0.0217, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.875819134993447, | |
| "grad_norm": 0.01119129080325365, | |
| "learning_rate": 2.0831310679611653e-05, | |
| "loss": 0.0307, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 1.884010484927916, | |
| "grad_norm": 0.002649197354912758, | |
| "learning_rate": 2.0679611650485438e-05, | |
| "loss": 0.0043, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 1.8922018348623855, | |
| "grad_norm": 0.03346943110227585, | |
| "learning_rate": 2.0527912621359223e-05, | |
| "loss": 0.0139, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 1.9003931847968545, | |
| "grad_norm": 0.00223180977627635, | |
| "learning_rate": 2.0376213592233008e-05, | |
| "loss": 0.0685, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.9085845347313237, | |
| "grad_norm": 0.0009488385985605419, | |
| "learning_rate": 2.0224514563106796e-05, | |
| "loss": 0.0004, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 1.916775884665793, | |
| "grad_norm": 0.0017646062187850475, | |
| "learning_rate": 2.007281553398058e-05, | |
| "loss": 0.0002, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 1.924967234600262, | |
| "grad_norm": 0.0024130092933773994, | |
| "learning_rate": 1.992111650485437e-05, | |
| "loss": 0.0003, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 1.9331585845347314, | |
| "grad_norm": 0.0008109256159514189, | |
| "learning_rate": 1.9769417475728154e-05, | |
| "loss": 0.0223, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.9413499344692005, | |
| "grad_norm": 0.0029513093177229166, | |
| "learning_rate": 1.9617718446601943e-05, | |
| "loss": 0.0009, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 1.9495412844036697, | |
| "grad_norm": 0.006773350294679403, | |
| "learning_rate": 1.9466019417475728e-05, | |
| "loss": 0.0058, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 1.957732634338139, | |
| "grad_norm": 0.003487163921818137, | |
| "learning_rate": 1.9314320388349516e-05, | |
| "loss": 0.0003, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 1.9659239842726082, | |
| "grad_norm": 0.002628578804433346, | |
| "learning_rate": 1.91626213592233e-05, | |
| "loss": 0.0115, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.9741153342070774, | |
| "grad_norm": 0.001333425403572619, | |
| "learning_rate": 1.901092233009709e-05, | |
| "loss": 0.0054, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 1.9823066841415464, | |
| "grad_norm": 0.0018065802287310362, | |
| "learning_rate": 1.8859223300970874e-05, | |
| "loss": 0.0467, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 1.9904980340760159, | |
| "grad_norm": 0.00119776523206383, | |
| "learning_rate": 1.8707524271844663e-05, | |
| "loss": 0.0194, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 1.998689384010485, | |
| "grad_norm": 0.049537669867277145, | |
| "learning_rate": 1.8555825242718448e-05, | |
| "loss": 0.0003, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9985255570117956, | |
| "eval_auc": 0.9973051409618573, | |
| "eval_f1": 0.9361702127659575, | |
| "eval_loss": 0.00864658784121275, | |
| "eval_precision": 0.9850746268656716, | |
| "eval_recall": 0.8918918918918919, | |
| "eval_runtime": 71.46, | |
| "eval_samples_per_second": 85.418, | |
| "eval_steps_per_second": 5.346, | |
| "step": 6104 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 9156, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.01 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.9610164221187195e+19, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |