CoolFace
Modelpublic

windgrin/c1

sourceHugging Faceupdated 1y agoView on Hugging Face
0likes2downloads
trainer_state.json1099 linesDownload Raw Back to root
1{2  "best_global_step": null,3  "best_metric": null,4  "best_model_checkpoint": null,5  "epoch": 1.0,6  "eval_steps": 500,7  "global_step": 152,8  "is_hyper_param_search": false,9  "is_local_process_zero": true,10  "is_world_process_zero": true,11  "log_history": [12    {13      "epoch": 0.006600660066006601,14      "grad_norm": 3.143493175506592,15      "learning_rate": 0.0,16      "loss": 2.5041,17      "step": 118    },19    {20      "epoch": 0.013201320132013201,21      "grad_norm": 4.06755256652832,22      "learning_rate": 4.347826086956522e-06,23      "loss": 2.5644,24      "step": 225    },26    {27      "epoch": 0.019801980198019802,28      "grad_norm": 3.0701375007629395,29      "learning_rate": 8.695652173913044e-06,30      "loss": 2.5095,31      "step": 332    },33    {34      "epoch": 0.026402640264026403,35      "grad_norm": 3.792125701904297,36      "learning_rate": 1.3043478260869566e-05,37      "loss": 2.426,38      "step": 439    },40    {41      "epoch": 0.033003300330033,42      "grad_norm": 4.194362640380859,43      "learning_rate": 1.739130434782609e-05,44      "loss": 1.973,45      "step": 546    },47    {48      "epoch": 0.039603960396039604,49      "grad_norm": 4.066171646118164,50      "learning_rate": 2.173913043478261e-05,51      "loss": 2.2242,52      "step": 653    },54    {55      "epoch": 0.0462046204620462,56      "grad_norm": 3.9235055446624756,57      "learning_rate": 2.608695652173913e-05,58      "loss": 2.5398,59      "step": 760    },61    {62      "epoch": 0.052805280528052806,63      "grad_norm": 3.5214486122131348,64      "learning_rate": 3.0434782608695656e-05,65      "loss": 1.9146,66      "step": 867    },68    {69      "epoch": 0.0594059405940594,70      "grad_norm": 3.2443058490753174,71      "learning_rate": 3.478260869565218e-05,72      "loss": 2.0756,73      "step": 974    },75    {76      "epoch": 0.066006600660066,77      "grad_norm": 3.280684471130371,78      "learning_rate": 3.91304347826087e-05,79      "loss": 2.3453,80      "step": 1081    },82    {83      "epoch": 0.07260726072607261,84      "grad_norm": 2.701324701309204,85      "learning_rate": 4.347826086956522e-05,86      "loss": 1.6499,87      "step": 1188    },89    {90      "epoch": 0.07920792079207921,91      "grad_norm": 2.205188035964966,92      "learning_rate": 4.782608695652174e-05,93      "loss": 1.8142,94      "step": 1295    },96    {97      "epoch": 0.0858085808580858,98      "grad_norm": 2.969386577606201,99      "learning_rate": 5.217391304347826e-05,100      "loss": 1.6555,101      "step": 13102    },103    {104      "epoch": 0.0924092409240924,105      "grad_norm": 3.1078085899353027,106      "learning_rate": 5.652173913043478e-05,107      "loss": 1.892,108      "step": 14109    },110    {111      "epoch": 0.09900990099009901,112      "grad_norm": 2.403679847717285,113      "learning_rate": 6.086956521739131e-05,114      "loss": 2.1263,115      "step": 15116    },117    {118      "epoch": 0.10561056105610561,119      "grad_norm": 3.7624003887176514,120      "learning_rate": 6.521739130434783e-05,121      "loss": 2.1824,122      "step": 16123    },124    {125      "epoch": 0.11221122112211221,126      "grad_norm": 2.571948766708374,127      "learning_rate": 6.956521739130436e-05,128      "loss": 2.2477,129      "step": 17130    },131    {132      "epoch": 0.1188118811881188,133      "grad_norm": 3.469731569290161,134      "learning_rate": 7.391304347826086e-05,135      "loss": 1.4427,136      "step": 18137    },138    {139      "epoch": 0.1254125412541254,140      "grad_norm": 2.207329750061035,141      "learning_rate": 7.82608695652174e-05,142      "loss": 1.5447,143      "step": 19144    },145    {146      "epoch": 0.132013201320132,147      "grad_norm": 2.514176368713379,148      "learning_rate": 8.260869565217392e-05,149      "loss": 1.9723,150      "step": 20151    },152    {153      "epoch": 0.13861386138613863,154      "grad_norm": 3.027451753616333,155      "learning_rate": 8.695652173913044e-05,156      "loss": 1.7312,157      "step": 21158    },159    {160      "epoch": 0.14521452145214522,161      "grad_norm": 5.07168436050415,162      "learning_rate": 9.130434782608696e-05,163      "loss": 2.5444,164      "step": 22165    },166    {167      "epoch": 0.15181518151815182,168      "grad_norm": 3.528111219406128,169      "learning_rate": 9.565217391304348e-05,170      "loss": 2.1284,171      "step": 23172    },173    {174      "epoch": 0.15841584158415842,175      "grad_norm": 3.0631463527679443,176      "learning_rate": 0.0001,177      "loss": 2.2181,178      "step": 24179    },180    {181      "epoch": 0.16501650165016502,182      "grad_norm": 3.331843852996826,183      "learning_rate": 0.00010434782608695653,184      "loss": 3.0526,185      "step": 25186    },187    {188      "epoch": 0.1716171617161716,189      "grad_norm": 2.444413661956787,190      "learning_rate": 0.00010869565217391305,191      "loss": 1.7363,192      "step": 26193    },194    {195      "epoch": 0.1782178217821782,196      "grad_norm": 2.5538690090179443,197      "learning_rate": 0.00011304347826086956,198      "loss": 1.5258,199      "step": 27200    },201    {202      "epoch": 0.1848184818481848,203      "grad_norm": 2.088798999786377,204      "learning_rate": 0.0001173913043478261,205      "loss": 1.8808,206      "step": 28207    },208    {209      "epoch": 0.19141914191419143,210      "grad_norm": 2.5747692584991455,211      "learning_rate": 0.00012173913043478263,212      "loss": 2.6931,213      "step": 29214    },215    {216      "epoch": 0.19801980198019803,217      "grad_norm": 2.9658546447753906,218      "learning_rate": 0.00012608695652173915,219      "loss": 2.0258,220      "step": 30221    },222    {223      "epoch": 0.20462046204620463,224      "grad_norm": 2.768965244293213,225      "learning_rate": 0.00013043478260869567,226      "loss": 1.8468,227      "step": 31228    },229    {230      "epoch": 0.21122112211221122,231      "grad_norm": 2.2689647674560547,232      "learning_rate": 0.0001347826086956522,233      "loss": 1.7859,234      "step": 32235    },236    {237      "epoch": 0.21782178217821782,238      "grad_norm": 2.06204891204834,239      "learning_rate": 0.0001391304347826087,240      "loss": 1.8043,241      "step": 33242    },243    {244      "epoch": 0.22442244224422442,245      "grad_norm": 2.958625078201294,246      "learning_rate": 0.0001434782608695652,247      "loss": 1.2597,248      "step": 34249    },250    {251      "epoch": 0.23102310231023102,252      "grad_norm": 2.6240015029907227,253      "learning_rate": 0.00014782608695652173,254      "loss": 1.9745,255      "step": 35256    },257    {258      "epoch": 0.2376237623762376,259      "grad_norm": 2.636274814605713,260      "learning_rate": 0.00015217391304347827,261      "loss": 1.8066,262      "step": 36263    },264    {265      "epoch": 0.24422442244224424,266      "grad_norm": 2.4695310592651367,267      "learning_rate": 0.0001565217391304348,268      "loss": 1.5221,269      "step": 37270    },271    {272      "epoch": 0.2508250825082508,273      "grad_norm": 3.572252035140991,274      "learning_rate": 0.00016086956521739132,275      "loss": 1.8853,276      "step": 38277    },278    {279      "epoch": 0.25742574257425743,280      "grad_norm": 5.93402624130249,281      "learning_rate": 0.00016521739130434784,282      "loss": 2.3075,283      "step": 39284    },285    {286      "epoch": 0.264026402640264,287      "grad_norm": 2.454044818878174,288      "learning_rate": 0.00016956521739130436,289      "loss": 1.5395,290      "step": 40291    },292    {293      "epoch": 0.2706270627062706,294      "grad_norm": 2.4715137481689453,295      "learning_rate": 0.00017391304347826088,296      "loss": 1.6654,297      "step": 41298    },299    {300      "epoch": 0.27722772277227725,301      "grad_norm": 2.5912485122680664,302      "learning_rate": 0.0001782608695652174,303      "loss": 2.3238,304      "step": 42305    },306    {307      "epoch": 0.2838283828382838,308      "grad_norm": 2.6347994804382324,309      "learning_rate": 0.00018260869565217392,310      "loss": 1.7519,311      "step": 43312    },313    {314      "epoch": 0.29042904290429045,315      "grad_norm": 2.668267011642456,316      "learning_rate": 0.00018695652173913045,317      "loss": 1.6689,318      "step": 44319    },320    {321      "epoch": 0.297029702970297,322      "grad_norm": 3.038886308670044,323      "learning_rate": 0.00019130434782608697,324      "loss": 1.968,325      "step": 45326    },327    {328      "epoch": 0.30363036303630364,329      "grad_norm": 2.065838575363159,330      "learning_rate": 0.0001956521739130435,331      "loss": 1.8324,332      "step": 46333    },334    {335      "epoch": 0.3102310231023102,336      "grad_norm": 1.9431511163711548,337      "learning_rate": 0.0002,338      "loss": 1.5543,339      "step": 47340    },341    {342      "epoch": 0.31683168316831684,343      "grad_norm": 2.310769557952881,344      "learning_rate": 0.00019999702094326033,345      "loss": 1.4833,346      "step": 48347    },348    {349      "epoch": 0.3234323432343234,350      "grad_norm": 2.8865110874176025,351      "learning_rate": 0.00019998808395053688,352      "loss": 1.7368,353      "step": 49354    },355    {356      "epoch": 0.33003300330033003,357      "grad_norm": 2.3484208583831787,358      "learning_rate": 0.0001999731895543058,359      "loss": 1.4685,360      "step": 50361    },362    {363      "epoch": 0.33663366336633666,364      "grad_norm": 2.802156686782837,365      "learning_rate": 0.00019995233864199215,366      "loss": 1.7877,367      "step": 51368    },369    {370      "epoch": 0.3432343234323432,371      "grad_norm": 2.18292498588562,372      "learning_rate": 0.00019992553245591694,373      "loss": 1.5654,374      "step": 52375    },376    {377      "epoch": 0.34983498349834985,378      "grad_norm": 2.082122325897217,379      "learning_rate": 0.00019989277259322314,380      "loss": 1.8541,381      "step": 53382    },383    {384      "epoch": 0.3564356435643564,385      "grad_norm": 2.40657377243042,386      "learning_rate": 0.0001998540610057806,387      "loss": 1.9715,388      "step": 54389    },390    {391      "epoch": 0.36303630363036304,392      "grad_norm": 2.077791929244995,393      "learning_rate": 0.00019980940000006954,394      "loss": 1.5949,395      "step": 55396    },397    {398      "epoch": 0.3696369636963696,399      "grad_norm": 2.2711575031280518,400      "learning_rate": 0.0001997587922370434,401      "loss": 1.0556,402      "step": 56403    },404    {405      "epoch": 0.37623762376237624,406      "grad_norm": 1.9230396747589111,407      "learning_rate": 0.00019970224073197017,408      "loss": 1.8958,409      "step": 57410    },411    {412      "epoch": 0.38283828382838286,413      "grad_norm": 2.0223827362060547,414      "learning_rate": 0.00019963974885425266,415      "loss": 1.8731,416      "step": 58417    },418    {419      "epoch": 0.38943894389438943,420      "grad_norm": 3.312962293624878,421      "learning_rate": 0.00019957132032722785,422      "loss": 1.8496,423      "step": 59424    },425    {426      "epoch": 0.39603960396039606,427      "grad_norm": 3.0161566734313965,428      "learning_rate": 0.00019949695922794506,429      "loss": 1.8099,430      "step": 60431    },432    {433      "epoch": 0.40264026402640263,434      "grad_norm": 2.2380363941192627,435      "learning_rate": 0.00019941666998692296,436      "loss": 1.6939,437      "step": 61438    },439    {440      "epoch": 0.40924092409240925,441      "grad_norm": 2.783853530883789,442      "learning_rate": 0.00019933045738788563,443      "loss": 2.1385,444      "step": 62445    },446    {447      "epoch": 0.4158415841584158,448      "grad_norm": 2.4008877277374268,449      "learning_rate": 0.0001992383265674776,450      "loss": 1.6338,451      "step": 63452    },453    {454      "epoch": 0.42244224422442245,455      "grad_norm": 2.317880868911743,456      "learning_rate": 0.0001991402830149576,457      "loss": 1.6309,458      "step": 64459    },460    {461      "epoch": 0.429042904290429,462      "grad_norm": 2.7851943969726562,463      "learning_rate": 0.00019903633257187185,464      "loss": 1.9731,465      "step": 65466    },467    {468      "epoch": 0.43564356435643564,469      "grad_norm": 2.795436143875122,470      "learning_rate": 0.00019892648143170566,471      "loss": 1.7634,472      "step": 66473    },474    {475      "epoch": 0.44224422442244227,476      "grad_norm": 3.0923502445220947,477      "learning_rate": 0.0001988107361395146,478      "loss": 1.7409,479      "step": 67480    },481    {482      "epoch": 0.44884488448844884,483      "grad_norm": 2.286417245864868,484      "learning_rate": 0.0001986891035915346,485      "loss": 1.7104,486      "step": 68487    },488    {489      "epoch": 0.45544554455445546,490      "grad_norm": 2.321275234222412,491      "learning_rate": 0.00019856159103477086,492      "loss": 1.8864,493      "step": 69494    },495    {496      "epoch": 0.46204620462046203,497      "grad_norm": 1.9481537342071533,498      "learning_rate": 0.0001984282060665662,499      "loss": 1.7544,500      "step": 70501    },502    {503      "epoch": 0.46864686468646866,504      "grad_norm": 2.094721794128418,505      "learning_rate": 0.00019828895663414836,506      "loss": 1.2463,507      "step": 71508    },509    {510      "epoch": 0.4752475247524752,511      "grad_norm": 2.0305490493774414,512      "learning_rate": 0.00019814385103415663,513      "loss": 1.7909,514      "step": 72515    },516    {517      "epoch": 0.48184818481848185,518      "grad_norm": 2.0635201930999756,519      "learning_rate": 0.00019799289791214725,520      "loss": 1.5155,521      "step": 73522    },523    {524      "epoch": 0.4884488448844885,525      "grad_norm": 1.9585940837860107,526      "learning_rate": 0.00019783610626207854,527      "loss": 1.6954,528      "step": 74529    },530    {531      "epoch": 0.49504950495049505,532      "grad_norm": 2.560070514678955,533      "learning_rate": 0.00019767348542577495,534      "loss": 1.8156,535      "step": 75536    },537    {538      "epoch": 0.5016501650165016,539      "grad_norm": 2.5949947834014893,540      "learning_rate": 0.00019750504509237046,541      "loss": 2.0184,542      "step": 76543    },544    {545      "epoch": 0.5082508250825083,546      "grad_norm": 2.281064748764038,547      "learning_rate": 0.00019733079529773125,548      "loss": 1.6881,549      "step": 77550    },551    {552      "epoch": 0.5148514851485149,553      "grad_norm": 2.1914496421813965,554      "learning_rate": 0.00019715074642385786,555      "loss": 1.4654,556      "step": 78557    },558    {559      "epoch": 0.5214521452145214,560      "grad_norm": 2.1156349182128906,561      "learning_rate": 0.00019696490919826647,562      "loss": 1.7675,563      "step": 79564    },565    {566      "epoch": 0.528052805280528,567      "grad_norm": 2.097763776779175,568      "learning_rate": 0.0001967732946933499,569      "loss": 2.0482,570      "step": 80571    },572    {573      "epoch": 0.5346534653465347,574      "grad_norm": 2.4619557857513428,575      "learning_rate": 0.00019657591432571778,576      "loss": 1.8658,577      "step": 81578    },579    {580      "epoch": 0.5412541254125413,581      "grad_norm": 2.268629312515259,582      "learning_rate": 0.0001963727798555164,583      "loss": 1.9668,584      "step": 82585    },586    {587      "epoch": 0.5478547854785478,588      "grad_norm": 2.8534958362579346,589      "learning_rate": 0.00019616390338572803,590      "loss": 1.5379,591      "step": 83592    },593    {594      "epoch": 0.5544554455445545,595      "grad_norm": 1.972926139831543,596      "learning_rate": 0.00019594929736144976,597      "loss": 1.3666,598      "step": 84599    },600    {601      "epoch": 0.5610561056105611,602      "grad_norm": 2.3708107471466064,603      "learning_rate": 0.00019572897456915202,604      "loss": 1.7144,605      "step": 85606    },607    {608      "epoch": 0.5676567656765676,609      "grad_norm": 2.909595489501953,610      "learning_rate": 0.00019550294813591682,611      "loss": 2.2708,612      "step": 86613    },614    {615      "epoch": 0.5742574257425742,616      "grad_norm": 1.8480359315872192,617      "learning_rate": 0.0001952712315286556,618      "loss": 1.3613,619      "step": 87620    },621    {622      "epoch": 0.5808580858085809,623      "grad_norm": 2.4477217197418213,624      "learning_rate": 0.00019503383855330668,625      "loss": 1.6463,626      "step": 88627    },628    {629      "epoch": 0.5874587458745875,630      "grad_norm": 2.253204345703125,631      "learning_rate": 0.00019479078335401297,632      "loss": 1.2088,633      "step": 89634    },635    {636      "epoch": 0.594059405940594,637      "grad_norm": 1.9275654554367065,638      "learning_rate": 0.00019454208041227907,639      "loss": 1.5648,640      "step": 90641    },642    {643      "epoch": 0.6006600660066007,644      "grad_norm": 1.9389451742172241,645      "learning_rate": 0.00019428774454610843,646      "loss": 1.2708,647      "step": 91648    },649    {650      "epoch": 0.6072607260726073,651      "grad_norm": 1.9256659746170044,652      "learning_rate": 0.0001940277909091206,653      "loss": 1.5638,654      "step": 92655    },656    {657      "epoch": 0.6138613861386139,658      "grad_norm": 2.4642438888549805,659      "learning_rate": 0.0001937622349896483,660      "loss": 1.9737,661      "step": 93662    },663    {664      "epoch": 0.6204620462046204,665      "grad_norm": 2.4450933933258057,666      "learning_rate": 0.00019349109260981452,667      "loss": 2.407,668      "step": 94669    },670    {671      "epoch": 0.6270627062706271,672      "grad_norm": 2.6247427463531494,673      "learning_rate": 0.00019321437992458995,674      "loss": 2.0388,675      "step": 95676    },677    {678      "epoch": 0.6336633663366337,679      "grad_norm": 2.9768080711364746,680      "learning_rate": 0.0001929321134208304,681      "loss": 1.8954,682      "step": 96683    },684    {685      "epoch": 0.6402640264026402,686      "grad_norm": 2.293891191482544,687      "learning_rate": 0.00019264430991629446,688      "loss": 2.3185,689      "step": 97690    },691    {692      "epoch": 0.6468646864686468,693      "grad_norm": 2.011204242706299,694      "learning_rate": 0.00019235098655864157,695      "loss": 1.7781,696      "step": 98697    },698    {699      "epoch": 0.6534653465346535,700      "grad_norm": 1.914803147315979,701      "learning_rate": 0.00019205216082441017,702      "loss": 1.5633,703      "step": 99704    },705    {706      "epoch": 0.6600660066006601,707      "grad_norm": 2.229295015335083,708      "learning_rate": 0.00019174785051797666,709      "loss": 1.422,710      "step": 100711    },712    {713      "epoch": 0.6666666666666666,714      "grad_norm": 1.9156501293182373,715      "learning_rate": 0.0001914380737704944,716      "loss": 1.6114,717      "step": 101718    },719    {720      "epoch": 0.6732673267326733,721      "grad_norm": 2.1045665740966797,722      "learning_rate": 0.0001911228490388136,723      "loss": 1.9957,724      "step": 102725    },726    {727      "epoch": 0.6798679867986799,728      "grad_norm": 1.971551775932312,729      "learning_rate": 0.0001908021951043814,730      "loss": 1.3397,731      "step": 103732    },733    {734      "epoch": 0.6864686468646864,735      "grad_norm": 2.331234931945801,736      "learning_rate": 0.00019047613107212312,737      "loss": 1.453,738      "step": 104739    },740    {741      "epoch": 0.693069306930693,742      "grad_norm": 2.3500189781188965,743      "learning_rate": 0.00019014467636930385,744      "loss": 1.5166,745      "step": 105746    },747    {748      "epoch": 0.6996699669966997,749      "grad_norm": 3.211228609085083,750      "learning_rate": 0.00018980785074437093,751      "loss": 2.0583,752      "step": 106753    },754    {755      "epoch": 0.7062706270627063,756      "grad_norm": 2.552626371383667,757      "learning_rate": 0.00018946567426577723,758      "loss": 2.0275,759      "step": 107760    },761    {762      "epoch": 0.7128712871287128,763      "grad_norm": 2.142106294631958,764      "learning_rate": 0.00018911816732078575,765      "loss": 1.5193,766      "step": 108767    },768    {769      "epoch": 0.7194719471947195,770      "grad_norm": 2.5598621368408203,771      "learning_rate": 0.00018876535061425453,772      "loss": 1.7457,773      "step": 109774    },775    {776      "epoch": 0.7260726072607261,777      "grad_norm": 2.658561944961548,778      "learning_rate": 0.0001884072451674034,779      "loss": 1.5258,780      "step": 110781    },782    {783      "epoch": 0.7326732673267327,784      "grad_norm": 2.122279167175293,785      "learning_rate": 0.00018804387231656118,786      "loss": 1.6165,787      "step": 111788    },789    {790      "epoch": 0.7392739273927392,791      "grad_norm": 2.6554064750671387,792      "learning_rate": 0.00018767525371189475,793      "loss": 1.5194,794      "step": 112795    },796    {797      "epoch": 0.7458745874587459,798      "grad_norm": 2.4760324954986572,799      "learning_rate": 0.00018730141131611882,800      "loss": 1.8324,801      "step": 113802    },803    {804      "epoch": 0.7524752475247525,805      "grad_norm": 1.9205211400985718,806      "learning_rate": 0.0001869223674031876,807      "loss": 1.5462,808      "step": 114809    },810    {811      "epoch": 0.759075907590759,812      "grad_norm": 2.451927423477173,813      "learning_rate": 0.00018653814455696756,814      "loss": 1.9891,815      "step": 115816    },817    {818      "epoch": 0.7656765676567657,819      "grad_norm": 2.2201361656188965,820      "learning_rate": 0.0001861487656698919,821      "loss": 1.6986,822      "step": 116823    },824    {825      "epoch": 0.7722772277227723,826      "grad_norm": 3.1413655281066895,827      "learning_rate": 0.00018575425394159653,828      "loss": 2.0776,829      "step": 117830    },831    {832      "epoch": 0.7788778877887789,833      "grad_norm": 2.1141433715820312,834      "learning_rate": 0.00018535463287753796,835      "loss": 1.366,836      "step": 118837    },838    {839      "epoch": 0.7854785478547854,840      "grad_norm": 3.3970096111297607,841      "learning_rate": 0.00018494992628759267,842      "loss": 1.6139,843      "step": 119844    },845    {846      "epoch": 0.7920792079207921,847      "grad_norm": 2.305530548095703,848      "learning_rate": 0.0001845401582846385,849      "loss": 1.4516,850      "step": 120851    },852    {853      "epoch": 0.7986798679867987,854      "grad_norm": 2.0770864486694336,855      "learning_rate": 0.00018412535328311814,856      "loss": 1.7341,857      "step": 121858    },859    {860      "epoch": 0.8052805280528053,861      "grad_norm": 2.0777578353881836,862      "learning_rate": 0.00018370553599758422,863      "loss": 1.5696,864      "step": 122865    },866    {867      "epoch": 0.8118811881188119,868      "grad_norm": 2.021235704421997,869      "learning_rate": 0.00018328073144122708,870      "loss": 1.1397,871      "step": 123872    },873    {874      "epoch": 0.8184818481848185,875      "grad_norm": 2.1600515842437744,876      "learning_rate": 0.00018285096492438424,877      "loss": 1.5525,878      "step": 124879    },880    {881      "epoch": 0.8250825082508251,882      "grad_norm": 2.407317638397217,883      "learning_rate": 0.00018241626205303243,884      "loss": 2.1211,885      "step": 125886    },887    {888      "epoch": 0.8316831683168316,889      "grad_norm": 2.115229845046997,890      "learning_rate": 0.00018197664872726205,891      "loss": 1.4519,892      "step": 126893    },894    {895      "epoch": 0.8382838283828383,896      "grad_norm": 2.243795156478882,897      "learning_rate": 0.00018153215113973396,898      "loss": 1.609,899      "step": 127900    },901    {902      "epoch": 0.8448844884488449,903      "grad_norm": 2.622382164001465,904      "learning_rate": 0.0001810827957741188,905      "loss": 1.9359,906      "step": 128907    },908    {909      "epoch": 0.8514851485148515,910      "grad_norm": 2.31684947013855,911      "learning_rate": 0.00018062860940351916,912      "loss": 1.9789,913      "step": 129914    },915    {916      "epoch": 0.858085808580858,917      "grad_norm": 1.9168429374694824,918      "learning_rate": 0.00018016961908887442,919      "loss": 1.3984,920      "step": 130921    },922    {923      "epoch": 0.8646864686468647,924      "grad_norm": 2.2554709911346436,925      "learning_rate": 0.00017970585217734844,926      "loss": 1.127,927      "step": 131928    },929    {930      "epoch": 0.8712871287128713,931      "grad_norm": 2.3152496814727783,932      "learning_rate": 0.00017923733630070002,933      "loss": 1.922,934      "step": 132935    },936    {937      "epoch": 0.8778877887788779,938      "grad_norm": 2.9799273014068604,939      "learning_rate": 0.00017876409937363677,940      "loss": 1.6876,941      "step": 133942    },943    {944      "epoch": 0.8844884488448845,945      "grad_norm": 1.8281869888305664,946      "learning_rate": 0.00017828616959215184,947      "loss": 1.5791,948      "step": 134949    },950    {951      "epoch": 0.8910891089108911,952      "grad_norm": 1.8797411918640137,953      "learning_rate": 0.00017780357543184397,954      "loss": 1.4774,955      "step": 135956    },957    {958      "epoch": 0.8976897689768977,959      "grad_norm": 2.052123785018921,960      "learning_rate": 0.00017731634564622085,961      "loss": 1.883,962      "step": 136963    },964    {965      "epoch": 0.9042904290429042,966      "grad_norm": 2.0069074630737305,967      "learning_rate": 0.00017682450926498607,968      "loss": 1.4696,969      "step": 137970    },971    {972      "epoch": 0.9108910891089109,973      "grad_norm": 1.6514910459518433,974      "learning_rate": 0.0001763280955923093,975      "loss": 1.4254,976      "step": 138977    },978    {979      "epoch": 0.9174917491749175,980      "grad_norm": 2.6039862632751465,981      "learning_rate": 0.00017582713420508052,982      "loss": 2.1567,983      "step": 139984    },985    {986      "epoch": 0.9240924092409241,987      "grad_norm": 1.9613580703735352,988      "learning_rate": 0.00017532165495114766,989      "loss": 1.0636,990      "step": 140991    },992    {993      "epoch": 0.9306930693069307,994      "grad_norm": 3.350720167160034,995      "learning_rate": 0.0001748116879475383,996      "loss": 1.5513,997      "step": 141998    },999    {1000      "epoch": 0.9372937293729373,1001      "grad_norm": 2.287201404571533,1002      "learning_rate": 0.00017429726357866515,1003      "loss": 2.229,1004      "step": 1421005    },1006    {1007      "epoch": 0.9438943894389439,1008      "grad_norm": 2.0688509941101074,1009      "learning_rate": 0.00017377841249451595,1010      "loss": 1.518,1011      "step": 1431012    },1013    {1014      "epoch": 0.9504950495049505,1015      "grad_norm": 2.2164456844329834,1016      "learning_rate": 0.00017325516560882703,1017      "loss": 1.2676,1018      "step": 1441019    },1020    {1021      "epoch": 0.9570957095709571,1022      "grad_norm": 1.946719765663147,1023      "learning_rate": 0.00017272755409724168,1024      "loss": 1.4389,1025      "step": 1451026    },1027    {1028      "epoch": 0.9636963696369637,1029      "grad_norm": 2.1992857456207275,1030      "learning_rate": 0.00017219560939545246,1031      "loss": 1.5815,1032      "step": 1461033    },1034    {1035      "epoch": 0.9702970297029703,1036      "grad_norm": 1.699536919593811,1037      "learning_rate": 0.0001716593631973283,1038      "loss": 1.1554,1039      "step": 1471040    },1041    {1042      "epoch": 0.976897689768977,1043      "grad_norm": 2.061931610107422,1044      "learning_rate": 0.00017111884745302632,1045      "loss": 1.7294,1046      "step": 1481047    },1048    {1049      "epoch": 0.9834983498349835,1050      "grad_norm": 1.7385263442993164,1051      "learning_rate": 0.00017057409436708782,1052      "loss": 1.2457,1053      "step": 1491054    },1055    {1056      "epoch": 0.9900990099009901,1057      "grad_norm": 2.110952615737915,1058      "learning_rate": 0.00017002513639651992,1059      "loss": 1.3175,1060      "step": 1501061    },1062    {1063      "epoch": 0.9966996699669967,1064      "grad_norm": 2.1564648151397705,1065      "learning_rate": 0.00016947200624886145,1066      "loss": 1.9294,1067      "step": 1511068    },1069    {1070      "epoch": 1.0,1071      "grad_norm": 2.933058977127075,1072      "learning_rate": 0.00016891473688023426,1073      "loss": 1.692,1074      "step": 1521075    }1076  ],1077  "logging_steps": 1,1078  "max_steps": 453,1079  "num_input_tokens_seen": 0,1080  "num_train_epochs": 3,1081  "save_steps": 500,1082  "stateful_callbacks": {1083    "TrainerControl": {1084      "args": {1085        "should_epoch_stop": false,1086        "should_evaluate": false,1087        "should_log": false,1088        "should_save": true,1089        "should_training_stop": false1090      },1091      "attributes": {}1092    }1093  },1094  "total_flos": 1.7929916510204808e+16,1095  "train_batch_size": 2,1096  "trial_name": null,1097  "trial_params": null1098}1099