CoolFace
Modelpublic

gpol13/codet5_sparse_lora_l1

sourceHugging Faceupdated 9mo agoView on Hugging Face
0likes7downloads
trainer_state.json1087 linesDownload Raw Back to root
1{2  "best_global_step": 7075,3  "best_metric": 0.957730518736978,4  "best_model_checkpoint": "/kaggle/working/sparse_lora/checkpoint-7075",5  "epoch": 5.0,6  "eval_steps": 500,7  "global_step": 7075,8  "is_hyper_param_search": false,9  "is_local_process_zero": true,10  "is_world_process_zero": true,11  "log_history": [12    {13      "epoch": 0.0353356890459364,14      "grad_norm": 87870.6484375,15      "learning_rate": 4.9653710247349825e-05,16      "loss": 4.5998,17      "step": 5018    },19    {20      "epoch": 0.0706713780918728,21      "grad_norm": 671269.4375,22      "learning_rate": 4.930035335689046e-05,23      "loss": 3.6646,24      "step": 10025    },26    {27      "epoch": 0.10600706713780919,28      "grad_norm": 103706.0,29      "learning_rate": 4.89469964664311e-05,30      "loss": 3.0394,31      "step": 15032    },33    {34      "epoch": 0.1413427561837456,35      "grad_norm": 60311.45703125,36      "learning_rate": 4.859363957597173e-05,37      "loss": 2.1607,38      "step": 20039    },40    {41      "epoch": 0.17667844522968199,42      "grad_norm": 56506.35546875,43      "learning_rate": 4.8240282685512374e-05,44      "loss": 2.0396,45      "step": 25046    },47    {48      "epoch": 0.21201413427561838,49      "grad_norm": 64563.61328125,50      "learning_rate": 4.788692579505301e-05,51      "loss": 1.8784,52      "step": 30053    },54    {55      "epoch": 0.24734982332155478,56      "grad_norm": 84701.9765625,57      "learning_rate": 4.753356890459364e-05,58      "loss": 1.823,59      "step": 35060    },61    {62      "epoch": 0.2826855123674912,63      "grad_norm": 80899.9765625,64      "learning_rate": 4.718021201413428e-05,65      "loss": 1.825,66      "step": 40067    },68    {69      "epoch": 0.31802120141342755,70      "grad_norm": 83637.8671875,71      "learning_rate": 4.682685512367491e-05,72      "loss": 1.6937,73      "step": 45074    },75    {76      "epoch": 0.35335689045936397,77      "grad_norm": 172655.1875,78      "learning_rate": 4.647349823321555e-05,79      "loss": 1.6309,80      "step": 50081    },82    {83      "epoch": 0.38869257950530034,84      "grad_norm": 212648.234375,85      "learning_rate": 4.612014134275618e-05,86      "loss": 1.5667,87      "step": 55088    },89    {90      "epoch": 0.42402826855123676,91      "grad_norm": 85414.8125,92      "learning_rate": 4.5766784452296823e-05,93      "loss": 1.5433,94      "step": 60095    },96    {97      "epoch": 0.45936395759717313,98      "grad_norm": 84665.4296875,99      "learning_rate": 4.541342756183746e-05,100      "loss": 1.5225,101      "step": 650102    },103    {104      "epoch": 0.49469964664310956,105      "grad_norm": 87850.546875,106      "learning_rate": 4.50600706713781e-05,107      "loss": 1.4696,108      "step": 700109    },110    {111      "epoch": 0.5300353356890459,112      "grad_norm": 105920.03125,113      "learning_rate": 4.470671378091873e-05,114      "loss": 1.5046,115      "step": 750116    },117    {118      "epoch": 0.5653710247349824,119      "grad_norm": 128489.2734375,120      "learning_rate": 4.435335689045937e-05,121      "loss": 1.4268,122      "step": 800123    },124    {125      "epoch": 0.6007067137809188,126      "grad_norm": 126656.5390625,127      "learning_rate": 4.4000000000000006e-05,128      "loss": 1.3749,129      "step": 850130    },131    {132      "epoch": 0.6360424028268551,133      "grad_norm": 90037.234375,134      "learning_rate": 4.364664310954063e-05,135      "loss": 1.4043,136      "step": 900137    },138    {139      "epoch": 0.6713780918727915,140      "grad_norm": 123195.953125,141      "learning_rate": 4.329328621908127e-05,142      "loss": 1.3655,143      "step": 950144    },145    {146      "epoch": 0.7067137809187279,147      "grad_norm": 106699.8984375,148      "learning_rate": 4.293992932862191e-05,149      "loss": 1.3125,150      "step": 1000151    },152    {153      "epoch": 0.7420494699646644,154      "grad_norm": 141920.890625,155      "learning_rate": 4.258657243816255e-05,156      "loss": 1.2784,157      "step": 1050158    },159    {160      "epoch": 0.7773851590106007,161      "grad_norm": 155978.375,162      "learning_rate": 4.223321554770318e-05,163      "loss": 1.307,164      "step": 1100165    },166    {167      "epoch": 0.8127208480565371,168      "grad_norm": 151003.53125,169      "learning_rate": 4.187985865724382e-05,170      "loss": 1.2263,171      "step": 1150172    },173    {174      "epoch": 0.8480565371024735,175      "grad_norm": 122458.796875,176      "learning_rate": 4.1526501766784455e-05,177      "loss": 1.1795,178      "step": 1200179    },180    {181      "epoch": 0.8833922261484098,182      "grad_norm": 137286.90625,183      "learning_rate": 4.1173144876325096e-05,184      "loss": 1.2285,185      "step": 1250186    },187    {188      "epoch": 0.9187279151943463,189      "grad_norm": 112117.640625,190      "learning_rate": 4.081978798586573e-05,191      "loss": 1.2263,192      "step": 1300193    },194    {195      "epoch": 0.9540636042402827,196      "grad_norm": 127897.46875,197      "learning_rate": 4.0466431095406356e-05,198      "loss": 1.1636,199      "step": 1350200    },201    {202      "epoch": 0.9893992932862191,203      "grad_norm": 121189.921875,204      "learning_rate": 4.0113074204947e-05,205      "loss": 1.1762,206      "step": 1400207    },208    {209      "epoch": 1.0,210      "eval_bertscore_f1": 0.9435882283983255,211      "eval_bleu": 0.34081729524428495,212      "eval_loss": 0.8283334970474243,213      "eval_meteor": 0.5505336031382576,214      "eval_rouge1": 0.741807000391657,215      "eval_rouge2": 0.561792451504467,216      "eval_runtime": 458.3629,217      "eval_samples_per_second": 14.102,218      "eval_steps_per_second": 0.881,219      "step": 1415220    },221    {222      "epoch": 1.0247349823321554,223      "grad_norm": 128025.5703125,224      "learning_rate": 3.975971731448763e-05,225      "loss": 1.1689,226      "step": 1450227    },228    {229      "epoch": 1.0600706713780919,230      "grad_norm": 117022.6953125,231      "learning_rate": 3.940636042402827e-05,232      "loss": 1.1721,233      "step": 1500234    },235    {236      "epoch": 1.0954063604240283,237      "grad_norm": 127291.71875,238      "learning_rate": 3.9053003533568905e-05,239      "loss": 1.1144,240      "step": 1550241    },242    {243      "epoch": 1.1307420494699647,244      "grad_norm": 432520.5,245      "learning_rate": 3.8699646643109545e-05,246      "loss": 1.1173,247      "step": 1600248    },249    {250      "epoch": 1.1660777385159011,251      "grad_norm": 133514.078125,252      "learning_rate": 3.834628975265018e-05,253      "loss": 1.1349,254      "step": 1650255    },256    {257      "epoch": 1.2014134275618376,258      "grad_norm": 196108.515625,259      "learning_rate": 3.799293286219082e-05,260      "loss": 1.1105,261      "step": 1700262    },263    {264      "epoch": 1.2367491166077738,265      "grad_norm": 141071.890625,266      "learning_rate": 3.763957597173145e-05,267      "loss": 1.0843,268      "step": 1750269    },270    {271      "epoch": 1.2720848056537102,272      "grad_norm": 132118.53125,273      "learning_rate": 3.728621908127209e-05,274      "loss": 1.1424,275      "step": 1800276    },277    {278      "epoch": 1.3074204946996466,279      "grad_norm": 160950.84375,280      "learning_rate": 3.693286219081272e-05,281      "loss": 1.1462,282      "step": 1850283    },284    {285      "epoch": 1.342756183745583,286      "grad_norm": 128804.078125,287      "learning_rate": 3.6579505300353355e-05,288      "loss": 1.0572,289      "step": 1900290    },291    {292      "epoch": 1.3780918727915195,293      "grad_norm": 157611.953125,294      "learning_rate": 3.6226148409893995e-05,295      "loss": 1.0987,296      "step": 1950297    },298    {299      "epoch": 1.4134275618374559,300      "grad_norm": 141798.90625,301      "learning_rate": 3.587279151943463e-05,302      "loss": 1.0541,303      "step": 2000304    },305    {306      "epoch": 1.4487632508833923,307      "grad_norm": 171915.71875,308      "learning_rate": 3.551943462897527e-05,309      "loss": 1.0487,310      "step": 2050311    },312    {313      "epoch": 1.4840989399293285,314      "grad_norm": 127521.7109375,315      "learning_rate": 3.51660777385159e-05,316      "loss": 0.9808,317      "step": 2100318    },319    {320      "epoch": 1.5194346289752652,321      "grad_norm": 144515.828125,322      "learning_rate": 3.4812720848056544e-05,323      "loss": 0.9844,324      "step": 2150325    },326    {327      "epoch": 1.5547703180212014,328      "grad_norm": 135642.578125,329      "learning_rate": 3.445936395759717e-05,330      "loss": 0.9523,331      "step": 2200332    },333    {334      "epoch": 1.5901060070671378,335      "grad_norm": 211087.34375,336      "learning_rate": 3.410600706713781e-05,337      "loss": 0.9904,338      "step": 2250339    },340    {341      "epoch": 1.6254416961130742,342      "grad_norm": 157448.421875,343      "learning_rate": 3.3752650176678445e-05,344      "loss": 1.0143,345      "step": 2300346    },347    {348      "epoch": 1.6607773851590106,349      "grad_norm": 185968.609375,350      "learning_rate": 3.3399293286219085e-05,351      "loss": 1.0191,352      "step": 2350353    },354    {355      "epoch": 1.696113074204947,356      "grad_norm": 138438.078125,357      "learning_rate": 3.304593639575972e-05,358      "loss": 1.0524,359      "step": 2400360    },361    {362      "epoch": 1.7314487632508833,363      "grad_norm": 151158.640625,364      "learning_rate": 3.269257950530035e-05,365      "loss": 0.9748,366      "step": 2450367    },368    {369      "epoch": 1.76678445229682,370      "grad_norm": 146153.25,371      "learning_rate": 3.233922261484099e-05,372      "loss": 0.9595,373      "step": 2500374    },375    {376      "epoch": 1.802120141342756,377      "grad_norm": 145571.171875,378      "learning_rate": 3.198586572438163e-05,379      "loss": 0.9884,380      "step": 2550381    },382    {383      "epoch": 1.8374558303886925,384      "grad_norm": 200024.28125,385      "learning_rate": 3.163250883392227e-05,386      "loss": 0.9961,387      "step": 2600388    },389    {390      "epoch": 1.872791519434629,391      "grad_norm": 1255579.375,392      "learning_rate": 3.1279151943462894e-05,393      "loss": 0.9381,394      "step": 2650395    },396    {397      "epoch": 1.9081272084805654,398      "grad_norm": 191306.1875,399      "learning_rate": 3.0925795053003535e-05,400      "loss": 0.9179,401      "step": 2700402    },403    {404      "epoch": 1.9434628975265018,405      "grad_norm": 139976.265625,406      "learning_rate": 3.057243816254417e-05,407      "loss": 0.9741,408      "step": 2750409    },410    {411      "epoch": 1.978798586572438,412      "grad_norm": 254590.484375,413      "learning_rate": 3.021908127208481e-05,414      "loss": 0.97,415      "step": 2800416    },417    {418      "epoch": 2.0,419      "eval_bertscore_f1": 0.9522204981784862,420      "eval_bleu": 0.4391766527339286,421      "eval_loss": 0.6113855838775635,422      "eval_meteor": 0.6141492660226205,423      "eval_rouge1": 0.7802485332799334,424      "eval_rouge2": 0.6456575876833133,425      "eval_runtime": 442.4344,426      "eval_samples_per_second": 14.61,427      "eval_steps_per_second": 0.913,428      "step": 2830429    },430    {431      "epoch": 2.0141342756183747,432      "grad_norm": 251050.734375,433      "learning_rate": 2.9865724381625443e-05,434      "loss": 0.9362,435      "step": 2850436    },437    {438      "epoch": 2.049469964664311,439      "grad_norm": 165134.1875,440      "learning_rate": 2.9512367491166083e-05,441      "loss": 0.9642,442      "step": 2900443    },444    {445      "epoch": 2.0848056537102475,446      "grad_norm": 167977.453125,447      "learning_rate": 2.9159010600706717e-05,448      "loss": 0.899,449      "step": 2950450    },451    {452      "epoch": 2.1201413427561837,453      "grad_norm": 137508.921875,454      "learning_rate": 2.8805653710247347e-05,455      "loss": 0.8828,456      "step": 3000457    },458    {459      "epoch": 2.1554770318021204,460      "grad_norm": 155117.59375,461      "learning_rate": 2.8452296819787988e-05,462      "loss": 0.8958,463      "step": 3050464    },465    {466      "epoch": 2.1908127208480566,467      "grad_norm": 155701.109375,468      "learning_rate": 2.809893992932862e-05,469      "loss": 0.9106,470      "step": 3100471    },472    {473      "epoch": 2.2261484098939928,474      "grad_norm": 148495.125,475      "learning_rate": 2.7745583038869262e-05,476      "loss": 0.9755,477      "step": 3150478    },479    {480      "epoch": 2.2614840989399294,481      "grad_norm": 179815.203125,482      "learning_rate": 2.7392226148409892e-05,483      "loss": 0.9842,484      "step": 3200485    },486    {487      "epoch": 2.2968197879858656,488      "grad_norm": 193006.0625,489      "learning_rate": 2.7038869257950533e-05,490      "loss": 0.8794,491      "step": 3250492    },493    {494      "epoch": 2.3321554770318023,495      "grad_norm": 181638.421875,496      "learning_rate": 2.6685512367491167e-05,497      "loss": 0.9415,498      "step": 3300499    },500    {501      "epoch": 2.3674911660777385,502      "grad_norm": 172265.5,503      "learning_rate": 2.6332155477031807e-05,504      "loss": 0.9453,505      "step": 3350506    },507    {508      "epoch": 2.402826855123675,509      "grad_norm": 142922.71875,510      "learning_rate": 2.597879858657244e-05,511      "loss": 0.8669,512      "step": 3400513    },514    {515      "epoch": 2.4381625441696113,516      "grad_norm": 153024.515625,517      "learning_rate": 2.562544169611307e-05,518      "loss": 0.9168,519      "step": 3450520    },521    {522      "epoch": 2.4734982332155475,523      "grad_norm": 325320.28125,524      "learning_rate": 2.5272084805653712e-05,525      "loss": 0.8523,526      "step": 3500527    },528    {529      "epoch": 2.508833922261484,530      "grad_norm": 180908.15625,531      "learning_rate": 2.491872791519435e-05,532      "loss": 0.8853,533      "step": 3550534    },535    {536      "epoch": 2.5441696113074204,537      "grad_norm": 171192.765625,538      "learning_rate": 2.4565371024734986e-05,539      "loss": 0.9041,540      "step": 3600541    },542    {543      "epoch": 2.579505300353357,544      "grad_norm": 148674.578125,545      "learning_rate": 2.421201413427562e-05,546      "loss": 0.865,547      "step": 3650548    },549    {550      "epoch": 2.614840989399293,551      "grad_norm": 189248.390625,552      "learning_rate": 2.3858657243816253e-05,553      "loss": 0.9255,554      "step": 3700555    },556    {557      "epoch": 2.65017667844523,558      "grad_norm": 159843.796875,559      "learning_rate": 2.350530035335689e-05,560      "loss": 0.8886,561      "step": 3750562    },563    {564      "epoch": 2.685512367491166,565      "grad_norm": 162496.359375,566      "learning_rate": 2.3151943462897528e-05,567      "loss": 0.8689,568      "step": 3800569    },570    {571      "epoch": 2.7208480565371023,572      "grad_norm": 135053.5625,573      "learning_rate": 2.2798586572438165e-05,574      "loss": 0.8275,575      "step": 3850576    },577    {578      "epoch": 2.756183745583039,579      "grad_norm": 249403.953125,580      "learning_rate": 2.24452296819788e-05,581      "loss": 0.8319,582      "step": 3900583    },584    {585      "epoch": 2.791519434628975,586      "grad_norm": 181885.328125,587      "learning_rate": 2.2091872791519436e-05,588      "loss": 0.8719,589      "step": 3950590    },591    {592      "epoch": 2.8268551236749118,593      "grad_norm": 156204.75,594      "learning_rate": 2.1738515901060073e-05,595      "loss": 0.8592,596      "step": 4000597    },598    {599      "epoch": 2.862190812720848,600      "grad_norm": 159614.03125,601      "learning_rate": 2.138515901060071e-05,602      "loss": 0.8381,603      "step": 4050604    },605    {606      "epoch": 2.8975265017667846,607      "grad_norm": 162379.78125,608      "learning_rate": 2.1031802120141344e-05,609      "loss": 0.8612,610      "step": 4100611    },612    {613      "epoch": 2.932862190812721,614      "grad_norm": 149608.25,615      "learning_rate": 2.067844522968198e-05,616      "loss": 0.838,617      "step": 4150618    },619    {620      "epoch": 2.968197879858657,621      "grad_norm": 145233.40625,622      "learning_rate": 2.0325088339222618e-05,623      "loss": 0.8411,624      "step": 4200625    },626    {627      "epoch": 3.0,628      "eval_bertscore_f1": 0.955309976823628,629      "eval_bleu": 0.47422911395097656,630      "eval_loss": 0.5273156762123108,631      "eval_meteor": 0.6439931966505934,632      "eval_rouge1": 0.7960674625670938,633      "eval_rouge2": 0.6807866947718111,634      "eval_runtime": 440.5737,635      "eval_samples_per_second": 14.672,636      "eval_steps_per_second": 0.917,637      "step": 4245638    },639    {640      "epoch": 3.0035335689045937,641      "grad_norm": 146904.109375,642      "learning_rate": 1.997173144876325e-05,643      "loss": 0.8212,644      "step": 4250645    },646    {647      "epoch": 3.03886925795053,648      "grad_norm": 141972.140625,649      "learning_rate": 1.9618374558303885e-05,650      "loss": 0.8332,651      "step": 4300652    },653    {654      "epoch": 3.0742049469964665,655      "grad_norm": 155438.34375,656      "learning_rate": 1.9265017667844522e-05,657      "loss": 0.8314,658      "step": 4350659    },660    {661      "epoch": 3.1095406360424027,662      "grad_norm": 157805.5,663      "learning_rate": 1.891166077738516e-05,664      "loss": 0.8432,665      "step": 4400666    },667    {668      "epoch": 3.1448763250883394,669      "grad_norm": 171890.296875,670      "learning_rate": 1.8558303886925797e-05,671      "loss": 0.8171,672      "step": 4450673    },674    {675      "epoch": 3.1802120141342756,676      "grad_norm": 159529.09375,677      "learning_rate": 1.8204946996466434e-05,678      "loss": 0.8462,679      "step": 4500680    },681    {682      "epoch": 3.215547703180212,683      "grad_norm": 291361.3125,684      "learning_rate": 1.7851590106007067e-05,685      "loss": 0.8525,686      "step": 4550687    },688    {689      "epoch": 3.2508833922261484,690      "grad_norm": 146320.515625,691      "learning_rate": 1.7498233215547705e-05,692      "loss": 0.8691,693      "step": 4600694    },695    {696      "epoch": 3.2862190812720846,697      "grad_norm": 142328.234375,698      "learning_rate": 1.714487632508834e-05,699      "loss": 0.8152,700      "step": 4650701    },702    {703      "epoch": 3.3215547703180213,704      "grad_norm": 164441.828125,705      "learning_rate": 1.679151943462898e-05,706      "loss": 0.7897,707      "step": 4700708    },709    {710      "epoch": 3.3568904593639575,711      "grad_norm": 181799.328125,712      "learning_rate": 1.643816254416961e-05,713      "loss": 0.89,714      "step": 4750715    },716    {717      "epoch": 3.392226148409894,718      "grad_norm": 161868.921875,719      "learning_rate": 1.6084805653710246e-05,720      "loss": 0.8342,721      "step": 4800722    },723    {724      "epoch": 3.4275618374558303,725      "grad_norm": 168403.046875,726      "learning_rate": 1.5731448763250883e-05,727      "loss": 0.7708,728      "step": 4850729    },730    {731      "epoch": 3.462897526501767,732      "grad_norm": 180798.71875,733      "learning_rate": 1.537809187279152e-05,734      "loss": 0.8452,735      "step": 4900736    },737    {738      "epoch": 3.498233215547703,739      "grad_norm": 194585.109375,740      "learning_rate": 1.5024734982332156e-05,741      "loss": 0.8543,742      "step": 4950743    },744    {745      "epoch": 3.53356890459364,746      "grad_norm": 162353.515625,747      "learning_rate": 1.4671378091872793e-05,748      "loss": 0.7865,749      "step": 5000750    },751    {752      "epoch": 3.568904593639576,753      "grad_norm": 161337.140625,754      "learning_rate": 1.4318021201413428e-05,755      "loss": 0.8459,756      "step": 5050757    },758    {759      "epoch": 3.604240282685512,760      "grad_norm": 174313.953125,761      "learning_rate": 1.3964664310954066e-05,762      "loss": 0.8914,763      "step": 5100764    },765    {766      "epoch": 3.639575971731449,767      "grad_norm": 163668.421875,768      "learning_rate": 1.3611307420494701e-05,769      "loss": 0.8457,770      "step": 5150771    },772    {773      "epoch": 3.674911660777385,774      "grad_norm": 164355.765625,775      "learning_rate": 1.3257950530035338e-05,776      "loss": 0.8266,777      "step": 5200778    },779    {780      "epoch": 3.7102473498233217,781      "grad_norm": 159919.046875,782      "learning_rate": 1.2904593639575973e-05,783      "loss": 0.7511,784      "step": 5250785    },786    {787      "epoch": 3.745583038869258,788      "grad_norm": 131121.03125,789      "learning_rate": 1.2551236749116607e-05,790      "loss": 0.8333,791      "step": 5300792    },793    {794      "epoch": 3.7809187279151946,795      "grad_norm": 160561.640625,796      "learning_rate": 1.2197879858657244e-05,797      "loss": 0.7992,798      "step": 5350799    },800    {801      "epoch": 3.8162544169611308,802      "grad_norm": 200662.859375,803      "learning_rate": 1.184452296819788e-05,804      "loss": 0.8494,805      "step": 5400806    },807    {808      "epoch": 3.851590106007067,809      "grad_norm": 158871.84375,810      "learning_rate": 1.1491166077738517e-05,811      "loss": 0.7689,812      "step": 5450813    },814    {815      "epoch": 3.8869257950530036,816      "grad_norm": 233291.453125,817      "learning_rate": 1.1137809187279152e-05,818      "loss": 0.7717,819      "step": 5500820    },821    {822      "epoch": 3.92226148409894,823      "grad_norm": 152039.84375,824      "learning_rate": 1.078445229681979e-05,825      "loss": 0.7991,826      "step": 5550827    },828    {829      "epoch": 3.9575971731448765,830      "grad_norm": 174898.390625,831      "learning_rate": 1.0431095406360425e-05,832      "loss": 0.7713,833      "step": 5600834    },835    {836      "epoch": 3.9929328621908127,837      "grad_norm": 185653.25,838      "learning_rate": 1.0077738515901062e-05,839      "loss": 0.8271,840      "step": 5650841    },842    {843      "epoch": 4.0,844      "eval_bertscore_f1": 0.9574022400700072,845      "eval_bleu": 0.49288907686192807,846      "eval_loss": 0.4886058270931244,847      "eval_meteor": 0.6607653874114606,848      "eval_rouge1": 0.8043075620747033,849      "eval_rouge2": 0.698973717724064,850      "eval_runtime": 437.782,851      "eval_samples_per_second": 14.765,852      "eval_steps_per_second": 0.923,853      "step": 5660854    },855    {856      "epoch": 4.028268551236749,857      "grad_norm": 189546.390625,858      "learning_rate": 9.724381625441696e-06,859      "loss": 0.8009,860      "step": 5700861    },862    {863      "epoch": 4.063604240282685,864      "grad_norm": 179902.28125,865      "learning_rate": 9.371024734982333e-06,866      "loss": 0.7955,867      "step": 5750868    },869    {870      "epoch": 4.098939929328622,871      "grad_norm": 134316.078125,872      "learning_rate": 9.017667844522968e-06,873      "loss": 0.8093,874      "step": 5800875    },876    {877      "epoch": 4.134275618374558,878      "grad_norm": 113333.8203125,879      "learning_rate": 8.664310954063605e-06,880      "loss": 0.7959,881      "step": 5850882    },883    {884      "epoch": 4.169611307420495,885      "grad_norm": 153999.46875,886      "learning_rate": 8.31095406360424e-06,887      "loss": 0.7883,888      "step": 5900889    },890    {891      "epoch": 4.204946996466431,892      "grad_norm": 284284.375,893      "learning_rate": 7.957597173144876e-06,894      "loss": 0.7858,895      "step": 5950896    },897    {898      "epoch": 4.240282685512367,899      "grad_norm": 139182.28125,900      "learning_rate": 7.6042402826855124e-06,901      "loss": 0.804,902      "step": 6000903    },904    {905      "epoch": 4.275618374558304,906      "grad_norm": 169185.328125,907      "learning_rate": 7.250883392226149e-06,908      "loss": 0.7745,909      "step": 6050910    },911    {912      "epoch": 4.310954063604241,913      "grad_norm": 200688.453125,914      "learning_rate": 6.897526501766785e-06,915      "loss": 0.7795,916      "step": 6100917    },918    {919      "epoch": 4.3462897526501765,920      "grad_norm": 135357.96875,921      "learning_rate": 6.544169611307421e-06,922      "loss": 0.8412,923      "step": 6150924    },925    {926      "epoch": 4.381625441696113,927      "grad_norm": 137914.25,928      "learning_rate": 6.190812720848057e-06,929      "loss": 0.811,930      "step": 6200931    },932    {933      "epoch": 4.41696113074205,934      "grad_norm": 179002.140625,935      "learning_rate": 5.837455830388693e-06,936      "loss": 0.7859,937      "step": 6250938    },939    {940      "epoch": 4.4522968197879855,941      "grad_norm": 139846.9375,942      "learning_rate": 5.484098939929329e-06,943      "loss": 0.801,944      "step": 6300945    },946    {947      "epoch": 4.487632508833922,948      "grad_norm": 161350.671875,949      "learning_rate": 5.1307420494699655e-06,950      "loss": 0.8203,951      "step": 6350952    },953    {954      "epoch": 4.522968197879859,955      "grad_norm": 153958.46875,956      "learning_rate": 4.777385159010601e-06,957      "loss": 0.7693,958      "step": 6400959    },960    {961      "epoch": 4.5583038869257955,962      "grad_norm": 154714.984375,963      "learning_rate": 4.424028268551237e-06,964      "loss": 0.7808,965      "step": 6450966    },967    {968      "epoch": 4.593639575971731,969      "grad_norm": 294880.1875,970      "learning_rate": 4.070671378091873e-06,971      "loss": 0.8194,972      "step": 6500973    },974    {975      "epoch": 4.628975265017668,976      "grad_norm": 480907.34375,977      "learning_rate": 3.717314487632509e-06,978      "loss": 0.8096,979      "step": 6550980    },981    {982      "epoch": 4.6643109540636045,983      "grad_norm": 319982.71875,984      "learning_rate": 3.363957597173145e-06,985      "loss": 0.7987,986      "step": 6600987    },988    {989      "epoch": 4.69964664310954,990      "grad_norm": 188937.359375,991      "learning_rate": 3.010600706713781e-06,992      "loss": 0.8162,993      "step": 6650994    },995    {996      "epoch": 4.734982332155477,997      "grad_norm": 166750.71875,998      "learning_rate": 2.6572438162544172e-06,999      "loss": 0.7726,1000      "step": 67001001    },1002    {1003      "epoch": 4.770318021201414,1004      "grad_norm": 179732.5625,1005      "learning_rate": 2.303886925795053e-06,1006      "loss": 0.7848,1007      "step": 67501008    },1009    {1010      "epoch": 4.80565371024735,1011      "grad_norm": 135768.640625,1012      "learning_rate": 1.9505300353356893e-06,1013      "loss": 0.8204,1014      "step": 68001015    },1016    {1017      "epoch": 4.840989399293286,1018      "grad_norm": 194958.0625,1019      "learning_rate": 1.5971731448763252e-06,1020      "loss": 0.8214,1021      "step": 68501022    },1023    {1024      "epoch": 4.876325088339223,1025      "grad_norm": 205979.40625,1026      "learning_rate": 1.2438162544169612e-06,1027      "loss": 0.7658,1028      "step": 69001029    },1030    {1031      "epoch": 4.911660777385159,1032      "grad_norm": 140958.078125,1033      "learning_rate": 8.904593639575972e-07,1034      "loss": 0.777,1035      "step": 69501036    },1037    {1038      "epoch": 4.946996466431095,1039      "grad_norm": 149027.953125,1040      "learning_rate": 5.371024734982332e-07,1041      "loss": 0.8286,1042      "step": 70001043    },1044    {1045      "epoch": 4.982332155477032,1046      "grad_norm": 161736.4375,1047      "learning_rate": 1.8374558303886926e-07,1048      "loss": 0.7922,1049      "step": 70501050    },1051    {1052      "epoch": 5.0,1053      "eval_bertscore_f1": 0.957730518736978,1054      "eval_bleu": 0.49562010024897835,1055      "eval_loss": 0.47903409600257874,1056      "eval_meteor": 0.6644886697409358,1057      "eval_rouge1": 0.8053692499714378,1058      "eval_rouge2": 0.7018033389816531,1059      "eval_runtime": 438.2906,1060      "eval_samples_per_second": 14.748,1061      "eval_steps_per_second": 0.922,1062      "step": 70751063    }1064  ],1065  "logging_steps": 50,1066  "max_steps": 7075,1067  "num_input_tokens_seen": 0,1068  "num_train_epochs": 5,1069  "save_steps": 500,1070  "stateful_callbacks": {1071    "TrainerControl": {1072      "args": {1073        "should_epoch_stop": false,1074        "should_evaluate": false,1075        "should_log": false,1076        "should_save": true,1077        "should_training_stop": true1078      },1079      "attributes": {}1080    }1081  },1082  "total_flos": 1.730048539557888e+16,1083  "train_batch_size": 16,1084  "trial_name": null,1085  "trial_params": null1086}1087