epoch 0
train loss: 0.10178063126471196
valid loss scaled: 0.08694887030887369
valid loss: 14.393254678046581
valid loss clipped: 14.393254678046581
train loss: 0.08234276634246003
valid loss scaled: 0.0941885423619173
valid loss: 15.591687502080195
valid loss clipped: 15.591687502080195
train loss: 0.07840566897624576
valid loss scaled: 0.10380323876951889
valid loss: 17.18327815406788
valid loss clipped: 17.18327815406788
train loss: 0.07422883822886046
valid loss scaled: 0.10199038650759494
valid loss: 16.883180520434045
valid loss clipped: 16.883180520434045
train loss: 0.0715406045230223
valid loss scaled: 0.08943446505286708
valid loss: 14.8047131664572
valid loss clipped: 14.8047131664572
train loss: 0.07021876769740436
valid loss scaled: 0.0936252853835668
valid loss: 15.4984476046049
valid loss clipped: 15.498044701389192
train loss: 0.0698000478439966
valid loss scaled: 0.0732450713218563
valid loss: 12.124766595640176
valid loss clipped: 12.1245106015308
train loss: 0.06787076175194112
valid loss scaled: 0.08167600706479376
valid loss: 13.520399589246514
valid loss clipped: 13.520399589246514
train loss: 0.06687028820999129
valid loss scaled: 0.07860967955406604
valid loss: 13.012812666808495
valid loss clipped: 13.012812666808495
train loss: 0.06485297715840126
valid loss scaled: 0.06548883445499123
valid loss: 10.840826069687282
valid loss clipped: 10.840826069687282
train loss: 0.06536078446808377
valid loss scaled: 0.06285778998197525
valid loss: 10.405289848890389
valid loss clipped: 10.402398559859241
train loss: 0.06282014927278873
valid loss scaled: 0.0596510515677964
valid loss: 9.874458507430498
valid loss clipped: 9.87346211871712
train loss: 0.06211401904304488
valid loss scaled: 0.08676563866997164
valid loss: 14.362925503357548
valid loss clipped: 14.362736109627358
train loss: 0.06245454523502525
valid loss scaled: 0.061674062543468834
valid loss: 10.209341911143197
valid loss clipped: 10.209341911143197
train loss: 0.06125578036664975
valid loss scaled: 0.07651956140507003
valid loss: 12.666819321757812
valid loss clipped: 12.665205579153804
train loss: 0.059952159220873956
valid loss scaled: 0.060298738548431624
valid loss: 9.981674467979698
valid loss clipped: 9.981674467979698
train loss: 0.058864656505384534
valid loss scaled: 0.06224854018969625
valid loss: 10.304438001377102
valid loss clipped: 10.304438001377102
train loss: 0.06033644216377098
valid loss scaled: 0.06860725018131864
valid loss: 11.357033358430243
valid loss clipped: 11.355784159072005
train loss: 0.05914900476107049
valid loss scaled: 0.06558627462957999
valid loss: 10.856954814387644
valid loss clipped: 10.855340320357298
train loss: 0.05801354450611692
valid loss scaled: 0.05499361709851459
valid loss: 9.103479951662178
valid loss clipped: 9.1033016162492
train loss: 0.05722296967684818
valid loss scaled: 0.057276662331433935
valid loss: 9.481407818900754
valid loss clipped: 9.481123608784491
train loss: 0.05615422017549816
valid loss scaled: 0.058946735056267775
valid loss: 9.757867643250142
valid loss clipped: 9.757867643250142
train loss: 0.0568488666286758
valid loss scaled: 0.05493375345527444
valid loss: 9.093570558704585
valid loss clipped: 9.093570558704585
train loss: 0.056790975754785165
valid loss scaled: 0.05718280344549578
valid loss: 9.465868729560718
valid loss clipped: 9.465666290138364
train loss: 0.056136715463394736
valid loss scaled: 0.053061117528704765
valid loss: 8.783576806818123
valid loss clipped: 8.783576806818123
train loss: 0.0553076835671036
valid loss scaled: 0.05756376005144751
valid loss: 9.528932262971411
valid loss clipped: 9.528929186851473
train loss: 0.05608131737157289
valid loss scaled: 0.05928617719392673
valid loss: 9.814053006356623
valid loss clipped: 9.813086909431942
train loss: 0.054209274725265544
valid loss scaled: 0.059807781220226314
valid loss: 9.900402019838845
valid loss clipped: 9.899803739726012
train loss: 0.05412653998906017
valid loss scaled: 0.05289933204229587
valid loss: 8.75679344111035
valid loss clipped: 8.75679344111035
train loss: 0.05518407843528995
valid loss scaled: 0.05688392864725153
valid loss: 9.416393057892193
valid loss clipped: 9.416393057892193
train loss: 0.054482056935617085
valid loss scaled: 0.05216977017299728
valid loss: 8.636023503699716
valid loss clipped: 8.636023503699716
train loss: 0.05399237402441963
valid loss scaled: 0.05409103409701342
valid loss: 8.954068173354056
valid loss clipped: 8.954068173354056
train loss: 0.05278416502551686
valid loss scaled: 0.05505897657177228
valid loss: 9.114298554400836
valid loss clipped: 9.114128444401343
train loss: 0.05283459274043561
valid loss scaled: 0.05330468837007633
valid loss: 8.823897722342918
valid loss clipped: 8.823469673021748
train loss: 0.05349978882139668
valid loss scaled: 0.05340672519610164
valid loss: 8.840791683139289
valid loss clipped: 8.840791683139289
train loss: 0.05244616329396176
valid loss scaled: 0.05313163808650701
valid loss: 8.795251288181513
valid loss clipped: 8.795251288181513
train loss: 0.05458822252512801
valid loss scaled: 0.05058266187593807
valid loss: 8.373303749237403
valid loss clipped: 8.372144978003309
train loss: 0.052178348999718266
valid loss scaled: 0.05477028984676988
valid loss: 9.066509788893315
valid loss clipped: 9.066509788893315
train loss: 0.05264708004419216
valid loss scaled: 0.05061255628492586
valid loss: 8.378252333491037
valid loss clipped: 8.378252333491037
epoch 0 done, EVALUATION ON FULL DEV:
valid loss scaled: 0.05304482890568026
valid loss: 8.780882018378279
valid loss clipped: 8.780282676016194
evaluation done
epoch 1
train loss: 0.05193906537483793
valid loss scaled: 0.05525216534943099
valid loss: 9.146276710629154
valid loss clipped: 9.146276710629154
train loss: 0.05052301682721696
valid loss scaled: 0.05013342859224921
valid loss: 8.298936329028248
valid loss clipped: 8.298936329028248
train loss: 0.05098356105392436
valid loss scaled: 0.05003979649887808
valid loss: 8.283437402089449
valid loss clipped: 8.283437402089449
train loss: 0.050088782917609716
valid loss scaled: 0.049286247248487895
valid loss: 8.15869774398909
valid loss clipped: 8.15869774398909
train loss: 0.04976562104718302
valid loss scaled: 0.049223300781573644
valid loss: 8.148276889380307
valid loss clipped: 8.148276889380307
train loss: 0.04991738679144195
valid loss scaled: 0.057702729077406346
valid loss: 9.551934437355772
valid loss clipped: 9.550634254813586
train loss: 0.049231458019302426
valid loss scaled: 0.04882268479584859
valid loss: 8.081957979949655
valid loss clipped: 8.081957979949655
train loss: 0.04899937393720371
valid loss scaled: 0.05770907825550645
valid loss: 9.552989047406287
valid loss clipped: 9.552741958522342
train loss: 0.04908777283465607
valid loss scaled: 0.05134136848123586
valid loss: 8.498893804182858
valid loss clipped: 8.49871165029239
train loss: 0.04764838316443966
valid loss scaled: 0.05261156122562749
valid loss: 8.709157880799538
valid loss clipped: 8.709138480005972
train loss: 0.048582658014159445
valid loss scaled: 0.058413308832729716
valid loss: 9.669566298033944
valid loss clipped: 9.668254240676646
train loss: 0.04743617173694683
valid loss scaled: 0.051687093210397744
valid loss: 8.556127664837945
valid loss clipped: 8.556127664837945
train loss: 0.04656181514429912
valid loss scaled: 0.0507906026516647
valid loss: 8.407722978198628
valid loss clipped: 8.407722978198628
train loss: 0.04762343910575039
valid loss scaled: 0.04810272429882445
valid loss: 7.962782613217737
valid loss clipped: 7.962782613217737
train loss: 0.048021365618576804
valid loss scaled: 0.051694526258934385
valid loss: 8.55735450556255
valid loss clipped: 8.55735450556255
train loss: 0.04710475399633604
valid loss scaled: 0.04621696889011772
valid loss: 7.650616536714813
valid loss clipped: 7.650616536714813
train loss: 0.04589946658188546
valid loss scaled: 0.04597697057310632
valid loss: 7.610890791241026
valid loss clipped: 7.610890791241026
train loss: 0.047086027235787566
valid loss scaled: 0.04816656697145074
valid loss: 7.973350046367094
valid loss clipped: 7.973350046367094
train loss: 0.04701215419012016