different runs of linear svm

output.txt

lr,reg: 0.000000 25000.000000
iteration 0 / 1000: loss 777.803914
iteration 100 / 1000: loss 283.905719
iteration 200 / 1000: loss 106.902133
iteration 300 / 1000: loss 41.978504
iteration 400 / 1000: loss 18.960358
iteration 500 / 1000: loss 9.918861
iteration 600 / 1000: loss 7.038409
iteration 700 / 1000: loss 5.862711
iteration 800 / 1000: loss 5.861156
iteration 900 / 1000: loss 4.950127
lr,reg: 0.000000 50000.000000
iteration 0 / 1000: loss 1576.391614
iteration 100 / 1000: loss 213.573595
iteration 200 / 1000: loss 33.410996
iteration 300 / 1000: loss 8.803453
iteration 400 / 1000: loss 6.376496
iteration 500 / 1000: loss 6.014604
iteration 600 / 1000: loss 5.697175
iteration 700 / 1000: loss 6.027580
iteration 800 / 1000: loss 6.380398
iteration 900 / 1000: loss 5.557037
lr,reg: 0.000000 75000.000000
iteration 0 / 1000: loss 2302.192696
iteration 100 / 1000: loss 115.567413
iteration 200 / 1000: loss 11.220674
iteration 300 / 1000: loss 6.122058
iteration 400 / 1000: loss 6.105310
iteration 500 / 1000: loss 5.393758
iteration 600 / 1000: loss 6.217518
iteration 700 / 1000: loss 6.282513
iteration 800 / 1000: loss 5.834316
iteration 900 / 1000: loss 5.699341
lr,reg: 0.000050 25000.000000
iteration 0 / 1000: loss 783.782094
iteration 100 / 1000: loss 385084372909859584717087001143760912384.000000
iteration 200 / 1000: loss 63651362731610826734156287764279972089541255963398671431902928096527908864.000000
iteration 300 / 1000: loss 10521060480788370788972691338183621805485937359691824725632075582840294365004305741378183083134231768587042816.000000
iteration 400 / 1000: loss 1739047035130233283857224214855694070162683605496794695664813627346084316787878516537529030680699512821157627777697312842194724252873755908374528.000000
iteration 500 / 1000: loss 287450547016400914879068088818483553391316915177104630442965325518091497191833639940561953712467590732051132457288900424796250638496515018010268595817027507210152525207453351018496.000000
iteration 600 / 1000: loss 47513273253037862394369298175007146978955906860961040679144655975297842748940284221394518040353801864036943877628984560650720253606762376929062799365211115331948388084964970815249358274425682512580561606672056844288.000000
iteration 700 / 1000: loss 7853563538666833376284751710154180999246403731001224771138121084840580895553258864834063202810415067371148302780505393460004597760620190649712938338212509010874353047710553514338824405239359333365138560744441498013329711481620619267553416068934402048.000000
iteration 800 / 1000: loss 1298131154370291438204713531188475463250459328798334127909120332957952421244968792560215719969747999689068693406722043624697162583697926961691226112019543708091611889912996856317524838983251466833632540832608405439020132774980921028912150558308162528595770673864426664580436387008872448.000000
/Users/samjoseph/Desktop/assignment1/cs231n/classifiers/linear_svm.py:104: RuntimeWarning: overflow encountered in double_scalars
  loss += reg * np.sum(W * W)
/Users/samjoseph/Desktop/assignment1/cs231n/classifiers/linear_svm.py:104: RuntimeWarning: overflow encountered in multiply
  loss += reg * np.sum(W * W)

iteration 900 / 1000: loss inf
lr,reg: 0.000050 50000.000000
iteration 0 / 1000: loss 1564.747763
iteration 100 / 1000: loss 4374447233693759452141017110462281607298357790049724541316732292421383387511371864860728706284363259667720824418716126019584.000000
iteration 200 / 1000: loss 11295915835903324874257049971873404516888286666930217798599181805590314669976657073195809150214131255773333754065148974808862454193476492990953140406718667773660569177460586474869962663920045192816926367818037153857546975338597546523668286275584.000000
iteration 300 / 1000: loss inf
iteration 400 / 1000: loss inf
iteration 500 / 1000: loss inf
iteration 600 / 1000: loss nan
iteration 700 / 1000: loss nan
iteration 800 / 1000: loss nan
iteration 900 / 1000: loss nan
lr,reg: 0.000050 75000.000000
iteration 0 / 1000: loss 2309.148320
iteration 100 / 1000: loss 8998917738452717014736850955540632047911044903922336687441255672476797643711862612429063541373484042630850787619875726338854221527878386303259530787892424782494302208.000000
iteration 200 / 1000: loss inf
iteration 300 / 1000: loss inf
iteration 400 / 1000: loss nan
iteration 500 / 1000: loss nan
iteration 600 / 1000: loss nan
iteration 700 / 1000: loss nan
iteration 800 / 1000: loss nan
iteration 900 / 1000: loss nan
lr,reg: 0.000100 25000.000000
iteration 0 / 1000: loss 791.041301
iteration 100 / 1000: loss 2635035150609258110460299157515991185416427862482001534596304989110929635044733863166443962655619829313577447021576747745280.000000
iteration 200 / 1000: loss 6804319196415475570039841210242538031963646190347797212880315256865788827704369595695228213427065043514974957075958410708031533572710531723575352758575143333112180873408170114638996461524523325636221391220603792410117723198870639307379735265280.000000
iteration 300 / 1000: loss inf
iteration 400 / 1000: loss inf
iteration 500 / 1000: loss inf
/Users/samjoseph/Desktop/assignment1/cs231n/classifiers/linear_svm.py:156: RuntimeWarning: overflow encountered in multiply
  dW += 2 * reg * W
/Users/samjoseph/Desktop/assignment1/cs231n/classifiers/linear_svm.py:96: RuntimeWarning: invalid value encountered in maximum
  margins = np.maximum(0, scores.T - correct_class_scores + delta) # => [10,500]
/Users/samjoseph/Desktop/assignment1/cs231n/classifiers/linear_svm.py:144: RuntimeWarning: invalid value encountered in greater
  margins_boolean = margins>0 # this makes margins_boolean of type bool
/Users/samjoseph/Desktop/assignment1/cs231n/classifiers/linear_classifier.py:79: RuntimeWarning: invalid value encountered in add
  self.W += - learning_rate * grad
/Users/samjoseph/Desktop/assignment1/cs231n/classifiers/linear_svm.py:96: RuntimeWarning: overflow encountered in subtract
  margins = np.maximum(0, scores.T - correct_class_scores + delta) # => [10,500]

iteration 600 / 1000: loss nan
iteration 700 / 1000: loss nan
iteration 800 / 1000: loss nan
iteration 900 / 1000: loss nan
lr,reg: 0.000100 50000.000000
iteration 0 / 1000: loss 1560.439934
iteration 100 / 1000: loss 115408796669660159643988993653019043035971899190424621552214070147189489207535817031052808863617431359186641488950484530996840608472908583593477838905767147612599581580208947231393962797553418240.000000
iteration 200 / 1000: loss inf
iteration 300 / 1000: loss inf
iteration 400 / 1000: loss nan
iteration 500 / 1000: loss nan
iteration 600 / 1000: loss nan
iteration 700 / 1000: loss nan
iteration 800 / 1000: loss nan
iteration 900 / 1000: loss nan
lr,reg: 0.000100 75000.000000
iteration 0 / 1000: loss 2320.133583
iteration 100 / 1000: loss 39432499218316571604612257864672826965763149021054229696075583940577556655246385822668374739650726089181881808716923256670805589278015530076464383064268742455241568025243843674079894204214917546685989692764185751072918917788267970560.000000
iteration 200 / 1000: loss inf
iteration 300 / 1000: loss nan
iteration 400 / 1000: loss nan
iteration 500 / 1000: loss nan
iteration 600 / 1000: loss nan
iteration 700 / 1000: loss nan
iteration 800 / 1000: loss nan
iteration 900 / 1000: loss nan
lr 1.000000e-07 reg 2.500000e+04 train accuracy: 0.372408 val accuracy: 0.377000
lr 1.000000e-07 reg 5.000000e+04 train accuracy: 0.360327 val accuracy: 0.374000
lr 1.000000e-07 reg 7.500000e+04 train accuracy: 0.349918 val accuracy: 0.357000
lr 5.000000e-05 reg 2.500000e+04 train accuracy: 0.048939 val accuracy: 0.047000
lr 5.000000e-05 reg 5.000000e+04 train accuracy: 0.100265 val accuracy: 0.087000
lr 5.000000e-05 reg 7.500000e+04 train accuracy: 0.100265 val accuracy: 0.087000
lr 1.000000e-04 reg 2.500000e+04 train accuracy: 0.100265 val accuracy: 0.087000
lr 1.000000e-04 reg 5.000000e+04 train accuracy: 0.100265 val accuracy: 0.087000
lr 1.000000e-04 reg 7.500000e+04 train accuracy: 0.100265 val accuracy: 0.087000
best validation accuracy achieved during cross-validation: 0.377000
/Users/samjoseph/Desktop/assignment1/cs231n/classifiers/linear_svm.py:96: RuntimeWarning: invalid value encountered in subtract
  margins = np.maximum(0, scores.T - correct_class_scores + delta) # => [10,500]