soumith/imagenet-multiGPU.torch

Issues

• Trained googlenet's outputs are negative value. Why negative?

  #106 opened 7 years ago by bemoregt
  1

• how to save my image data into torch format (.t7)

  #109 opened 6 years ago by zq1314
  3

• Can not continue training

  #110 opened 6 years ago by mhmtsarigul
  0

• Training stops while running main.lua , "nClasses is reported different in the data loader, and in the commandline options"

  #87 opened 8 years ago by DKP-90
  1

• just want to say "Thanks to Soumith" for this useful toolbox

  #108 opened 7 years ago by Ammarah-Farooq
  0

• Why not call cutorch.synchronizeAll()

  #107 opened 7 years ago by farleylai
  0

• Training time about Multi GPUs and Single GPU

  #97 opened 8 years ago by xieshuqin
  1

• vggbn training's validation accuracy is very low

  #63 opened 9 years ago by cxy7452
  2

• AlexNet Top-1 accuracy on test dataset only reach 45% when train with multiple GPUs

  #88 opened 7 years ago by yiheng
  3

• How can I retrain this googlenet network?

  #104 opened 7 years ago by bemoregt
  2

• I used just 2 class training. But test outputs are 4 values. Why?

  #105 opened 7 years ago by bemoregt
  0

• Retraining an existing model(No details)

  #103 opened 7 years ago by anmolbhatia28
  1

• Problem with Thread ???

  #102 opened 8 years ago by hnanhtuan
  0

• Data loading time is different across threads

  #101 opened 8 years ago by JiayunLi
  0

• "nn/Linear.lua:69: input must be vector or matrix"

  #100 opened 8 years ago by bemoregt
  0

• testing extremely slow

  #99 opened 8 years ago by varghesealex90
  0

• Data loading

  #98 opened 8 years ago by varghesealex90
  0

• SqueezeNet issue

  #92 opened 8 years ago by rracinskij
  2

• Class has zero samples

  #72 opened 8 years ago by vesvis
  3

• Could not find any image file in the given input paths

  #95 opened 8 years ago by yusufzaferaydin
  0

• Resume training fails - weights/gradients/input tensors on different GPUs

  #90 opened 8 years ago by jwjohnson314
  0

• nGPU=4 is slower than nGPU=1

  #49 opened 9 years ago by chienlinhuang1116
  14

• A question about predicting individual images

  #77 opened 8 years ago by jsongcse
  3

• Dropout in AlexNet

  #89 opened 8 years ago by joeyhng
  0

• Too much time per epoch

  #75 opened 8 years ago by Viresh-R
  6

• random loading data sequence

  #65 opened 9 years ago by zijunwei
  1

• Train Alexnet with 4 GPUs seems slower than one

  #59 opened 9 years ago by yiheng
  2

• cunn backend no longer supported

  #85 opened 8 years ago by williford
  0

• Vggbn

  #66 opened 8 years ago by sinaghs
  0

• syncParameters() bug without consequences

  #79 opened 8 years ago by tomsercu
  1

• Code does not resize images to 256x256

  #76 opened 8 years ago by sozubek
  1

• Googlenet with Torch has accuracy of 0 %

  #82 opened 8 years ago by antoajayraj
  0

• alexnetowtbn runs exit with illegal memory access error

  #81 opened 8 years ago by antoajayraj
  0

• low accuracy on alexnetowtbn

  #55 opened 9 years ago by mrastegari
  12

• Top-1 error of 45.07

  #80 opened 8 years ago by Viresh-R
  0

• attempt to call global 'unpack' (a nil value)

  #78 opened 8 years ago by karandwivedi42
  1

• training on multiple GPUs gives nan

  #74 opened 8 years ago by arashno
  1

• Which Class Labels To Indices In Output Tensor?

  #73 opened 8 years ago by hsoule
  4

• Alexnet code for fbcunn

  #71 opened 9 years ago by antoajayraj
  5

• Newbie question: Is my model making any progress

  #70 opened 9 years ago by livenletdie
  2

• Error attempt to index field 'THNN' (a nil value)

  #69 opened 9 years ago by mlagunas
  1

• Torch.save on dataset class error with gm package

  #64 opened 9 years ago by vrkrishn
  1

• License for Repo

  #67 opened 9 years ago by nvram83
  2

• issue with clearState

  #54 opened 9 years ago by mrastegari
  2

• Wrong model definition

  #62 opened 9 years ago by Atcold
  2

• utils.lua nn.DataParallelTable requires 'cunn'

  #60 opened 9 years ago by anuragranj
  0

• opts.backend not supported, breaks model.lua

  #56 opened 9 years ago by anuragranj
  2

• What is the best way to test the1-bit quantization routines?

  #52 opened 9 years ago by chienlinhuang1116
  0

• How can I use fbcunn libs in the exmaple codes?

  #50 opened 9 years ago by chienlinhuang1116
  0

• mattorch.load with multi threading.

  #45 opened 9 years ago by skrealworld
  2