Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / soumith/imagenet-multigpu.torch issues and pull requests

#110 - Can not continue training

Issue - State: closed - Opened by mhmtsarigul almost 6 years ago

#109 - how to save my image data into torch format (.t7)

Issue - State: open - Opened by zq1314 about 6 years ago - 3 comments

#107 - Why not call cutorch.synchronizeAll()

Issue - State: open - Opened by farleylai about 7 years ago

#106 - Trained googlenet's outputs are negative value. Why negative?

Issue - State: open - Opened by bemoregt over 7 years ago - 1 comment

#104 - How can I retrain this googlenet network?

Issue - State: open - Opened by bemoregt over 7 years ago - 2 comments

#103 - Retraining an existing model(No details)

Issue - State: open - Opened by anmolbhatia28 over 7 years ago - 1 comment

#102 - Problem with Thread ???

Issue - State: open - Opened by hnanhtuan over 7 years ago

#101 - Data loading time is different across threads

Issue - State: closed - Opened by JiayunLi over 7 years ago

#100 - "nn/Linear.lua:69: input must be vector or matrix"

Issue - State: open - Opened by bemoregt over 7 years ago

#99 - testing extremely slow

Issue - State: open - Opened by varghesealex90 over 7 years ago

#98 - Data loading

Issue - State: closed - Opened by varghesealex90 over 7 years ago

#97 - Training time about Multi GPUs and Single GPU

Issue - State: open - Opened by xieshuqin over 7 years ago - 1 comment

#96 - Model convergence changes; option to initialize weights

Pull Request - State: closed - Opened by dheerajsreedhar over 7 years ago - 1 comment

#94 - Enabling Autotune

Pull Request - State: closed - Opened by dheerajsreedhar almost 8 years ago - 1 comment

#93 - Benchmark

Pull Request - State: closed - Opened by dheerajsreedhar almost 8 years ago

#92 - SqueezeNet issue

Issue - State: closed - Opened by rracinskij almost 8 years ago - 2 comments

#91 - fix the last mini-batch

Pull Request - State: closed - Opened by ljk628 almost 8 years ago - 1 comment

#89 - Dropout in AlexNet

Issue - State: open - Opened by joeyhng about 8 years ago

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

Issue - State: closed - Opened by yiheng about 8 years ago - 3 comments

#86 - Re-enable the cunn backend.

Pull Request - State: closed - Opened by williford about 8 years ago - 1 comment

#85 - cunn backend no longer supported

Issue - State: closed - Opened by williford about 8 years ago

#84 - Make vggbn.lua use `nClasses` instead of 1000.

Pull Request - State: closed - Opened by williford about 8 years ago - 1 comment

#83 - Fixes #79 - DPT does not need syncParameters anymore

Pull Request - State: closed - Opened by tomsercu about 8 years ago - 1 comment

#82 - Googlenet with Torch has accuracy of 0 %

Issue - State: open - Opened by antoajayraj about 8 years ago

#81 - alexnetowtbn runs exit with illegal memory access error

Issue - State: open - Opened by antoajayraj about 8 years ago

#80 - Top-1 error of 45.07

Issue - State: open - Opened by Viresh-R about 8 years ago

#79 - syncParameters() bug without consequences

Issue - State: closed - Opened by tomsercu about 8 years ago - 1 comment

#78 - attempt to call global 'unpack' (a nil value)

Issue - State: closed - Opened by karandwivedi42 about 8 years ago - 1 comment

#77 - A question about predicting individual images

Issue - State: open - Opened by jsongcse about 8 years ago - 3 comments

#76 - Code does not resize images to 256x256

Issue - State: closed - Opened by sozubek over 8 years ago - 1 comment

#75 - Too much time per epoch

Issue - State: closed - Opened by Viresh-R over 8 years ago - 6 comments

#74 - training on multiple GPUs gives nan

Issue - State: closed - Opened by arashno over 8 years ago - 1 comment

#73 - Which Class Labels To Indices In Output Tensor?

Issue - State: open - Opened by hsoule over 8 years ago - 4 comments

#72 - Class has zero samples

Issue - State: open - Opened by vesvis over 8 years ago - 3 comments

#71 - Alexnet code for fbcunn

Issue - State: closed - Opened by antoajayraj over 8 years ago - 5 comments

#70 - Newbie question: Is my model making any progress

Issue - State: closed - Opened by livenletdie over 8 years ago - 2 comments

#69 - Error attempt to index field 'THNN' (a nil value)

Issue - State: closed - Opened by ghost over 8 years ago - 1 comment

#68 - SqueezeNet implementation

Pull Request - State: closed - Opened by windscope over 8 years ago - 5 comments

#67 - License for Repo

Issue - State: closed - Opened by nvram83 over 8 years ago - 2 comments

#66 - Vggbn

Issue - State: closed - Opened by sinaghs over 8 years ago

#65 - random loading data sequence

Issue - State: open - Opened by zijunwei over 8 years ago - 1 comment

#64 - Torch.save on dataset class error with gm package

Issue - State: closed - Opened by vrkrishn over 8 years ago - 1 comment

#63 - vggbn training's validation accuracy is very low

Issue - State: open - Opened by cxy7452 over 8 years ago - 2 comments

#62 - Wrong model definition

Issue - State: open - Opened by Atcold over 8 years ago - 2 comments

#61 - fixes issue #60

Pull Request - State: closed - Opened by anuragranj over 8 years ago - 1 comment

#60 - utils.lua nn.DataParallelTable requires 'cunn'

Issue - State: closed - Opened by anuragranj over 8 years ago

#59 - Train Alexnet with 4 GPUs seems slower than one

Issue - State: closed - Opened by yiheng over 8 years ago - 2 comments

#58 - Network-in-Network

Pull Request - State: closed - Opened by szagoruyko over 8 years ago

#57 - Backend help update

Pull Request - State: closed - Opened by szagoruyko over 8 years ago

#56 - opts.backend not supported, breaks model.lua

Issue - State: closed - Opened by anuragranj over 8 years ago - 2 comments

#55 - low accuracy on alexnetowtbn

Issue - State: open - Opened by mrastegari over 8 years ago - 12 comments

#54 - issue with clearState

Issue - State: closed - Opened by mrastegari over 8 years ago - 2 comments

#53 - Update dataset.lua

Pull Request - State: closed - Opened by anuragranj over 8 years ago

#51 - remove unused code

Pull Request - State: closed - Opened by jhjin over 8 years ago

#51 - remove unused code

Pull Request - State: closed - Opened by jhjin over 8 years ago

#50 - How can I use fbcunn libs in the exmaple codes?

Issue - State: open - Opened by chienlinhuang1116 over 8 years ago

#49 - nGPU=4 is slower than nGPU=1

Issue - State: closed - Opened by chienlinhuang1116 over 8 years ago - 14 comments

#48 - remove defaultDir and hardcoded save path

Pull Request - State: closed - Opened by ljk628 over 8 years ago - 1 comment

#47 - DPT Synchronize() performance fix. GPU flag fix

Pull Request - State: closed - Opened by lukacf over 8 years ago - 1 comment

#46 - clearState

Pull Request - State: closed - Opened by szagoruyko over 8 years ago

#45 - mattorch.load with multi threading.

Issue - State: open - Opened by skrealworld over 8 years ago - 2 comments

#44 - Retrain throws error

Pull Request - State: closed - Opened by ppwwyyxx over 8 years ago - 2 comments

#43 - png file in ILSVRC12

Issue - State: closed - Opened by ppwwyyxx over 8 years ago - 2 comments

#42 - nn backend select bugfix

Pull Request - State: closed - Opened by szagoruyko over 8 years ago

#41 - Handle auxiliary classifiers properly

Pull Request - State: open - Opened by apaszke over 8 years ago

#40 - Testing should process batches in chunks too

Pull Request - State: closed - Opened by apaszke over 8 years ago - 1 comment

#39 - Cannot train googlenet

Issue - State: open - Opened by ppwwyyxx over 8 years ago - 1 comment

#38 - Replace remaining cudnn functions

Pull Request - State: closed - Opened by jhjin over 8 years ago - 1 comment

#37 - Remove separate model definition files for each backend and use cudnn.convert

Pull Request - State: closed - Opened by szagoruyko over 8 years ago - 1 comment

#36 - When do we need to use cutorch.synchronize

Issue - State: closed - Opened by eriche2016 over 8 years ago - 1 comment

#35 - some bug in vggbn_cudnn.lua

Issue - State: closed - Opened by buttomnutstoast over 8 years ago - 1 comment

#34 - A tiny tiny bug

Issue - State: closed - Opened by WendyShang over 8 years ago - 1 comment

#33 - Multi Label Example

Issue - State: closed - Opened by lolz0r over 8 years ago - 1 comment

#32 - Changing GPU device id to >1 gives an error.

Issue - State: open - Opened by vislab2013 over 8 years ago - 4 comments

#31 - Fix satellite thread Tensor indexing bug

Pull Request - State: closed - Opened by Atcold almost 9 years ago - 2 comments

#30 - opt was expecting

Issue - State: closed - Opened by pedropgusmao almost 9 years ago

#29 - Corrects logic when using epochNumber without giving a LR.

Pull Request - State: closed - Opened by pedropgusmao almost 9 years ago

#28 - Predict bug attempt to call global 'testHook' (a nil value)

Issue - State: closed - Opened by hamsal almost 9 years ago - 1 comment

#27 - Clean save and load of the DataParallelTable

Pull Request - State: closed - Opened by albanD almost 9 years ago - 2 comments

#26 - Allows for models with different input sizes

Pull Request - State: closed - Opened by pedropgusmao almost 9 years ago - 1 comment

#25 - Parameterizes nb of classes in output layer

Pull Request - State: closed - Opened by jhjin almost 9 years ago

#24 - Data loading time

Issue - State: closed - Opened by Atcold almost 9 years ago - 4 comments

#23 - Some help debugging?

Issue - State: closed - Opened by Atcold almost 9 years ago - 3 comments

#22 - added classification feature

Pull Request - State: open - Opened by cbarokk almost 9 years ago - 1 comment

#21 - Threaded data loading

Issue - State: closed - Opened by dasguptar almost 9 years ago - 2 comments

#20 - CPU version

Issue - State: closed - Opened by srp1970 almost 9 years ago - 3 comments

#19 - Getting not supported error for DataParallelTable type

Issue - State: closed - Opened by srp1970 almost 9 years ago - 2 comments

#18 - Do not normalize images two times

Pull Request - State: closed - Opened by mfigurnov almost 9 years ago - 2 comments

#17 - Remove fbcunn*, use DataParallelTable instead

Pull Request - State: closed - Opened by soumith almost 9 years ago - 1 comment

#16 - README: Instructions for installing GNU tools on OSX

Pull Request - State: closed - Opened by Smerity almost 9 years ago - 3 comments

#15 - multi GPU: select GPU to use and #to use

Issue - State: closed - Opened by culurciello about 9 years ago - 1 comment

#14 - assert fails on non-square images

Pull Request - State: closed - Opened by bobbens about 9 years ago - 1 comment

#13 - std calculation

Issue - State: closed - Opened by bobbens about 9 years ago - 1 comment

#12 - Different performance when model is reloaded

Issue - State: closed - Opened by dileepc about 9 years ago - 7 comments