messages listlengths 2 8 | kind stringclasses 2
values | thread_id int64 17.3k 206k | url stringlengths 48 283 | created_at stringlengths 20 27 | tier stringclasses 2
values | answer_origin stringclasses 2
values | answer_post_numbers listlengths 1 5 | answer_authors listlengths 1 3 |
|---|---|---|---|---|---|---|---|---|
[
{
"role": "user",
"content": "GPU utilized 99% but Cudnn not used, extremely slow\n\nHello!\n\nTo work with a remote GPU server running on CentOS 6.9, I had to install pytorch from source (the packaged version as an incompatibility with glibc). I ran 20 batches of my training process with the autograd profi... | answer | 21,035 | https://discuss.pytorch.org/t/gpu-utilized-99-but-cudnn-not-used-extremely-slow/21035 | 2018-07-11T19:48:41.271000Z | gold | verbatim | [
6
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "GPU utilized 99% but Cudnn not used, extremely slow\n\nHello!\n\nTo work with a remote GPU server running on CentOS 6.9, I had to install pytorch from source (the packaged version as an incompatibility with glibc). I ran 20 batches of my training process with the autograd profi... | clarification | 21,035 | https://discuss.pytorch.org/t/gpu-utilized-99-but-cudnn-not-used-extremely-slow/21035 | 2018-07-11T19:48:41.271000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "How to split test and train data keeping equal proportions of each class?\n\nSuppose I have a dataset with the following classes:\n\nClass A: 3000 items\nClass B: 1000 items\nClass C: 2000 items\n\nI want to split this dataset in two parts so that there are 25% data in test set... | answer | 21,063 | https://discuss.pytorch.org/t/how-to-split-test-and-train-data-keeping-equal-proportions-of-each-class/21063 | 2018-07-12T12:59:32.552000Z | gold | verbatim | [
7
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Padding setting of torch.nn.ConvTranspose2d\n\nI am a little confused about the padding setting of the torch.nn.ConvTranspose2d.\n\nAs I see the document, it seems that the padding of the deconvolution is calculated by some settings of convolution. **padding(deconv) = K(conv) ... | answer | 21,066 | https://discuss.pytorch.org/t/padding-setting-of-torch-nn-convtranspose2d/21066 | 2018-07-12T13:35:23.967000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Empty state_dict after moving model to GPU\n\nHi all,\n\nI am encountering a weird phenomenon when trying to save the parameters of a custom model that has being moved to the GPU in Pytorch 0.4.0. When I call model.state_dict(), the resulting OrderedDict is empty, see the follo... | answer | 21,069 | https://discuss.pytorch.org/t/empty-state-dict-after-moving-model-to-gpu/21069 | 2018-07-12T14:14:24.413000Z | silver | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Training accuracy (loss) increases (decreases) in a zigzag way?\n\nMy training log is very strange that first batch always performs better then later batches in an epoch.\n\nHere is my training log\n\n* training loss [image]\n\n* training accuracy [image]\n\nAs you see, they de... | answer | 21,112 | https://discuss.pytorch.org/t/training-accuracy-loss-increases-decreases-in-a-zigzag-way/21112 | 2018-07-13T15:05:14.938000Z | silver | verbatim | [
6
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Training accuracy (loss) increases (decreases) in a zigzag way?\n\nMy training log is very strange that first batch always performs better then later batches in an epoch.\n\nHere is my training log\n\n* training loss [image]\n\n* training accuracy [image]\n\nAs you see, they de... | clarification | 21,112 | https://discuss.pytorch.org/t/training-accuracy-loss-increases-decreases-in-a-zigzag-way/21112 | 2018-07-13T15:05:14.938000Z | silver | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "CNN Calculated Padded Error\n\nI'm trying to replicate a Keras model which starts out like this:\n\n```\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=======================================... | answer | 21,113 | https://discuss.pytorch.org/t/cnn-calculated-padded-error/21113 | 2018-07-13T15:32:07.732000Z | gold | verbatim | [
5
] | [
"colesbury"
] |
[
{
"role": "user",
"content": "CNN Calculated Padded Error\n\nI'm trying to replicate a Keras model which starts out like this:\n\n```\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=======================================... | clarification | 21,113 | https://discuss.pytorch.org/t/cnn-calculated-padded-error/21113 | 2018-07-13T15:32:07.732000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "CNN With 3 Conv Layer\n\nHi\nI want to make a CNN with 3 convolution layers and 3 pooling layers, the net is like this : \nconvolution-pooling-convolution-pooling-convolution-pooling-fc1-fc2\nthe pooling is 2x2 and its stride is 1 and the convolutions are 5x5\nthe output sizes ... | answer | 21,149 | https://discuss.pytorch.org/t/cnn-with-3-conv-layer/21149 | 2018-07-14T21:08:17.678000Z | silver | verbatim | [
4
] | [
"zeakey"
] |
[
{
"role": "user",
"content": "How to get the max (and the index of the max value) of tensors in a list\n\nLet s say I have a list consist of K tensors \n```\nL = [ torch.rand(B,C,D,D) for _ in range(K)]\n\n```\nfor simplicity lets say `B=1`\nI want to find the max value and the corresponding index to the ma... | answer | 21,173 | https://discuss.pytorch.org/t/how-to-get-the-max-and-the-index-of-the-max-value-of-tensors-in-a-list/21173 | 2018-07-15T18:21:52.028000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Conv1d diffuclties\n\nI have read the other similar posts but have not been able to solve the issue and get Conv1d to work. \n\nInitially I was getting this error: Runtime error: Expected object of type torch.DoubleTensor but found type torch.FloatTensor for argument #2 'weigh... | answer | 21,198 | https://discuss.pytorch.org/t/conv1d-diffuclties/21198 | 2018-07-16T08:36:12.283000Z | gold | composed | [
2,
3
] | [
"Tank",
"ptrblck"
] |
[
{
"role": "user",
"content": "Performing mini-batch gradient descent or stochastic gradient descent on a mini-batch\n\nHello, I have created a data-loader object, I set the parameter batch size equal to five and I run the following code. I would like some clarification, is the following code performing mini... | answer | 21,235 | https://discuss.pytorch.org/t/performing-mini-batch-gradient-descent-or-stochastic-gradient-descent-on-a-mini-batch/21235 | 2018-07-16T19:01:46.787000Z | gold | verbatim | [
4
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Checking Whether Network is Dead, Pytorch?\n\nDoes anyone know a convenient and easy way to understand whether your NN is suffering from dead RELU's in Pytorch?"
},
{
"role": "assistant",
"content": "By dead ReLUs you mean a negative input and thus a zero result?\nI... | answer | 21,266 | https://discuss.pytorch.org/t/checking-whether-network-is-dead-pytorch/21266 | 2018-07-17T09:07:32.952000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "No_grad() vs requires_grad\n\nI know this is a very basic question, but it's my first day with pytorch and I can't seem to figure it out? What is the difference between no_grad() and requires_grad, and when to use each of them, and when/how to mix them?\n\nThanks guys.\n\nBest,... | answer | 21,272 | https://discuss.pytorch.org/t/no-grad-vs-requires-grad/21272 | 2018-07-17T11:45:16.271000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Building complex CNN\n\nIs it possible to build a CNN like the following easily in Pytorch?\n\nX_train[0:99] -> Con1 -> Conv2 - MaxPool / + X[99:105] -> linear1 -> linear2 ->Ouput \n\nAs in we are adding more information to the fully connected layers than simply what the conv ... | answer | 21,325 | https://discuss.pytorch.org/t/building-complex-cnn/21325 | 2018-07-18T09:32:19.130000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Plt.imshow shows weird data augmentation when I have none\n\nThe below piece of code gives me back really weird data augmented image. There are no other transformations besides the Resize. I am using pytorch 0.40\n\n[image]\n```\ntransformed_dataset = data_sets(csv_path=csv_pat... | answer | 21,354 | https://discuss.pytorch.org/t/plt-imshow-shows-weird-data-augmentation-when-i-have-none/21354 | 2018-07-18T21:11:42.192000Z | gold | verbatim | [
5
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Share weight in Resnet model\n\nHi,\n\nI have a fine-tuned ResNet model, but I have no idea how I can do to share a weight for this architecture!\n\nBasically, the idea is to share the weight only for the base_network.\n[image]\nThank you<\n```python\nclass ResNetModel(nn.Modul... | answer | 21,371 | https://discuss.pytorch.org/t/share-weight-in-resnet-model/21371 | 2018-07-19T06:47:22.331000Z | gold | verbatim | [
4
] | [
"chenglu"
] |
[
{
"role": "user",
"content": "Weight Initialization\n\nI am trying to use xavier_uniform_ initialization for batch and layer normalization layers however I am receiving a valueerror: Fan in and fan out can not be computed for tensor with less than 2 dimensions. Is there anyway I can rectify this problem?\n\... | answer | 21,410 | https://discuss.pytorch.org/t/weight-initialization/21410 | 2018-07-20T03:02:16.204000Z | silver | composed | [
2,
4,
5
] | [
"WR01",
"ptrblck"
] |
[
{
"role": "user",
"content": "How to use custom convolution module instead of torch.nn.conv2d?\n\nI am trying to use the example given below to construct a custom convolution module which will be used in the network instead of the torch.nn.conv2d.\n```\ninp = torch.randn(1, 3, 10, 12)\nw = torch.randn(2, 3,... | answer | 21,458 | https://discuss.pytorch.org/t/how-to-use-custom-convolution-module-instead-of-torch-nn-conv2d/21458 | 2018-07-21T00:11:44.822000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Model.eval() in predicting step yields SyntaxError\n\nI adopted a 2-layer neural network in the tutorial:\n \n model = nn.Sequential(\n nn.Linear(D_in, H),\n nn.Tanh(),\n nn.Linear(H, D_out),\n )\nThen I trained the model, save and load the model i... | answer | 21,468 | https://discuss.pytorch.org/t/model-eval-in-predicting-step-yields-syntaxerror/21468 | 2018-07-21T05:12:39.245000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Unclear about Weighted BCE Loss\n\nHey there super people!\nI am having issues understanding the `BCELoss` `weight` parameter. I am having a binary classification issue, I have an RNN which for each time step over a sequence produces a binary classification. Precisely, it produ... | answer | 21,486 | https://discuss.pytorch.org/t/unclear-about-weighted-bce-loss/21486 | 2018-07-21T16:30:20.993000Z | gold | verbatim | [
4
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Is it possible to Free-up a Dataloader?\n\nI am using different Dataloaders for train set and test set, so in all, I have 2 Dataloaders. \n\nI do training and testing in every epoch. \n\nIs there a way I can free up the Dataloader not being used (for eg. free-up the train datal... | answer | 21,498 | https://discuss.pytorch.org/t/is-it-possible-to-free-up-a-dataloader/21498 | 2018-07-21T21:49:41.693000Z | gold | verbatim | [
6
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Basic requirements for doing GPU-based computations\n\nHi,\nI'm using PyTorch tensor library for tensor based computations. I'm implementing a particular type of spiking neural network and I do not use PyTorch's Network or Module. I just use tensor and its methods plus some fun... | answer | 21,516 | https://discuss.pytorch.org/t/basic-requirements-for-doing-gpu-based-computations/21516 | 2018-07-22T11:23:23.656000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Simulate GPU for testing?\n\nI'm often in a situation where on my local machine (without a GPU) my code runs fine, but I'd like to test that all the tensors are added to the correct device if there's a GPU present. I want to test this before I commit my code, but do so would re... | answer | 21,521 | https://discuss.pytorch.org/t/simulate-gpu-for-testing/21521 | 2018-07-22T18:38:53.588000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "AssertionError: nn criterions don't compute the gradient w.r.t. targets - please mark these variables as volatile or not requiring gradients\n\nI tried to extract the feature by adding the VGG16 network to the predicted image and ground truth, and use L1 loss to establish the e... | answer | 21,542 | https://discuss.pytorch.org/t/assertionerror-nn-criterions-dont-compute-the-gradient-w-r-t-targets-please-mark-these-variables-as-volatile-or-not-requiring-gradients/21542 | 2018-07-23T08:58:41.427000Z | gold | verbatim | [
6
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Does batch norm statistic maintain when switching train and val in training?\n\nI am using batch norm in my network. For each epoch in training, I will perform evaluate loss in validation set. Will the batch norm statistic in train() mode maintain in next epoch because I want t... | answer | 21,553 | https://discuss.pytorch.org/t/does-batch-norm-statistic-maintain-when-switching-train-and-val-in-training/21553 | 2018-07-23T10:38:22.052000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Kernel size can't greater than actual input size\n\nHi,\nI'm getting the following error.\n\n```\nRuntimeError: Calculated padded input size per channel: (1017 x 1). \nKernel size: (3 x 3). \nKernel size can't greater than actual input size at /opt/conda/conda-bld/pytorch_15245... | answer | 21,579 | https://discuss.pytorch.org/t/kernel-size-cant-greater-than-actual-input-size/21579 | 2018-07-23T19:18:44.562000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Sub() received map but expected some tensors\n\nGood morning guys. \nI am trying to run a bunch of (complicated) code but I keep getting the following weird error:\n\n```\nINFO:root:Random state initialized with seed 28794055 \nINFO:root:Ani1 will be loaded...\nINFO:root:creat... | answer | 21,631 | https://discuss.pytorch.org/t/sub-received-map-but-expected-some-tensors/21631 | 2018-07-24T10:01:20.353000Z | silver | composed | [
7
] | [
"nicoliKim"
] |
[
{
"role": "user",
"content": "Sub() received map but expected some tensors\n\nGood morning guys. \nI am trying to run a bunch of (complicated) code but I keep getting the following weird error:\n\n```\nINFO:root:Random state initialized with seed 28794055 \nINFO:root:Ani1 will be loaded...\nINFO:root:creat... | clarification | 21,631 | https://discuss.pytorch.org/t/sub-received-map-but-expected-some-tensors/21631 | 2018-07-24T10:01:20.353000Z | silver | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Difference between (nn.Linear + nn.CrossEntropyLoss) and (nn.LogSoftmax + nn.NLLLoss)\n\nIn a multi-class classification, I sometimes see the following two implementations:\n* `nn.Linear` + `nn.CrossEntropyLoss`\n* `nn.LogSoftmax` + `nn.NLLLoss`\n\nAre they both the same in ter... | answer | 21,634 | https://discuss.pytorch.org/t/difference-between-nn-linear-nn-crossentropyloss-and-nn-logsoftmax-nn-nllloss/21634 | 2018-07-24T11:01:25.819000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Output of resnet34 network depends on the batch size\n\nHello,\n\nI am experimenting with resnet34 network in pytorch and I have noticed that sometimes the output of some layers depends on the batch size. Here is my code\n<pre><code class=\"lang-python\">x1 = torch.rand(100, 3,... | answer | 21,647 | https://discuss.pytorch.org/t/output-of-resnet34-network-depends-on-the-batch-size/21647 | 2018-07-24T16:51:38.702000Z | gold | verbatim | [
3
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "PCA while using Dataloaders\n\nHow do I apply PCA (from sklearn.decomposition) to an image dataset that I am loading using Dataloaders? \nIs there a better way to load an image dataset in pytorch that would allow me to apply PCA?"
},
{
"role": "assistant",
"content"... | answer | 21,650 | https://discuss.pytorch.org/t/pca-while-using-dataloaders/21650 | 2018-07-24T17:21:28.246000Z | silver | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "How to use focalloss in segmentation?\n\nHello, I’m doing a project about semantic segmentation with 2 class using U-net , but my data was unbalanced, so I think maybe use focalloss is a good idea for my project, and I was use BCEloss, so changed the final layer from 1 channel ... | answer | 21,695 | https://discuss.pytorch.org/t/how-to-use-focalloss-in-segmentation/21695 | 2018-07-25T10:16:29.269000Z | gold | verbatim | [
4
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Size mismatch of input if more than one CUDA device\n\nHey guys,\n\nI'm implementing some RL and got stuck at a, in my opinion, weird behaviour.\nI'll use `DataParallel` and the `device`-tag to move my Nets/ Data to the available device(s).\nUsing CPU and one CUDA device everyt... | answer | 21,697 | https://discuss.pytorch.org/t/size-mismatch-of-input-if-more-than-one-cuda-device/21697 | 2018-07-25T11:39:44.191000Z | gold | verbatim | [
7
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Size mismatch of input if more than one CUDA device\n\nHey guys,\n\nI'm implementing some RL and got stuck at a, in my opinion, weird behaviour.\nI'll use `DataParallel` and the `device`-tag to move my Nets/ Data to the available device(s).\nUsing CPU and one CUDA device everyt... | clarification | 21,697 | https://discuss.pytorch.org/t/size-mismatch-of-input-if-more-than-one-cuda-device/21697 | 2018-07-25T11:39:44.191000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "How can I use different losses to update different branches respectively, and sum grad together to update master(main) branch\n\nI am facing a multi-label problem. My network has 2 branches, which are separated from a conv layers.\nThere is the sketch of my network\n\n```\nclas... | answer | 21,700 | https://discuss.pytorch.org/t/how-can-i-use-different-losses-to-update-different-branches-respectively-and-sum-grad-together-to-update-master-main-branch/21700 | 2018-07-25T11:50:21.860000Z | silver | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "How to define the kernel size for conv2d\n\nif I wanna do simple convolution i can use conv2d.\nI want to simply compute the edge of my input (the input has the size of `BxCxDxD`), and my edge kernel has the size of `3x3`, how should i expand my edge kernel so when i do `conv2... | answer | 21,777 | https://discuss.pytorch.org/t/how-to-define-the-kernel-size-for-conv2d/21777 | 2018-07-26T14:50:32.799000Z | gold | verbatim | [
5
] | [
"inkplay"
] |
[
{
"role": "user",
"content": "Torch has not attribute load_state_dict?\n\nHi. I am trying to load a model with:\n\n```\nimport torch\n\nimport pyautogui as mouse\nimport cv2 \n\nfrom ScreenRecorder import Record,IniRecord,Frame\n\ndef start(model):\n sc_ini = Frame()\n monitor = sc_ini.get()\n sc =... | answer | 21,781 | https://discuss.pytorch.org/t/torch-has-not-attribute-load-state-dict/21781 | 2018-07-26T19:29:58.920000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "I am confused about updating input channels and output channels when you are trying to progressively grow a model\n\nHi I have a couple questions and I am inexperienced so please bear with me.\n\nI am trying to make a progressive autoencoder and I have thought a couple ways of ... | answer | 21,798 | https://discuss.pytorch.org/t/i-am-confused-about-updating-input-channels-and-output-channels-when-you-are-trying-to-progressively-grow-a-model/21798 | 2018-07-26T23:51:20.850000Z | gold | verbatim | [
3
] | [
"inkplay"
] |
[
{
"role": "user",
"content": "How to get MNIST data from torchvision with three channels for some pretrained model like VGG?\n\nI think this should be encountered by many people, and is not difficult, but when I want to use the MNIST data from torchvision with a model pretrained from torchvision.models, suc... | answer | 21,872 | https://discuss.pytorch.org/t/how-to-get-mnist-data-from-torchvision-with-three-channels-for-some-pretrained-model-like-vgg/21872 | 2018-07-28T01:31:36.517000Z | gold | verbatim | [
2
] | [
"deJQK"
] |
[
{
"role": "user",
"content": "Convolution and pooling layers need a method to calculate output size\n\nIs there a simple way to calculate the output size of multiple convolution and pooling layers? I mean the automatical way. Suppose I have 10 such layers, then when I decide to change the first layer, I wou... | answer | 21,895 | https://discuss.pytorch.org/t/convolution-and-pooling-layers-need-a-method-to-calculate-output-size/21895 | 2018-07-28T13:41:36.872000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "TuneError: Error running trial: 'bool' object is not callable\n\nWhile doing hyperparameter optimization using Ray Tune in pytorch, getting the below error:\n<pre>\n<code>\nTraceback (most recent call last):\n File \"/home/vsl3/anaconda2/lib/python2.7/site-packages/ray/worker.... | answer | 21,940 | https://discuss.pytorch.org/t/tuneerror-error-running-trial-bool-object-is-not-callable/21940 | 2018-07-29T13:15:12.629000Z | gold | verbatim | [
4
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "TuneError: Error running trial: 'bool' object is not callable\n\nWhile doing hyperparameter optimization using Ray Tune in pytorch, getting the below error:\n<pre>\n<code>\nTraceback (most recent call last):\n File \"/home/vsl3/anaconda2/lib/python2.7/site-packages/ray/worker.... | clarification | 21,940 | https://discuss.pytorch.org/t/tuneerror-error-running-trial-bool-object-is-not-callable/21940 | 2018-07-29T13:15:12.629000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Failed to install(update?) pytorch 0.4.0 from 0.2.0_3\n\nHi. I'm beginner of pytorch, and I'm trying to version up pytorch from 0.2.0_3 to 0.4.0.\n\nUsing conda environment, I tried to follow the instruction they gave(https://pytorch.org/)\n\nI followed codes they gave, (like c... | answer | 21,970 | https://discuss.pytorch.org/t/failed-to-install-update-pytorch-0-4-0-from-0-2-0-3/21970 | 2018-07-30T05:05:17.806000Z | gold | verbatim | [
4
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Out of memory without a hit on the GPU\n\nHi!\nI can monitor my GPU and I can see no hit on my GPU. But I get an out of memory error.\nI even set the batch size to 1. I'm feeding two images of size (224,224); (448,448).\n\nRuning of a GPU Quatro K1200\n```python\nRuntimeError: ... | answer | 21,988 | https://discuss.pytorch.org/t/out-of-memory-without-a-hit-on-the-gpu/21988 | 2018-07-30T09:07:07.675000Z | gold | composed | [
2,
3,
4,
5
] | [
"Deeply",
"jpainam",
"ptrblck"
] |
[
{
"role": "user",
"content": "Out of memory without a hit on the GPU\n\nHi!\nI can monitor my GPU and I can see no hit on my GPU. But I get an out of memory error.\nI even set the batch size to 1. I'm feeding two images of size (224,224); (448,448).\n\nRuning of a GPU Quatro K1200\n```python\nRuntimeError: ... | clarification | 21,988 | https://discuss.pytorch.org/t/out-of-memory-without-a-hit-on-the-gpu/21988 | 2018-07-30T09:07:07.675000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "How to normalize a tensor to have values between a and b\n\nIf i have tensor `A = torch.rand(30,500,50,50)` what is the smartest and fastest way to normalize each layer (the layers in A.size(1)) to have values between a and b.\nThe naive way is:\n\nB = torch.zeros(A.size())\n\n... | answer | 22,030 | https://discuss.pytorch.org/t/how-to-normalize-a-tensor-to-have-values-between-a-and-b/22030 | 2018-07-30T17:41:12.019000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Why is it when I call require_grad = False on all my params my weights in the network would still update?\n\nWhat I am trying to do right now is to write a multi layer conv2d encoder and freeze the weights from updating for the earlier layers. This hopefully would give me back ... | answer | 22,126 | https://discuss.pytorch.org/t/why-is-it-when-i-call-require-grad-false-on-all-my-params-my-weights-in-the-network-would-still-update/22126 | 2018-07-31T17:43:14.969000Z | gold | verbatim | [
6
] | [
"SimonW"
] |
[
{
"role": "user",
"content": "[SOLVED] Gradient Descent and Batches\n\nHey, I wanted to ask if there is a difference of feeding a batch to the model and then update the weights or feeding data solely to the model and sum up the loss and update after the batch size. I am aware that this depends also on the u... | answer | 22,181 | https://discuss.pytorch.org/t/solved-gradient-descent-and-batches/22181 | 2018-08-01T10:09:55.829000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "DataLoader for a LSTM Model with a Sliding Window\n\nI am working on a LSTM model and trying to use a DataLoader to provide the data. I am using stock price data and my dataset consists of:\n* Date (string)\n* Closing Price (float)\n* Price Change (float)\n\nRight now I am jus... | answer | 22,235 | https://discuss.pytorch.org/t/dataloader-for-a-lstm-model-with-a-sliding-window/22235 | 2018-08-01T22:42:25.406000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Can i use deep learning to measure the similarity between two variable length voice sequences?\n\nUsually we use DTW(Dynamic Time warpping) to measure the similarity between two variabel voice sequences. However DTW is time-cunsuming and not easy to run in the GPU since too muc... | answer | 22,269 | https://discuss.pytorch.org/t/can-i-use-deep-learning-to-measure-the-similarity-between-two-variable-length-voice-sequences/22269 | 2018-08-02T07:40:27.872000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Issues with torch.utils.data.random_split\n\nI'm getting an error when I use a DataLoader based on split dataset through `torch.utils.data.random_split()`.\nI have tried creating a dataloader with the un-split dataset and everything works fine. So I assume it's something do wi... | answer | 22,298 | https://discuss.pytorch.org/t/issues-with-torch-utils-data-random-split/22298 | 2018-08-02T14:37:57.446000Z | silver | verbatim | [
6
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Change learning of only one layer\n\nHello,\n\nI have seen that is possible to set diferent learnign rates for diferent layers using Per-parameter options. But in my case, I have a lot of diferent modules, and I only want to set a diferent learnign rate for just a layer. \n\nIf... | answer | 22,299 | https://discuss.pytorch.org/t/change-learning-of-only-one-layer/22299 | 2018-08-02T14:53:22.449000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Loading pretrained parameters to a model\n\nI have created a model, trained it, evaluated and saved with \n```\ntorch.save(model.state_dict(), 'pathToSavedModel')\n``` \nIf I create the same model in a different .py file, how can I load the pretrained parameters from an old mod... | answer | 22,324 | https://discuss.pytorch.org/t/loading-pretrained-parameters-to-a-model/22324 | 2018-08-02T19:01:06.453000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Missing keys & unexpected keys in state_dict when loading self trained model\n\nHi, there:\n\nI've encountered this problem and got stucked for a while. I have a labeled image dataset in a considerable large scale and I chose to train a vgg16 on it just starting from [pytorch's... | answer | 22,379 | https://discuss.pytorch.org/t/missing-keys-unexpected-keys-in-state-dict-when-loading-self-trained-model/22379 | 2018-08-03T11:43:15.513000Z | gold | verbatim | [
3
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Can some one give me a working example of nn.ModuleDict?\n\nThe code from documentation will give back an error if you add \n\n```model = MyModule()```\n\n```\nclass MyModule(nn.Module):\n def __init__(self):\n super(MyModule, self).__init__()\n self.choices = ... | answer | 22,447 | https://discuss.pytorch.org/t/can-some-one-give-me-a-working-example-of-nn-moduledict/22447 | 2018-08-04T22:09:16.829000Z | gold | composed | [
2,
3
] | [
"inkplay",
"ptrblck"
] |
[
{
"role": "user",
"content": "Some problem about valuation model(about the torch.nn.Module.eval() method)\n\nThe code is below:\n```\nimport torch\nimport matplotlib.pyplot as plt\nfrom torch.autograd import Variable\n\n# torch.manual_seed(1) # reproducible\n\nN_SAMPLES = 20\nN_HIDDEN = 300\n\n# training... | answer | 22,468 | https://discuss.pytorch.org/t/some-problem-about-valuation-model-about-the-torch-nn-module-eval-method/22468 | 2018-08-05T12:03:38.373000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Optimizers memory usage\n\nHi there, \nI was training a model with SGD and decided to move to Adam. \n\nI was using batch size 20 for SGD, however max BS i can use with Adam is 2.\n\n```\n optimizer = torch.optim.SGD([{'params': model.unet_model.parameters()},{'params': mode... | answer | 22,473 | https://discuss.pytorch.org/t/optimizers-memory-usage/22473 | 2018-08-05T14:48:42.930000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "How to rotate 90 and 270 degrees of 5D tensor\n\nAccording to https://github.com/pytorch/pytorch/pull/7873, the flip function have been merged in pytorch. I am using version 0.41 and it works for 5D tensor. My 5D tensor is `BxCxDxHxW`. For flipping, I just use \n```\nimages=ima... | answer | 22,476 | https://discuss.pytorch.org/t/how-to-rotate-90-and-270-degrees-of-5d-tensor/22476 | 2018-08-05T16:37:59.679000Z | silver | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Training epoch and Neural Net class of the tutorial inputs the info to wrong params\n\nhi this is my first post, so if not suitable sorry.\n\nbasically I am developing a NN that uses images. However, I receive this error:\n\nRuntimeError: Given groups=1, weight[5, 3, 436, 436],... | answer | 22,479 | https://discuss.pytorch.org/t/training-epoch-and-neural-net-class-of-the-tutorial-inputs-the-info-to-wrong-params/22479 | 2018-08-05T17:56:26.747000Z | gold | verbatim | [
4
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Can someone help on porting this snippet of torch to Pytorch?\n\nHi, I'm trying to train on ImageNet the same way it is done in torch(based on [this](https://github.com/facebook/fb.resnet.torch)). All transformations are present in Pytorch, except one. AlexNet lighting method. ... | answer | 22,501 | https://discuss.pytorch.org/t/can-someone-help-on-porting-this-snippet-of-torch-to-pytorch/22501 | 2018-08-06T06:02:57.801000Z | gold | verbatim | [
3
] | [
"Shisho_Sama"
] |
[
{
"role": "user",
"content": "Dealing with imbalanced datasets in pytorch\n\nI am trying to find a way to deal with imbalanced data in pytorch. I was used to Keras' `class_weight`, although I am not sure what it really did (I think it was a matter of penalizing more or less certain classes).\n\nThe only sol... | answer | 22,596 | https://discuss.pytorch.org/t/dealing-with-imbalanced-datasets-in-pytorch/22596 | 2018-08-07T13:37:47.850000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "How to do sort subsampling\n\nCan any one help me with this question?\nLets say i have a 4x4 tensor and i want to do subsampling in the following way, so for each 2x2 block i put the smallest elements together,\nthen the second smallest elements, and so on, so the out put will ... | answer | 22,616 | https://discuss.pytorch.org/t/how-to-do-sort-subsampling/22616 | 2018-08-07T20:31:20.761000Z | gold | verbatim | [
9
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "RuntimeError: CUDA error: unknown error\n\nI faced to something weird error when I try to move Tensor to GPU.\nFirst my machine spec is \n\nUbuntu 16.04\nnvidia driver: 390.77\nGPU: TitanX (Pascal)\n<anaconda3>\n Torch 0.4.1 (with cuda9.0)\n\nI didn't install any cuda Toolkit ... | answer | 22,634 | https://discuss.pytorch.org/t/runtimeerror-cuda-error-unknown-error/22634 | 2018-08-08T05:44:25.328000Z | gold | composed | [
4,
6
] | [
"ojh3636",
"ptrblck"
] |
[
{
"role": "user",
"content": "Dataloader access two items at the same time\n\nHello,\n\nI am looping on dataloader, but I want to access two items and the same time\nExample:\n\n```\ndataset1 = ImageFolder(input_folder, transform=transform)\ntrain_loader_a= DataLoader(dataset=dataset1, batch_size=batch_size... | answer | 22,664 | https://discuss.pytorch.org/t/dataloader-access-two-items-at-the-same-time/22664 | 2018-08-08T12:55:25.453000Z | silver | composed | [
4,
5,
6
] | [
"mhusseinsh",
"ptrblck"
] |
[
{
"role": "user",
"content": "Resize image before calling torchvision.utils.save_image\n\nHi!\nI'm using `save_image` after some conv layer to output the image. But since the output shape of the tensor is `torch.Size([32, 1, 3, 3])`. I end up having very tiny images. How can I resize before calling save_ima... | answer | 22,670 | https://discuss.pytorch.org/t/resize-image-before-calling-torchvision-utils-save-image/22670 | 2018-08-08T14:03:19.614000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Is there a way to prepend a module to the beginning or a ModuleList?\n\nI am trying to build a progressive autoencoder, for the encoder part I would like to prepend more layer on top of a ModuleList, how can I achieve that? I want to avoid copying and remaking my model every ti... | answer | 22,688 | https://discuss.pytorch.org/t/is-there-a-way-to-prepend-a-module-to-the-beginning-or-a-modulelist/22688 | 2018-08-08T17:18:08.287000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Does Maxpool2d have any learnable parameter?\n\nI am wondering if maxpool2d in pytorch as any learnable parameter? and if so what is that?\nI saw people use ` self.pool1 = nn.MaxPool2d(2, 2) , self.pool2 = nn.MaxPool2d(2, 2),` etc in their models. So i assume the... | answer | 22,689 | https://discuss.pytorch.org/t/does-maxpool2d-have-any-learnable-parameter/22689 | 2018-08-08T17:21:47.618000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Saving and restoring model weights at the mini-batch level?\n\nIn the event that a classification model is being trained on a large amount of data (~3,000,000 input images per epoch), what is the recommended approach for implementing [checkpoint](https://discuss.pytorch.org/t/t... | answer | 22,699 | https://discuss.pytorch.org/t/saving-and-restoring-model-weights-at-the-mini-batch-level/22699 | 2018-08-08T21:38:14.589000Z | gold | verbatim | [
6
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Which year ImageNet are torchvision models pretrained on?\n\naccording to the documentation https://pytorch.org/docs/stable/torchvision/models.html , pretrained models are pretrained on ImageNet. But I'd like to know **exactly the year** of the ImageNet dataset used. \nDoes it ... | answer | 22,816 | https://discuss.pytorch.org/t/which-year-imagenet-are-torchvision-models-pretrained-on/22816 | 2018-08-10T03:24:56.513000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Applying conv2d filter to all channels seperately, is my solution efficient?\n\nHi,\n\nFor a given input of size (batch, channels, width, height) I would like to apply a 2-strided convolution with a single fixed 2D-filter to each channel of each batch, resulting in an output of... | answer | 22,840 | https://discuss.pytorch.org/t/applying-conv2d-filter-to-all-channels-seperately-is-my-solution-efficient/22840 | 2018-08-10T08:51:45.126000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "How to change Spatial Transformer Network to support different image sizes\n\nFollowing the tutorial:\nhttps://pytorch.org/tutorials/intermediate/spatial_transformer_tutorial.html\n\nWhat would I need to do to adapt the STN network to work with specific B=batch size, H=image he... | answer | 22,911 | https://discuss.pytorch.org/t/how-to-change-spatial-transformer-network-to-support-different-image-sizes/22911 | 2018-08-11T09:40:44.760000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Multiply one tensor of channel 1 with a tensors with multiple channel across all channel\n\nHi. I have a tensor `y` of shape `torch.Size([32, 1, 4, 2])` and `x` of shape `torch.Size([32, 2048, 4, 2])`\nI want to multiply `x*y` across all `x` channel. I have tried \n```python\n#... | answer | 22,972 | https://discuss.pytorch.org/t/multiply-one-tensor-of-channel-1-with-a-tensors-with-multiple-channel-across-all-channel/22972 | 2018-08-12T15:10:10.825000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "State dictonary holds references to internal state of network\n\nWhen getting a state dict using `<module>.state_dict()` the dictionary references the internal parameters of the model. Meaning, once the model changes, the dict will also change. Usually this doesn't really impac... | answer | 23,012 | https://discuss.pytorch.org/t/state-dictonary-holds-references-to-internal-state-of-network/23012 | 2018-08-13T08:15:13.877000Z | gold | verbatim | [
4
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Cuda.is_available() is True on CPU?\n\nHi everyone,\n\nI tried to use such a function to determine cuda availability\n```\ndef get_torch_device():\n if torch.cuda.is_available():\n return torch.device(\"cuda:0\")\n return torch.device(\"cpu\")\n\n```\nand it return... | answer | 23,029 | https://discuss.pytorch.org/t/cuda-is-available-is-true-on-cpu/23029 | 2018-08-13T12:18:21.677000Z | gold | verbatim | [
5
] | [
"roaffix"
] |
[
{
"role": "user",
"content": "How to remove an element from a 1-d tensor by index?\n\nSo I have a 1-d tensor _T_ and an index _i_ and need to remove i-th element from a tensor T, much like in pure python T.remove(i).\n\nI've tried to do this:\n```\ni = 2\nT = torch.tensor([1,2,3,4,5])\nT = torch.cat([T[0:i]... | answer | 23,109 | https://discuss.pytorch.org/t/how-to-remove-an-element-from-a-1-d-tensor-by-index/23109 | 2018-08-14T12:36:40.376000Z | gold | verbatim | [
3
] | [
"ptab"
] |
[
{
"role": "user",
"content": "How can I read a image pair at the same time\n\nfor the Consumer-to-shop Clothes Retrieval dataset, it need to read a pair image at the same time, it will be okay as followed code?\n\n```\n\nimg1 = read('...')\n\nimg2 = read('...')\n\nimg = [img1, img2]\n\n``` \nthanks for rep... | answer | 23,151 | https://discuss.pytorch.org/t/how-can-i-read-a-image-pair-at-the-same-time/23151 | 2018-08-15T02:32:59.876000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Got 'RuntimeError: CUDNN_STATUS_INTERNAL_ERROR' in nn.functional.py conv3d()\n\nHi, \n\nI was trying to train GoogLeNet on a server with multiple GPUs using python3 + pytorch 0.3.1.post2. However, I keep getting this error: 'RuntimeError: CUDNN_STATUS_INTERNAL_ERROR'. It seems ... | answer | 23,230 | https://discuss.pytorch.org/t/got-runtimeerror-cudnn-status-internal-error-in-nn-functional-py-conv3d/23230 | 2018-08-16T04:11:10.131000Z | gold | composed | [
2,
3
] | [
"JustnLiu",
"ptrblck"
] |
[
{
"role": "user",
"content": "Some problems with WeightedRandomSampler\n\nDear groupers,\n I work on an unbalanced dataset. There are six class in my dataset. The first class has 568330 samples, the second class has 43000 samples, the third class has 34900, the fourth class has 20910, the fifth class... | answer | 23,242 | https://discuss.pytorch.org/t/some-problems-with-weightedrandomsampler/23242 | 2018-08-16T07:34:59.586000Z | gold | verbatim | [
34
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "RuntimeError: Expected object of type torch.LongTensor but found type torch.FloatTensor\n\nfc = nn.Linear(128, 128)\nlin = fc1(torch.tensor([16,128]))\n\nCan anyone please tell me why this error appears?.. it works when editing the last line to be\nlin = fc1(torch.Tensor(16,128... | answer | 23,289 | https://discuss.pytorch.org/t/runtimeerror-expected-object-of-type-torch-longtensor-but-found-type-torch-floattensor/23289 | 2018-08-16T17:55:14.282000Z | gold | verbatim | [
6
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "About the input problem of the densenet201?\n\nclass DenseNet(nn.Module):\n \"\"\"Densenet-BC model class\n Args:\n growth_rate (int) - how many filters to add each layer (k in paper)\n block_config (list of 4 ints) - how many layers in each pooling block\n ... | answer | 23,443 | https://discuss.pytorch.org/t/about-the-input-problem-of-the-densenet201/23443 | 2018-08-19T14:31:08.930000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Non-deterministic tensor index arithmetic on cuda\n\nThe code in this [gist](https://gist.github.com/bgobbi/042bbee9b0bedf70995f88c6ce3a9208) does some indexing and simple arithmetic with the exact same inputs 20 000 times.\nIt is supposed to find the index of the first and las... | answer | 23,446 | https://discuss.pytorch.org/t/non-deterministic-tensor-index-arithmetic-on-cuda/23446 | 2018-08-19T15:14:01.647000Z | gold | verbatim | [
11
] | [
"colesbury"
] |
[
{
"role": "user",
"content": "Pytorch DQN tutorial - where is autograd?\n\nhttps://pytorch.org/tutorials/intermediate/reinforcement_q_learning.html\n\nwhile the comments in the tutorial specify that autograd is used, it is never explicitly declared (that I can see). In supervised learning, the inputs are us... | answer | 23,460 | https://discuss.pytorch.org/t/pytorch-dqn-tutorial-where-is-autograd/23460 | 2018-08-19T21:03:08.328000Z | gold | verbatim | [
6
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Pytorch DQN tutorial - where is autograd?\n\nhttps://pytorch.org/tutorials/intermediate/reinforcement_q_learning.html\n\nwhile the comments in the tutorial specify that autograd is used, it is never explicitly declared (that I can see). In supervised learning, the inputs are us... | clarification | 23,460 | https://discuss.pytorch.org/t/pytorch-dqn-tutorial-where-is-autograd/23460 | 2018-08-19T21:03:08.328000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "UnsatisfiableError when installing specific versions of pytorch and cudatoolkit\n\nHi,\n\nI would like to install pytorch version 0.4.1 with cuda 8.0 with python 3.6.\nWhen I run `conda info pytorch`, there is one block as follows:\n```\npytorch 0.4.1 py36_cuda8.0.61_cudnn7.1.2... | answer | 23,462 | https://discuss.pytorch.org/t/unsatisfiableerror-when-installing-specific-versions-of-pytorch-and-cudatoolkit/23462 | 2018-08-19T21:09:44.448000Z | gold | verbatim | [
2
] | [
"smth"
] |
[
{
"role": "user",
"content": "How to train with imbalance class for classification?\n\nHello all, I have read some related question about imbalance classification; however, I did not find the answer. My dataset is imbalance class that shows the distribution in below\n\n[image]\n\nI am using `WeightedRandomS... | answer | 23,468 | https://discuss.pytorch.org/t/how-to-train-with-imbalance-class-for-classification/23468 | 2018-08-20T01:35:24.746000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "About Normalization using pre-trained vgg16 networks\n\nI am trying to use the given vgg16 network to extract features (not fine-tuning) for my own task dataset,such as UCF101, rather than Imagenet. Since vgg16 is trained on ImageNet, for image normalization, I see a lot of pe... | answer | 23,560 | https://discuss.pytorch.org/t/about-normalization-using-pre-trained-vgg16-networks/23560 | 2018-08-21T05:59:03.923000Z | gold | verbatim | [
6
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Using nn.function.interpolate() inside nn.Sequential()\n\n`nn.Upsample()` is depecated in `pytorch version > 0.4.0` in favor of `nn.functional.interpolate()`\nI'm not able to use `interpolate()` inside `nn.Sequential()`:\n\nBelow is my network:\n\n```\nclass MeeDecoder(torch.nn... | answer | 23,588 | https://discuss.pytorch.org/t/using-nn-function-interpolate-inside-nn-sequential/23588 | 2018-08-21T12:15:52.113000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Linear layer default weight initialization\n\nThe default Linear layer weight initialization mechanism isn't clear to me.\n\nIf I use default initialization, without calling tensor.nn.init.XX or reset_parameters(), I get different weight values than when I do explicitly initial... | answer | 23,610 | https://discuss.pytorch.org/t/linear-layer-default-weight-initialization/23610 | 2018-08-21T18:35:48.656000Z | silver | verbatim | [
3
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "RuntimeError with pretrained inception v3 model\n\nHi, I want to finetuning the pretrained inception v3 model, but get type mismatch error. the following is my code.\n```\n#!/usr/bin/env python3\n\nimport torch\nimport torch.nn as nn\nimport torch.optim as optim\nfrom torchvisi... | answer | 23,736 | https://discuss.pytorch.org/t/runtimeerror-with-pretrained-inception-v3-model/23736 | 2018-08-23T08:53:05.500000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "How to change the mean and var value of the BN layer for a pretrained model?\n\nI have a pretrained model, and want to apply it to a new dataset. From the AdaBN paper, I want to recaculate the the mean and var for each neuros and update the value for each BN layer. How could I ... | answer | 23,752 | https://discuss.pytorch.org/t/how-to-change-the-mean-and-var-value-of-the-bn-layer-for-a-pretrained-model/23752 | 2018-08-23T13:23:41.443000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "My CPU RAM is getting full but my GPU RAM is underutilized\n\nHi,\nI'm using Convolutional Autoencoder Network. While training my CPU RAM (30GB) is getting fully used in just 20 epochs but my GPU memory (8GB) is used only 5%.\n\nHow should I approach to use my GPU better and re... | answer | 23,786 | https://discuss.pytorch.org/t/my-cpu-ram-is-getting-full-but-my-gpu-ram-is-underutilized/23786 | 2018-08-24T05:04:04.015000Z | gold | verbatim | [
4
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Convolutional 1D NN with different sequence lengths\n\nHi. \n\nI have a hard time understanding the size of tensors and how they change shape thorughout my network.\n\nMy data consist of 1000s of .csv files which each have 1000s of data samples. Some have few samples like 10 so... | answer | 23,816 | https://discuss.pytorch.org/t/convolutional-1d-nn-with-different-sequence-lengths/23816 | 2018-08-24T13:49:34.940000Z | silver | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Why does groups kwarg in conv2d have different requirements in functional?\n\nThe convolutional layers (e.g. [nn.Conv2d](https://pytorch.org/docs/stable/nn.html#torch.nn.Conv2d)) require `groups` to divide both `in_channels` and `out_channels`. The functional convolutions (e.g.... | answer | 23,837 | https://discuss.pytorch.org/t/why-does-groups-kwarg-in-conv2d-have-different-requirements-in-functional/23837 | 2018-08-24T20:44:59.549000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Get same output with different image on pre-trained ResNet18\n\nI used `torchvision.model.resnet18` to classify images, but it seems output same result with different input.\nHere is my test code\n```\nres_model = torchvision.models.resnet18(pretrained=True)\nnormalize = transf... | answer | 23,918 | https://discuss.pytorch.org/t/get-same-output-with-different-image-on-pre-trained-resnet18/23918 | 2018-08-26T12:11:47.850000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Pytorch trained model on Webcam\n\nIs there any way to do real time image classification with webcam using Pytorch trained model?"
},
{
"role": "assistant",
"content": "You could just grab the webcam frame, preprocess it accordingly to your pipeline, and feed it int... | answer | 23,928 | https://discuss.pytorch.org/t/pytorch-trained-model-on-webcam/23928 | 2018-08-26T18:51:48.836000Z | gold | verbatim | [
11
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Error in training inception-v3\n\nI have a model that was written using models from torchvision and I wanna test the performance with inception-v3. However, with the same model structure and imput images (size 224 x 224), I got the following error. \n\n RuntimeError: Calcula... | answer | 23,933 | https://discuss.pytorch.org/t/error-in-training-inception-v3/23933 | 2018-08-26T22:44:11.060000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "What happens if I pair a GTX1080 with a GTX1080TI to train in Pytorch?\n\nhello everyone,\nIs it possible to use two different GPUs for training the same network? I mean in a multi-gpu scenario, or should all the GPUs be the same?\nI'm planning to get a new GTX1080TI(from Asus,... | answer | 23,946 | https://discuss.pytorch.org/t/what-happens-if-i-pair-a-gtx1080-with-a-gtx1080ti-to-train-in-pytorch/23946 | 2018-08-27T05:51:16.649000Z | gold | verbatim | [
2
] | [
"ptrblck"
] |
[
{
"role": "user",
"content": "Tensor.to_cuda() is slow when use our own .whl file built from source code\n\nWe have several offline machines and want to deploy pytorch on them. To do this, We need to build a wheel package containing all dependencies (just like the official wheel package) in one machine and ... | answer | 23,948 | https://discuss.pytorch.org/t/tensor-to-cuda-is-slow-when-use-our-own-whl-file-built-from-source-code/23948 | 2018-08-27T06:44:14.663000Z | gold | composed | [
1,
2,
4
] | [
"YuxiaoXu",
"ptrblck"
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.