Write TensorBoard events with simple function call.
The current release (v2.6.3) is tested with PyTorch 2.6 / torchvision 0.21.0 / tensorboard 2.19.0 on Python 3.9 to 3.12
Support
scalar,image,figure,histogram,audio,text,graph,onnx_graph,embedding,pr_curve,mesh,hyper-parametersandvideosummaries.
pip install tensorboardX
or build from source:
pip install 'git+https://github.com/lanpa/tensorboardX'
You can optionally install crc32c to speed up.
pip install crc32c
Starting from tensorboardX 2.1, You need to install soundfile for the add_audio() function (200x speedup).
pip install soundfile
- Clone the files in https://github.com/lanpa/tensorboardX/tree/master/examples
- Run the demo script: e.g.
python examples/demo.py - Start TensorBoard with
tensorboard --logdir runs
# demo.pyimporttorchimporttorchvision.utilsasvutilsimportnumpyasnpimporttorchvision.modelsasmodelsfromtorchvisionimportdatasetsfromtensorboardXimportSummaryWriterresnet18=models.resnet18(False) writer=SummaryWriter() sample_rate=44100freqs= [262, 294, 330, 349, 392, 440, 440, 440, 440, 440, 440] forn_iterinrange(100): dummy_s1=torch.rand(1) dummy_s2=torch.rand(1) # data grouping by `slash`writer.add_scalar('data/scalar1', dummy_s1[0], n_iter) writer.add_scalar('data/scalar2', dummy_s2[0], n_iter) writer.add_scalars('data/scalar_group',{'xsinx': n_iter*np.sin(n_iter), 'xcosx': n_iter*np.cos(n_iter), 'arctanx': np.arctan(n_iter)}, n_iter) dummy_img=torch.rand(32, 3, 64, 64) # output from networkifn_iter%10==0: x=vutils.make_grid(dummy_img, normalize=True, scale_each=True) writer.add_image('Image', x, n_iter) dummy_audio=torch.zeros(sample_rate*2) foriinrange(x.size(0)): # amplitude of sound should in [-1, 1]dummy_audio[i] =np.cos(freqs[n_iter//10] *np.pi*float(i) /float(sample_rate)) writer.add_audio('myAudio', dummy_audio, n_iter, sample_rate=sample_rate) writer.add_text('Text', 'text logged at step:'+str(n_iter), n_iter) forname, paraminresnet18.named_parameters(): writer.add_histogram(name, param.clone().cpu().data.numpy(), n_iter) # needs tensorboard 0.4RC or laterwriter.add_pr_curve('xoxo', np.random.randint(2, size=100), np.random.rand(100), n_iter) dataset=datasets.MNIST('mnist', train=False, download=True) images=dataset.test_data[:100].float() label=dataset.test_labels[:100] features=images.view(100, 784) writer.add_embedding(features, metadata=label, label_img=images.unsqueeze(1)) # export scalar data to JSON for external processingwriter.export_scalars_to_json("./all_scalars.json") writer.close()
TensorboardX now supports logging directly to Comet. Comet is a free cloud based solution that allows you to automatically track, compare and explain your experiments. It adds a lot of functionality on top of tensorboard such as dataset management, diffing experiments, seeing the code that generated the results and more.
This works out of the box and just require an additional line of code. See a full code example in this Colab Notebook
To add more ticks for the slider (show more image history), check #44 or tensorflow/tensorboard#1138
![@dependabot[bot]](https://avatars.githubusercontent.com/in/29110?s=64&v=4){kind=small}
