Marco Willi

Deep learning researcher.

[Marco Willi @mwilli_]

Happy New Year! In 2020, I trained 417 models over 1,851 hours, with 1,140 hours of GPU training time. Get insights on your model training with @wandb! #myyearwrapped

[Marco Willi @mwilli_]

Vielen Dank für die Unterstützung Monika! War extrem wertvoll bei deinem Kurs mitzumachen zu dürfen :) Nur zu empfehlen: #3WeeksofTwitt

[Marco Willi @mwilli_]

Cool music generated with VQ-VAE models @michaelhgraber & @3WeeksofTwitt

[OpenAI @OpenAI]

Introducing Jukebox, a neural net that generates music, including rudimentary singing, as raw audio in a variety of genres and artist styles. We're releasing a tool for everyone to explore the generated samples, as well as the model and code: https://t.co/EUq7hNZv62

[Marco Willi @mwilli_]

Interesting resources at: https://t.co/IuDi9H5nDe

[Kosta Derpanis (sabbatical in Zurich) @CSProfKGD]

Giant list of #DeepLearning/#MachineLearning/#ComputerVision online course material with videos

[Marco Willi @mwilli_]

Impressive image generation quality with quantization.

[Aäron van den Oord @avdnoord]

VQVAE-2 finally out!

Powerful autoregressive models in a hierarchical compressed latent space. No modes were collapsed in the creation of these samples ;)

Arixv: https://t.co/nQH5dzQhib
With @catamorphist and @vinyals

More samples and details 👇 [thread]

[Marco Willi @mwilli_]

Very interesting and useful to reduce the number of labels required.

[Aäron van den Oord @avdnoord]

Unsupervised pre-training now outperforms supervised learning on ImageNet for any data regime (see figure) and also for transfer learning to Pascal VOC object detection
https://t.co/MvRu3dqTxk

[Aäron van den Oord @avdnoord]

Unsupervised pre-training now outperforms supervised learning on ImageNet for any data regime (see figure) and also for transfer learning to Pascal VOC object detection
https://t.co/MvRu3dqTxk

[Marco Willi @mwilli_]

Interesting adversarial attack by employing "off-distribution" actions