I'm not quite sure what this person means by "the decoder".

Rodrigo Carvalho <rodrigo.d.carvalho@kcl.ac.uk>

Dear reader,

I'm very interested in TimeGPT model, specifically in the decoder. I'm doing some research about brain decoding, for this matter I'm using a multi-modal architecture. I want to incorporate the latent space of TimeGPT. However, I can't find this information in the documentation. Is there any possibility of using just the model decoder?

Best,

Rodrigo

The person thinks it's a transformer behing TimeGPT, which made of an encoder and a decoder. So, the answer here is no, but I guess we don't want to divulge the fact that's it's not a transformer. At least not before the paper is updated.

@Marco beat me to it, but yes 😅

yeah, but you get more points for taking the time to get the paper and take a sreenshot

😄 thanks both!