search

timeseriesAI / tsai

Time series Timeseries Deep Learning Machine Learning Python Pytorch fastai | State-of-the-art Deep Learning library for Time Series and Sequences in Pytorch / fastai
https://timeseriesai.github.io/tsai/
Apache License 2.0
5.07k stars 633 forks source link

Question about which architectures can be used for which problem #691

Closed karllandheer closed 1 year ago

karllandheer commented 1 year ago

Hello, is there a list of which arch types work for regression? I found all arch type names in the code, however when I try some, for example LSTM, it gives invalid argument for custom_head, so I assume some aren't built for regression? Or is there something I'm missing for some architectures?

MOREDataset commented 1 year ago

I got the same problem! They used to work before. I think there must be an issue with the new update!

karllandheer commented 1 year ago

Another quick question: is there an easy way to print out the training curves to a pd dataframe?

karllandheer commented 1 year ago

Hello, another quick question: is there a easy way to set the x and y data for a model? Say you have a model saved and want to load it and re-train on a new set of data, can this be done easily? I don't see anything like set_data

oguiza commented 1 year ago

@karllandheer, Please, do not open unrelated issues within an issue. You may open a different issue or a discussion (that allows a chat).

oguiza commented 1 year ago

Hi @karllandheer, Can you please post a code snippet to reproduce the initial issue that you raised? I believe most (if not all) architectures should work on regression tasks,

oguiza commented 1 year ago

Closed due to lack of response