We're getting closer to the point of wanting to search through a lot of hyperparameters for our full model. It looks like it's still feasible to just train it on my computer, but me and viva were discussing whether we should look into being able to train a lot of models simultaneously using some cloud service when we need to deal with hyperparameter optimization. As what to do for this is open ended, discussion on this would be helpful. We may not need it for this project, but might end up finding it helpful in the future to have a convenient way to train a lot of models with the only difference being hyperparameters.
We're getting closer to the point of wanting to search through a lot of hyperparameters for our full model. It looks like it's still feasible to just train it on my computer, but me and viva were discussing whether we should look into being able to train a lot of models simultaneously using some cloud service when we need to deal with hyperparameter optimization. As what to do for this is open ended, discussion on this would be helpful. We may not need it for this project, but might end up finding it helpful in the future to have a convenient way to train a lot of models with the only difference being hyperparameters.