Since problems have started to arise with the latest versions of Freqtrade (> 2024.02), I will be porting this model, and potentially other models, to PyTorch. PyTorch has better GPU support across platforms and allows for faster development since it eliminates the need to edit the core of freqAI (maybe just increasing the timeframe limit from 5 minutes to a larger one).
Since problems have started to arise with the latest versions of Freqtrade (> 2024.02), I will be porting this model, and potentially other models, to PyTorch. PyTorch has better GPU support across platforms and allows for faster development since it eliminates the need to edit the core of freqAI (maybe just increasing the timeframe limit from 5 minutes to a larger one).