khuangaf/CryptocurrencyPrediction ? reverse-engineered prompt

Build me a Python project that predicts Bitcoin prices with deep learning. I want it to pull historical crypto price data, preferably from Poloniex if that still makes sense, clean and preprocess it, then train and compare an LSTM, a GRU, and a 1D CNN, along with simple baseline and linear regression models. Use the past 256 time steps as input and predict the next 16, based on five minute data, with proper scaling before training.

Please include the full flow, from data collection to dataset creation, training, evaluation, and plots of the results so I can visually compare the models. I also want saved result images and a simple way to run each model from scripts or notebooks. Try to match the kind of outputs shown in the repo, especially the comparison charts and final prediction plots.

If the original setup is very old, update it carefully so it still follows the same idea and behavior. Look up current docs online if you need to.