I'm just using the default Samsung Notes app and the default screen recorder. I've looked into paid options, but they don't really offer anything extra and I'm pretty happy with this setup. To do dual screen, I just half the screen and have the notes app up twice
Hey, I am a student and am trying to create a time series forecasting model. I'm totally new to ML models, I know the question I'm asking doesn't even make sense. But it would be great if you could explain. Using an LSTM model trained on data from the last three years, which takes the past ten days' data as input to predict the output for the eleventh day and beyond, differs from a decoder-only model for time series forecasting. If i am trying only to predict for next day.
By decoder-only, do you mean decoder-only transformer? If so, then the task idea is the same. One hard part with transformers for time series forecasting is encoding the data in a way that the model can utilize efficiently. For an LSTM, you can just keep throwing scalar or vector data into the hidden state, but for transformers, you need to find a way to tokenize the data in a meaningful way. However, once tokenized, you can probably treat training like text data, making it an autoregressive model.
I love you man. Keep it going, your backlog of videos is a keg of dry powder waiting to explode.
Keep up the good work, man!
I do not understand it, but can we use it to do stock market forecast, sports forecast. How do I use the tool, is there a tutorial for it? Thank you .
What tools are you using to make this paper reading session?
I'm just using the default Samsung Notes app and the default screen recorder. I've looked into paid options, but they don't really offer anything extra and I'm pretty happy with this setup. To do dual screen, I just half the screen and have the notes app up twice
Hey, I am a student and am trying to create a time series forecasting model. I'm totally new to ML models, I know the question I'm asking doesn't even make sense. But it would be great if you could explain. Using an LSTM model trained on data from the last three years, which takes the past ten days' data as input to predict the output for the eleventh day and beyond, differs from a decoder-only model for time series forecasting. If i am trying only to predict for next day.
By decoder-only, do you mean decoder-only transformer? If so, then the task idea is the same. One hard part with transformers for time series forecasting is encoding the data in a way that the model can utilize efficiently. For an LSTM, you can just keep throwing scalar or vector data into the hidden state, but for transformers, you need to find a way to tokenize the data in a meaningful way. However, once tokenized, you can probably treat training like text data, making it an autoregressive model.