Movie Rating Prediction

#1

I initially built an lstm network which had 47,81,202 parameters to train,so it was taking too much time to train,So then i thought of adding pre-trained model–glove but since input dimensions were high like 40,000 sentences,1423 max len of padding and 50 dimensions of each word.So ram got crashed on google colab.I tried to keep the top 30,000 features.
Pls help me out in this.
Link to colab ntbk–
https://colab.research.google.com/drive/1QRpsMiK832x1pM_sIFnMlNC8_mA9o1PX

#2

Hey @Nik1, yes ram will get crash, you need to use something like train generator, it is covered in image captioning videos, go through them, and than build your own train_generator, than you will not face problem of ram crashing.