In the context of N-gram language models, what is the 'sparsity problem'?

1. ক
   The model requires too much memory to store N-grams
   
2. খ
   Many N-grams that appear in test data never appeared in the training corpus
   
3. গ
   The model is too slow to compute conditional probabilities
   
4. ঘ
   The model fails to handle long-range dependencies