Questions for LLaMA: Open and Efficient Foundation Language Models
-----------------------------------------------------------------------------------------------

Please send your answers to: bravoma@cs.uni-freiburg.de by 13:30 on 07.06.2022
    
    Question 1. Name the three major contributions of the paper. (~3 sentences)

    Question 2. The authors implemented an efficient causal multi-head attention. In your own words, explain how this attention is modified and why it is more efficient compared to the traditional approach. (~2 sentences)

    Question 3. In Figure 2, the plots depict the models' performance on various tasks during training. However, for SIQA and WinoGrande, the models' behavior does not follow to the "larger model, better performance" rule. What do you believe could be the reason behind this discrepancy? (~2 sentences)