Please read at least one paper from each paper group. For each paper you read, submit brief answers for the following:

  • The title of the paper
  • Problem definition
  • Technical challenges
  • Summary of approach/main ideas
  • Evaluation methods and results
  • Identify limitations. How would you change the approach to improve?
  • Discussion points

Paper group 1:

  • Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, MarieAnne Lachaux, Timoth´ee Lacroix, Baptiste Rozi`ere, Naman Goyal, Eric Hambro, Faisal Azhar, Aurelien Rodriguez, Armand Joulin, Edouard Grave, and Guillaume Lample. LLaMA: Open and Efficient Foundation Language Models. 2023. arXiv: 2302.13971 [cs.CL]. Link

  • Hugo Touvron et al. Llama 2: Open Foundation and Fine-Tuned Chat Models. 2023. arXiv: 2307.09288 [cs.CL]. Link
  • Additional Question: How would you use LLMs for robotics?

Paper group 2:

  • Sai Vemprala, Rogerio Bonatti, Arthur Bucker, and Ashish Kapoor. ChatGPT for Robotics: Design Principles and Model Abilities. 2023. arXiv:2306.17582 [cs.AI]. https://arxiv.org/abs/2306.17582Links to an external site.
  • Additional Question: The above paper is one of the first works that used LLMs for robotics. Among the papers that cited the above paper, which one sounds most intriguing to you? Google Scholar