11/24/23: Happy Thanksgiving!

Machine Learning Study Group

Welcome! We meet from 4:00-4:45 p.m. CT. Anyone can join. Feel free to attend any or all sessions, or ask to be removed from the invite list as we have no wish to send unneeded emails of which we all certainly get too many.

 Contacts: jdberleant@ualr.edu and mgmilanova@ualr.edu

---

Agenda & Minutes

90th Meeting, Nov. 24, 2023

• Schedule
• Today: we completed "week" 1 of the 3-week quasi-cycle (which consists of reading in week 1, and HuggingFace labs in weeks 2 and 3, interspersed with weeks with other activities that happen from time to time; also a week might take more than a week if it seems warranted).

• Current activities
  
• Readings: We are reading
  
• Sparks of Artificial General Intelligence: Early experiments with GPT-4 (https://arxiv.org/abs/2303.12712). 
• Today we got through "makes us comfortable with saying that GPT-4 is a significant step towards AGI." on p. 4. Next is to examine Figures 1.1 and 1.2, then to continue with the partial par. at the bottom of p. 4.
  
• Lab exercises:
  
• We are continuing with the HuggingFace lab program. We did question answering last time we worked on this, and will go on to the next item. 
• See chapter 1, part 3, https://huggingface.co/learn/nlp-course/chapter1/3?fw=tf.
  

 

11/17/23: Talk by AM

Machine Learning Study Group

Welcome! We start at 4:00 CT, and go for about 40 minutes. New members are welcome and anyone can join. You may attend any or all sessions. Feel free to ask to be removed from the invite list at any time, as we have no wish to send unneeded emails of which we all certainly get too many. To join, send a request to jdberleant@ualr.edu or mgmilanova@ualr.edu.

 

Agenda & Minutes

89th Meeting, Nov. 17, 2023

• Schedule
• Today: talk by AM. Attendees thought the presentation was very well done! Congratulations.
  
• Next time we will try to complete "week" 1 of the 3-week quasi-cycle (which consists of reading in week 1, and HuggingFace labs in weeks 2 and 3, interspersed with weeks with other activities that happen from time to time; also a week might take more than a week if it seems warranted).

• Current activities
  
• Readings: We are reading
  
• Sparks of Artificial General Intelligence: Early experiments with GPT-4 (https://arxiv.org/abs/2303.12712). We read the abstract to decide whether to read more of the article. Suggested by MC. 8/18/23 vote was 4 6/7
• We are up to "Figure 1.1, we display some preliminary examples" in the second paragraph in section 1. 
• Lab exercises:
  
• We will continue with the HuggingFace lab program. We did question answering last time we worked on this, and will go on to the next item. 
• See chapter 1, part 3, https://huggingface.co/learn/nlp-course/chapter1/3?fw=tf.
  

 

11/10/23: Discuss Thomason lecture; question on code generation using LLMs; a bit of reading

Machine Learning Study Group

Welcome! We start at 4:00 CT, and go for about 40 minutes. New members are welcome and anyone can join. You may attend any or all sessions. Feel free to ask to be removed from the invite list at any time, as we have no wish to send unneeded emails of which we all certainly get too many. To join, send a request to jdberleant@ualr.edu or mgmilanova@ualr.edu.

 

Agenda & Minutes

88th Meeting, Nov. 10, 2023

• Schedule
• We are in "week" 1 of the 3-week quasi-cycle (which consists of reading in week 1, and HuggingFace labs in weeks 2 and 3, interspersed with weeks with other activities that happen from time to time; also a week might take more than a week if it seems warranted).
• We will discuss a problem to be posed by MH.
• Next week we will continue with the reading, to go with the flow.
• The following week will be lab activity. We did question answering last time. See chapter 1, part 3, https://huggingface.co/learn/nlp-course/chapter1/3?fw=tf. We will go on to the next item.
• The subsequent week we will continue on the HuggingFace lab program. 

• Today
• Updates. Today's talk "Language-Guided Robots and Language Models" by J. Thomason.
• Used syntax to disambiguate polysemous words.
• Maybe there is a Markov chain involved in the algorithms.
• "I saw a _____ climb a tree." We agree with LLMs that it's cat more than dog, but for different reasons - we understand the world but LLMs look at word co-ocurrence frequencies.
  
• What color is a sheep? Us - white. LLMs - black. (GPT4 did better though).
• Robotics control with NL now no longer requires a restricted vocabulary, as it did a few years ago.
• Using LLMs for code generation. See WizardCoder generative AI for example.
  
• Readings: We are reading
  
• Sparks of Artificial General Intelligence: Early experiments with GPT-4 (https://arxiv.org/abs/2303.12712). We read the abstract to decide whether to read more of the article. Suggested by MC. 8/18/23 vote was 4 6/7
• Today we got up to "Figure 1.1, we display some preliminary examples" in the second paragraph in section 1. We can start from there next time.

The meeting ended here.

 

10/3/23: 87th Meeting

 Machine Learning Study Group

Welcome! We start at 4:00 CT, and go for about 40 minutes. New members are welcome and anyone can join. You may attend any or all sessions. Feel free to ask to be removed from the invite list at any time, as we have no wish to send unneeded emails of which we all certainly get too many. To join, send a request to jdberleant@ualr.edu.

 

Agenda & Minutes

87th Meeting, Nov. 3, 2023

 

It was a good meeting! But let's be honest, do you really care what the details of the meeting were?