Question 1 [15 marks] a. What are Large Language Models? b. How do Large Language Models work? c. Provide potential use cases of Large Language Models in engineering.
Added by Elizabeth W.
Close
Step 1
They are trained on massive amounts of text data, allowing them to learn patterns and relationships between words and phrases. This enables them to perform various tasks, such as translation, summarization, question answering, and text generation. Show more…
Show all steps
Your feedback will help us improve your experience
Aparna Shakti and 56 other AP CS educators are ready to help you.
Ask a new question
Labs
Want to see this concept in action?
Explore this concept interactively to see how it behaves as you change inputs.
Key Concepts
Recommended Videos
Your product manager is working for a website that sells clothing fabrics. You're working on an Al natural language processing product that allows customers to do natural language searches. What will be one of the biggest challenges with your new Al system? Select an answer: Humans tend to describe things as qualities such as "dark" and "soft" which are difficult to translate into search results. Like computers, humans are very precise in their language and so it'll be difficult to record all that data. Natural language processing requires a lot of experts to program the system. Customers might be intimidated by the strong Al system that could ask difficult follow-up questions.
Aparna S.
Categorize which type of NLP (NATURAL LANGUAGE PROCESSING) application applies to each of the following use-cases: a. A model that allocates which mail folder an email should be sent to (work, friends, promotions, important), like Cmail's inbox tabs. b. A model that helps decide what grade to award an essay question. This can be used by a university professor who grades a lot of classes or essay competitions. c. A model that provides assistive technology for doctors to provide their diagnosis. Remember, doctors ask questions, so the model will use the patients' answers to provide a probable diagnosis for the doctor to weigh and make decisions. Think about your environment — school, work, or home — and think of a problem you face that can be solved by Natural Language Processing. Be as descriptive as possible and show what value it will add to the environment in question. For example, if you are a stand-up comedian, you can come up with a model that analyzes the topics and areas your audience responds to the most. This can be done by, say, identifying tweets (or Facebook posts) that have the most 'retweets' (or 'likes'). The tweets or posts will then be categorized into topics, and if you want to create a new comic routine, you can stick to the topics mined using your followers' reactions! So, think of a scenario that applies to your life/work/school and how you can apply NLP to your environment, and save that as nlp.txt. Read up on any innovative technology using NLP (by companies such as Google or IBM, for instance) and write a brief summary about the technology, what it achieves/does, and an overview of how it works. To take an example, you may have noticed Cmail's auto-response suggestions on your incoming emails. If I send an email to your Cmail address asking for an appointment, on opening the mail you would notice Cmail's automatically suggested response options such as "Yes, that works for me" and "Sorry, I'm not available at that time.
Supreeta N.
3. Recurrent neural networks, or RNNs, are a family of neural networks for processing sequential data and commonly used in natural language processing (NLP) and time series analysis (TSA) due to their ability to retain memory of previous inputs. RNNs have shown promising results in detecting patterns of potential money laundering activities by analyzing transaction data over time. In light of this, we pose the following queries: (a) (2 points) Suppose a training examples are sentences (e.g., a sequence of words from a given snippest text posted on social media). Which of the following refers to the ( j^{ ext {th }} ) word in the ( i^{ ext {th }} ) in the training examples? 1. ( x^{(i)<j>} ) 2. ( x^{<i>(j)} ) 3. ( x^{(j)<i>} ) 4. ( x^{<j>(i)} ) (b) (2 points) Consider this RNN: Figure 3: A recurrent neural network (RNN). To prevent overfitting and improve generalization performance. This regularization term penalizes large weights in the network, leading to a simpler model that is less likely to memorize noise in the training data. term to the weights. The new objective is: This specific type of RNN architecture is appropriate when: 1. ( T_{x}=T_{y} ) 2. ( T_{x}<T_{y} ) 3. ( T_{x}>T_{y} ) 4. ( T_{x}=1 ) (c) (2 points) You are training an RNN, and find that your weights and activations are all taking on the value of NaN ("Not a Number"). Which of these is the most likely cause of this problem?
Shu N.
Recommended Textbooks
Computer Science and Information Technology
Introduction to Programming Using Python
Computer Science - An Overview
Transcript
18,000,000+
Students on Numerade
Trusted by students at 8,000+ universities
Watch the video solution with this free unlock.
EMAIL
PASSWORD