Chat GPT
How do you create an effective prompt in ChatGPT?
- Be Clear and Specific - Provide context so ChatGPT can understand the situation - Keep in Concise: ChatGPT might have a harder time understanding longer text. - Use Proper grammar and punctuation - Avoid Leading: Try to avoid leading or biasing the model by providing too much information about the topic - Test and Refine: Test your prompt and refine it as needed.
When creating a new chat, What three rows show on your screen?
- Examples (Things you can choose to lookup) - Capabilities (What chat GPT can do) - Limitations (What chat GPT can't do)
ChatGPT vs ChatGPT plus
- The biggest difference between ChatGPT vs ChatGPT plus is speed. ChatGPT plus is much faster with response time. - ChatGPT plus is always available even when the demand is high.
What is a prompt/query in Chat GPT?
A text input that you send to ChatGPT's model and the model will generate a response based on that Input.
How do you regenerate a prompt response?
After the prompt response has been fully generated or stopped, you click the "regenerate response" button located above the chat GPT search bar.
What is an embedding?
An Embedding is a vector representation of a piece of data that is meant to preserve aspects of it's content and its meaning. Chunks of data that are similar in some way will tend to have embeddings that are closer together than unrelated data.
How do you stop a prompt response from fully generating?
By clicking the "Stop Generating" button located above the chat GPT search bar.
How does Chat GPT work?
ChatGPT is a language model developed by openAI that uses artificial intelligence and machine learning algorithms to generate Human like responses to text inputs.
What are embeddings useful for?
Embeddings are useful for search, clustering, recommendations, anomaly, detection, classifications.
What are tokens?
GPT and embeddings models process text in chunks called tokens Tokens represent commonly occurring sequences of characters.
What does GPT mean?
Generative pre-trained transformer
What are the four playground models?
Text-devinci: Most capable model Text-Curie: Very capable, but faster and lower cost than text-devinci Text-Babbage: Capable of straightforward tasks, very fast, and lower cost. Text-ada: capable of simple tasks, usually the fastest model in the GPT-3 series, and the lowest cost.