Browsing is great for information retrieval and massively reduces hallucinations, but I feel that it is easy for models that browse to lose some of the “magic” of large language models. What I mean by magic is when language models give rich, organic responses reflecting the
PROMPT ENGINEERING
-
ASU Develops Personalized AI Tutors and Avatar Study Assistants
By
–
ASU plans to build a personalized AI tutor for students, allow students to create AI avatars for study help and broaden the university’s prompt engineering course.
-
Flow Engineering Boosts Code Generation Performance Significantly
By
–
Prompt engineering (or rather "Flow engineering") intensifies for code generation. Great reading and a reminder of how much alpha there is (pass@5 19% to 44%) in moving from a naive prompt:answer paradigm to a "flow" paradigm, where the answer is constructed iteratively.
-
GPT Database: Browse and Rank Custom GPTs by Niche
By
–
Lastly, for some inspiration and ideas on what GPT to build, check out our GPT database. It has every GPT you need to know about, including categorization by niche and ranking:
-
Create high-performing GPTs with these optimized prompts
By
–
Creating #GPTs on #ChatGPT is easy. But creating truly useful and high-performing GPTs is another story. Unless you use these prompts → https://youtu.be/8of5c3XidvA With them, you will create unique and optimized GPTs for the GPT Store.
-
Keep up with AI: learn prompt engineering with godofprompt
By
–
Keeping up with AI is easy. Just learn prompt engineering with @godofprompt and you are free of working as a slave.
-
Full prompt engineering guide from OpenAI
By
–
Here's the full prompt engineering guide if you want to explore it in more detail: https://
platform.openai.com/docs/guides/pr
ompt-engineering/strategy-test-changes-systematically
… -
Systematic testing and evaluation of ChatGPT outputs
By
–
Strategy 6: Systematically Test and Evaluate ChatGPT Outputs. Examples:
– Implement a model-based evaluation: check if ChatGPT's answers include specific known facts.
– Use a systematic approach: restate the fact, find the closest citation in ChatGPT's answer, and judge its -
Strategy 4: Give the Model Time to Think
By
–
Strategy 4: Give the Model Time to 'Think'. Examples:
– Instruct ChatGPT to develop a solution before concluding.
– Use inner monologue or queries sequence to aid reasoning.
– Ask ChatGPT to review previous responses for completeness. -
Strategy 2: Provide Reference Text for ChatGPT Answers
By
–
Strategy 2: Provide Reference Text. Examples:
– Instruct ChatGPT to use reference text for answers.
– Direct ChatGPT to answer with citations from reference text.