Query Analysis What do the following methods have in common? StepBack Prompting
HyDE
Query Structure They are all examples of **query analysis**, a technique to optimize queries to a RAG system We've added a brand new docs section on this! https://
python.langchain.com/docs/use_cases
/query_analysis/
…
PROMPT ENGINEERING
-
Query Analysis Techniques for RAG Systems Optimization
By
–
-
RAG Benchmarking Challenges and LLM Context Understanding Issues
By
–
Here is a great diagram that showcases the challenges with RAG benchmarking and LLM context understanding. RAG systems are complex because of these following 4 issues. Even stuffing the context of LLM does not really alleviate the issue and usually confuses the LLM. The
-
Claude-3 Steerability Features Excite Megaprompt User
By
–
For a megaprompt addict like myself, the new Claude-3 steerability sounds AMAZING. (Also, love that I may not get as many false-positive refusals, like when I ask the model to write a spikey point of view…)
-
ChatGPT Memory Function: Intriguing Yet Limited Capabilities
By
–
I tried ChatGPT's memory function and found it intriguing but limited The memory function in ChatGPT is like a fine-tuning procedure. Getting the results you want, however, can be frustrating at times. https://
zdnet.com/article/i-trie
d-chatgpts-memory-function-and-found-it-intriguing-but-limited/
… @OpenAI -
New Model Underperforms GPT-4 in Real Use Cases
By
–
It does some things worse than GPT-4 in real use cases we have played with, but that could be prompting. We don't know yet.
-
Prompting Art and Science: Practical Advice Guide
By
–
All of those weird prompting tricks (giving tips, threatening the AI) only work sometimes. The truth is that prompting is often more art than science, yet prompting is still very important. I try to reconcile these facts, and give some prompting advice:
-
GPT-4 powered weather-based activity recommendation system
By
–
1) I allow for sunny, partly sunny, partly cloudy (sneaky hack: I check for match text against “sunny” or “partly”) 2) Activity ideas are customized to me and the weather report and my location 3) Activity ideas are written by GPT-4 Total time spent: <1 hr
-
ChatGPT Debugging Tips: Test-Driven AI Development Approach
By
–
Learnings: Leverage ChatGPT and online tutorials Screenshot errors to ChatGPT for fixes Test one step at a time (ex: get weather, make sure that works, then move on), turn on all SHOW WHEN RUN in beginning then when perfected you can turn it off Try try try again!
-
PlanGPT: Multi-Approach Framework for Spatial Planning AI
By
–
10/ PlanGPT – combines multiple approaches like retrieval augmentation, fine-tuning, tool usage, and more; the proposed framework is applied to urban and spatial planning but there are a lot of insights and practical tips that apply to other domains.
-
LLMs as Practical Tools for Everyday Task Automation
By
–
LLMs act like Swiss Army knives, especially if you don’t code a lot I keep having GPT-4 solve tiny problems that would have been annoying, like removing line breaks from text that I copy from PDFs, or a little GPT that turns emails into .ics appointments