It’s so funny you said that! Spent all night hacking on a version of the “teaching skills” feature with OI!
PROMPT ENGINEERING
-
Toolkit for Custom GPT Development and Pre-built Templates
By
–
I created a GPT Builder Toolkit with hundreds of copy and paste GPTs. Discover it here: https://
godofprompt.ai/custom-gpt-too
lkit
… -
Using ChatGPT to generate social media content from transcripts
By
–
Twitter content creation isn’t easy. But ChatGPT can generate you weeks worth of content in just 5 minutes. Here’s a mega-prompt to generate 50 tweets from a transcript: I want you to write tweets for me based on a transcript. These tweets are characterised by being
-
Retrieval-Augmented Generation Basics Explained
By
–
Good morning fellow AI enthusiast! This week's iteration, and first one of the year, focuses on RAG! We go back to the basics. https://
open.substack.com/pub/louisbouch
ard/p/what-is-retrieval-augmented-generation?r=25qlky&utm_campaign=post&utm_medium=web&showWelcome=true
… -
Focus on Red Teaming and Adversarial Data for LLM Post-Training
By
–
Red teaming and adversarial data for post-training. I used to do more of what most people imagine when they think of prompt engineering (helping businesses with LLM app development etc.) but demand for adversarial is insane right now so that’s my main focus
-
Raw completions and default system prompts in AI models
By
–
Agreed that raw completions is useful in many cases and that people want it. This is generally a model level issue, there’s no default system prompt in the API and you can see similar outputs.
-
Human Persuasion Techniques as GPT-3.5 Prompt Jailbreaks for Policy Violations
By
–
The success rate of 40 human persuasion techniques as GPT-3.5 prompt jailbreaks for violating each of 14 OpenAI usage policies:
-
SDXL Fine-tune Enables Realistic Tattoo Design Generation
By
–
New SDXL fine-tune – "fresh ink" https://
replicate.com/fofr/sdxl-fres
h-ink
… – make tattoos that look like they were just inked
– use inpainting to try out a tattoo
– use lora weights with controlnets to try out designs -
Retrieval Augmented Generation: Augmenting LLMs with Memory
By
–
Good morning AI enthusiasts! After a long break – here's the first video of the year! What is Retrieval Augmented Generation (RAG) – Augmenting LLMs with a Memory Learn more in the video: https://
youtu.be/LAfrShnpVIk p.s. the What's AI Podcast is starting again next week! -
RAG vs Fine-tuning: When to Use Each Approach
By
–
When should you use RAG vs fine-tuning? When you have a custom knowledge base and want a ChatGPT-like interface then RAG makes sense. It has multiple components that can make it tricky to get right but it’s easier to implement than fine-tuning. When you want the model to adapt