Ask me your LLM prompting/fine-tuning questions. Have an hour, will answer as many as I can.
PROMPT ENGINEERING
-
Better LLM Reward Model Performance Without Numerical Scales
By
–
If you’re using out of the box LLMs as a reward model/sample rater, here’s a trick to get way better performance: Don’t ask the LLM to rate examples on a numerical scale (i.e. 1-5). The model will almost always choose 1 or 5. Instead, use words as rating options (“very bad”,
-
Fine-tuning LLMs: Progressive Data Selection Strategy
By
–
High alpha technique for fine-tuning LLMs: First, fine-tune your model on all your data Then, use AI to pick a smaller set of really good examples, and tune on them Finally, hand-pick and tune on a few hundred incredible examples >> Huge performance gains <<
-
Prompt template for AI-generated landing page copy
By
–
Here's the prompt:
— God of Prompt (@godofprompt) 23 mars 2024
"Using Russel Brunsons' "Hook-Story-Offer" framework, I want you to write a landing page text for me.
For context, [INSERT CONTEXT]
My backstory is: I used to be a [INSERT LOW POINT], then I found [YOUR PRODUCT / SOLUTION], and now I help others do the same… pic.twitter.com/eO2GGv2xIJHere's the prompt: "Using Russel Brunsons' "Hook-Story-Offer" framework, I want you to write a landing page text for me. For context, [INSERT CONTEXT] My backstory is: I used to be a [INSERT LOW POINT], then I found [YOUR PRODUCT / SOLUTION], and now I help others do the same
-
Using Prompt Engineering for Landing Page Creation
By
–
We're going to apply a framework in a prompt to guide us in crafting a high-converting landing page. Here’s how:
-
AI Limitations: Self-Explanation and Jailbreaking Techniques
By
–
Continuing on the theme, a microcosm of why AI is weird – I ask it to explain itself and it makes some stuff up as justification (AI can't interrogate its own thoughts). Then I essentially do some lightweight jailbreaking by using its own logic against it.
-
Emergent AI Guardrails: Judgment Without Direct System Prompts
By
–
I am almost certain there is no direct system prompt or tuning that makes Claude not want to tell you stories featuring humans and dinosaurs, it is just some weird emergent thing. This is why guardrails seem so challenging, a lot of them are the AI "using its judgement," oddly.
-
HyperWrite: AI Writing Assistant with Fast Internet Access
By
–
Use HyperWrite, it’s a really great assistant (especially for writing) + has super-fast internet access
-
Large Models Improve Smaller Model Performance via Prompt Engineering
By
–
Rohit just merged this to the `gpt-prompt-engineer` repo. Now, you can use any large model to improve the performance of a smaller model! https://
github.com/mshumer/gpt-pr
ompt-engineer
… -
Long Context LLMs and RAG: Exploring Limitations
By
–
RAG for long context LLMs: Will long context LLMs really kill RAG? This is a talk @RLanceMartin gave at a few recent meetups that pulls together threads from a few different projects related to this question. Multi-needle in a haystack shows limitations in long-context