Experimenting with using a language model to improve the input prompt, then use that output as the actual prompt for the model, then returning the result. It’s a bit of a play on the “critique” approach. Some of the outputs were interesting but I need a better way to evaluate the results.
import sys import openai MODEL = "gpt-3.5-turbo-16k" IMPROVER_PROMPT = """ You are an expert prompt writer for a language model.I’ve been working through a series on nix-flakes. It’s well written and shows some interesting applications of the tool set. I’m still trying to wrap my head around exactly where nix could fit in in my development lifecycle. It seems like it wraps up builds and package management into one. Sort of like docker, bazel, pip/npm/brew all in one. The tutorial has shown some useful variations and has convinced me flakes is the way to go, but I need to spend some more time better understanding the primitives as well.Facebook (Meta, whatever) announced Threads today to launch on July 6th. Given how much worse it feels like Twitter has become (my experience only), on one hand, I could see people migrating here because no great alternative has really emerged. On the other, Facebook has zero “public” products where the user experience is even palatable for me, personally (I use Whatsapp but it’s basically iMessage). Instagram and Facebook both rapidly became completely intolerable for me due to their content.A simple shell function to setup a Python project scaffold. It’s idempotent, so it won’t overwrite an existing folder or env.
pproj () { mkdir -p $1 cd $1 python -m venv env . env/bin/activate }I’ve been following Jason’s working experimenting with different abstractions for constructing prompts and structuring responses. I’ve long felt that building prompts with strings is not the type of developer experience that will win the day. On the other hand, I’m weary of the wrong abstraction that would move the developer too far away from the actual prompt, which would make it harder to construct good prompts and steer the model. I’m not sure if this is an ORM vs.Did some work with Clojure destructuring.
Unpack values into specific variables.
user=> (let [[a b c] [1 2 3]] (println a b c)) 1 2 3 nil Unpack the first N items, ignoring the rest.
user=> (let [[a b] [1 2 3]] (println a b)) 1 2 nil Unpack the first N items to variables and capture the rest as an array.
user=> (let [[a b & rst] [1 2 3 4 5]] (println a b rst)) 1 2 (3 4 5) nilDoing math with a non-big decimal number and a big decimal number can cast down.
user=> (* 0.1 101M) 10.100000000000001 user=> (bigdec (* 0.1 101M)) 10.100000000000001MHeard the phrase “if someone wins the lottery” used today to describe a teammate leaving a team. I much prefer this to the more morbid alternatives.
I tried gpt-engineer today. I liked the approach and the setup instructions are good. I think I remember needing to use Python 3.11 instead of 3.8 that I was running, but beyond that the readme instructions we on point.
Process You start by creating a project folder with a plaintext prompt.I’ve been thinking about the concept of “prompt overfitting”. In this context, there is a distinction between model overfitting and prompt overfitting. Say you want to use a large language model as a classifier. You may give it several example inputs and the expected outputs. I don’t have hard data to go by, but it feels meaningful to keep the prompt generic or abstract where possible rather than enumerating overly specific cases in a way that obfuscates the broader pattern you’re hoping to apply.Richard WM Jones booted Linux 292,612 to find a bug where it hangs on boot. I loved reading the recounting of his process to accomplish this, by bisecting through the different versions of Linux and boot each thousands of times to determine whether the version contained the bug.
Georgi Gerganov started a company, ggml.ai, to run language models on commodity hardware.