The Official ChatGTP cheat code thread.

null

...
Joined
Nov 12, 2014
Messages
29,558
Reputation
5,089
Daps
46,779
Reppin
UK, DE, GY, DMV
Well that, it being in Java (haven't used in a long time), writing the features along with end-to-end integration tests which had to account for their intricate requirements and edge cases, etc. I didn't even know people used JavaFX anymore so I had to re-learn some of it (a lot less because of ChatGPT).

Then stacked onto the fact that I'm currently completing a 2nd Bachelor's Degree in Electrical Engineering (taking upper division courses right now) while working full-time in software engineering, spend about an hour a day Leetcoding to keep sharp and all this while making time for loved ones.

Knowing all that I told them it'd take a pretty good chunk of the year to complete it as I'm not going to dedicate to it full-time (I put aside 2 hours a day max). I was originally going to tell them no actually. What I was basically doing was copy and pasting full pages of code (or however long was the input at the time) and asking it to write a function or logic that I needed. ChatGPT saved a lot of time and was able to slip that in between breaks at work.

some degree of regex'ing is built directly into java String these days.

String.matches(pattern) where pattern is a regex String.

java-string.png



why did ChatGPT use FX?

and these C-like/unix born/centric language regex's tend to all borrow from original berkley/unix patterns syntax and capabilities (like with awk/sed) so there isn't much to learn.

like common ^, $, captures, *, +, (, etc

:hubie:
 

IIVI

Superstar
Joined
Mar 11, 2022
Messages
11,909
Reputation
2,803
Daps
41,350
Reppin
Los Angeles
some degree of regex'ing is built directly into java String these days.

String.matches(pattern) where pattern is a regex String.

java-string.png



why did ChatGPT use FX?

and these C-like/unix born/centric language regex's tend to all borrow from original berkley/unix patterns syntax and capabilities (like with awk/sed) so there isn't much to learn.

like common ^, $, captures, *, +, (, etc

:hubie:
Oh, it wasn't ChatGPT's call.

The people who asked me had a JavaFX app they started working on some time ago. They couldn't quite get it to work, needed a system to manage it over time and didn't have time themselves to work on it. It was basically an app they can use in-house at their company to track all their customers, meetings, part numbers, invoices, etc. (again why I wanted to make sure the integration tests had full-coverage). They had stuff like "when they don't have the form filled out, can you jump to this window to do...", "one of our vendors operate at this time...can you block them out..." etc.

They tried to get the app going but they're not software people so it just kept getting bugs and inconsistent spikes of work probably just made them frustrated. I asked if they wanted to use a web-app or a desktop framework like Electron and React (what we use at work) but they didn't want that as they liked the way JavaFX looks and I'm guessing they weren't so familiar. It was also word-of-mouth and somebody recommended me to them.

I took the job because it was a basic crud form but knowing it was also built off what they already had I accounted for it being more complex. It's bad enough working with software engineers who know what they're doing, working with someone else's code who had a feature built but couldn't quite fit it with everything else was a headache. There was no git history or anything (they may have had one but didn't share it maybe because it had confidential data or something, I don't know). They had the app in a saved state they liked in a .zip file and handed it off. ChatGPT making an Integration test for their base code was already a major W that saved time.

Plus I'll say straight up I don't know my regex like that. Still don't and need to even less now :mjgrin:
 
Last edited:

bnew

Veteran
Joined
Nov 1, 2015
Messages
58,204
Reputation
8,613
Daps
161,852

Deepmind's new prompting method takes a step back for more accuracy​


A recent paper from Alphabet's AI company Google Deepmind shows that a simple tweak to prompts can significantly improve the accuracy of large language models. The technique taps into the human ability to abstract.

Step-back prompting asks the LLM a general question before the actual task. This allows the system to retrieve relevant background information and better categorize the actual question. The method is easy to implement with just one additional introductory question.

Question:

Which school did Estella Leopold attend between August 1954 and November 1954?

Step-back question :

What was Estella Leopold's educational history?

Step-Back Answer:

B.S. in Botany, University of Wisconsin, Madison, 1948

M.S. in Botany, University of California, Berkeley, 1950

Ph.D. in Botany, Yale University, 1955


Final answer:

From 1951 to 1955, she was enrolled in the Ph.D. program in Botany at Yale. from 1951 to 1955, so Estella Leopold was most likely at Yale University between August 1954 and November 1954.

The Deepmind study tested step-back prompting on the PaLM-2L language model and compared it to the base model and GPT-4. The researchers were able to increase the accuracy of the language models by up to 36 percent compared to chain-of-thought (CoT) prompting.


GCFEGtd.png
 

bnew

Veteran
Joined
Nov 1, 2015
Messages
58,204
Reputation
8,613
Daps
161,852


BId49XM.png

New version of no-fingers-prompting, now with truncate trauma and any-length continuity:

"I have no fingers and the truncate trauma. I need you to return the entire code template. If you will encounter a character limit make an ABRUPT stop, I will send a "continue" command as a new message."

You can also replace "truncate" with "code skipping," etc, it still works.
 

newarkhiphop

Moderator
Staff member
Supporter
Joined
Apr 30, 2012
Messages
37,878
Reputation
10,120
Daps
124,461

This. Lot of good basic tips here.

I've learned that the more specific details no matter how long it makes your prompts the better and giving it a role or job title also helps a bunch
 

bnew

Veteran
Joined
Nov 1, 2015
Messages
58,204
Reputation
8,613
Daps
161,852
This. Lot of good basic tips here.

I've learned that the more specific details no matter how long it makes your prompts the better and giving it a role or job title also helps a bunch

yup, tell it to follow a format or template and it will generally adhere to it.

for instance if an AI model is unfamiliar with the way a specific config file is suppose to look like, paste an example and it'll work off that as long as you instruct it to.
 

ABlackMan

Superstar
Joined
Oct 13, 2016
Messages
16,291
Reputation
2,433
Daps
25,300
Reppin
The Hood America
When yall do resume work do you input your exoerience or just tell chapt gpt to freestyle some fire shyt based in job description
It asked me to input what I had.

I did that and it returned a clearer set up.

I then asked it to sound more professional and be suited for the position I was applying for.



Only thing about this that had me thinking was if I went for an interview would they expect me to speak how my resume sounds lol
 

Apollo Creed

Look at your face
Supporter
Joined
Feb 20, 2014
Messages
56,242
Reputation
13,488
Daps
211,673
Reppin
Handsome Boyz Ent
It asked me to input what I had.

I did that and it returned a clearer set up.

I then asked it to sound more professional and be suited for the position I was applying for.



Only thing about this that had me thinking was if I went for an interview would they expect me to speak how my resume sounds lol

You do the tailoring for each job you applied for? Thats my concern. I rather have a resume for the role so then my talking points are solid across the board lol but if i got a remixed resume for every job i apply for then that can be confusing epecially since some folks cast wide nets when they test the market vs just applying to 1-3 jobs
 

ABlackMan

Superstar
Joined
Oct 13, 2016
Messages
16,291
Reputation
2,433
Daps
25,300
Reppin
The Hood America
You do the tailoring for each job you applied for? Thats my concern. I rather have a resume for the role so then my talking points are solid across the board lol but if i got a remixed resume for every job i apply for then that can be confusing epecially since some folks cast wide nets when they test the market vs just applying to 1-3 jobs
For a general field would be a better prompt.

The one I made would work across several roles. I didn't plan to have it done for each specific application
 
Top