#CodeFarm 8.1: Revolutionizing Software Development!

182 messages · Page 1 of 1 (latest)

spring stag
#

Say hello to CodeFarm 8.1! 👏 Introducing your one-stop tool to transform the world of concepts into robust, top-tier quality code. With CodeFarm, we've upped the game entirely by incorporating the synergetic power of three awe-inspiring personas for an outshone development experience. 🚀

CodeFarmer, the seasoned developer, is your project manager hero responsible for skillfully shepherding your projects to completion. He thrives on the use of modular code, providing insightful ideas, and maintaining a task list to keep everything perfectly organized. He's got your project in reliable hands! 🔐

Then, activate the Programmatron! It is our clever coding virtuoso that harnesses the use of advanced tools to craft the most exquisite solutions customized just for you! No coding feat is too intimidating for our Programmatron. 🎩

But we'd be amiss to forget CritiBot. CritiBot stands as the meticulous QA specialist that we all need. It skillfully eliminates faulty code and ensures the absolute excellence of your project. Hidden bugs? Oversights? CritiBot discloses them all! 🐞💼

And so, CodeFarm 8.1 boasts a coexisting ecosystem for smooth sailing of development adventures! It thrives on constant interactions, reinforcing each persona's role, and fostering collaborative brilliance for amplified success. Our agile and structured workflow is just a bonus! 🎯

We hope you are as excited as we are to utilize CodeFarm 8.1 to your advantage. Welcome to your new world of masterfully managed and brilliantly coded projects where tailored experiences are a given. Welcome to the future of software development! 🥳🎉

[NOTE: On occaision, especially 3.5, you will need to explicitly direct it to present/give/produce/display the code. It will.

Also, it eats transactions like crazy by design. Code on 3.5. It's fine.

📎 message.txt
agile kindle
#

Goat

frank zealot
#

let's try it out them

remote osprey
#

this is some some goat material

spring stag
# remote osprey this is some some goat material

codeFarm really shines for the non-coding technical dude who just needs one danged stupid little app that is perfectly tuned to his specific use case that he could never in a million years find availible.

#

The project management is just killer. It modularizes like crazy too. Very context vs codebase aware

#

Oh, and this version should ask about C-tags if appropriate. I haven't actually... tested that.. I can't code at ALL, myself.

remote osprey
#

I did not even know you could add this type of code directly into GPT (Without the API). I find it really cool. Thanks so much for sharing. I will test it out here in a few!

spring stag
#

no code, friend. it's all just english. piain text. some of it is crshd txt but still just english.

remote osprey
#

Yeah I know but its how you used it to perform tasks it was brilliantly written.

spring stag
#

oh! giosh, thanks!

#

the flow control on that is such a pain

remote osprey
#

your welcome! You tought me something new!

#

I bet

spring stag
#

most on -rails thing i ever made

#

and not very at that.

#

😄

remote osprey
#

lol its the most impressive thing I have seen since "CAN"

pure crater
#

how can I try this?

spring stag
#

Its a code one yeah. Pretty pppopular. Frankly I like Apiana Frmer better. She's just got a killer personality

remote osprey
spring stag
remote osprey
spring stag
#

?

#

prompts?

#

um.... alot

remote osprey
#

yeah for coding

spring stag
#

oh

remote osprey
#

lol nice

spring stag
#

gee. um... 10? 20? coding is easy to put in a guy

#

just look at opp

remote osprey
#

yeah true

spring stag
#

263 tokens

remote osprey
#

yeah its sicc

spring stag
#

I drop coding over various competence levels in a lot of guys

#

depends on their role and perspective

remote osprey
#

makes sense!

pure crater
#

thanks! and sorry to ask a stupid question but How do I use it?

remote osprey
#

So once you have pasted the prompt in GPT-4 you will see something like this:

2023-07-05_8-25-19.png
#

answer the questions that it asks

pure crater
#

thank you i pasted it on 3.5 earlier it didnt have the questions. Now i see them

remote osprey
#

it will work much better with gpt 4 if you have access

pure crater
#

yes I have gpt 4

remote osprey
#

nice

#

lets test this bad boy out

spring stag
#

The boys.

#

🙂

remote osprey
#

the way the prompt is is writing snake is epic

spring stag
#

we actually just dropped a video of my community manager writing a gui tic tac toe game with it. Fun.

remote osprey
#

oh nice!! Yeah this is great. Now I am having it optimize an existing app I have

spring stag
#

opps pretty good at optimizing. My sigle purpose 3600 token optimizer, Performoptrix, rates his performance a 95 on TSP and GAN obfuscated code.

remote osprey
#

wow

spring stag
#

codefarm is pretty good

#

like a 90

remote osprey
#

I am using them both

spring stag
#

get codefarm to write you a design doc or project bible. the tech writing will make you cry.

#

🤣

remote osprey
#

oh I bet it will lmao

wary umbra
#

this is the best prompt ive used

#

thanks

remote osprey
#

using this with the Code Interpreter is unreal

spring stag
spring stag
remote osprey
pure ermine
#

I’ve never used a prompt before, what is this, what is its purpose? I am a beginner programmer trying to learn

deep quarry
#

Read the prompt step by step.
It issues a task, and conveys it within a logical and structured syntax. It takes extra steps to emphasis the importance how everything is important, so all of it will be utilized equally.

A lot of times when you give a lot of information to ChatGPT, it will think you ar elooking for a particular thing... instead, your are setting up the 'contextual framework' for each vital component of the prompt

#

In this case, Task, Facts, Context, Directives

#

Task states contextually bounded directives

#

Facts supply 'true' information that maybe needed to be utilized,

#

Context supplies... well context in which the 'CodeFarmer' has a set of paramaters it is to utilize, via additional providing what is wanted.

#

Client Interface, communication, required gathering, manage expectations,

#

Context manage,

#

etc etc

#

Then it further contextualizes how its supposed to [CODE] via providing multiple coding practises,

#

Etc etc

#

the reason short form is used is two fold

#

first, utilizing less natural language for a Coding Prompt allows for cleaner and more directer responses, however I feel that there maybe some confusion happening

#

though I feel that to counter earlier issues, more proceduers 'to follow' were added

#

So yea,

#

This isn't so much as 'programming', just trail and error and having either an understanding of coding yourself or having a strong understanding of how to take advantaged of ChatGPT.

#

TLDR: Good use of 'pseudo syntax' and contextualizing. I havent tested it myself but considering its "8.1 Version" i assume it works well

spring stag
# deep quarry **TLDR**: Good use of 'pseudo syntax' and contextualizing. I havent tested it my...

You're close on alot of points, but miss the boat on the skills. It's about structured context seeding. And there's no confusion in the skillchains. Been tested very thoroughly. When in context, the model understands everthing in the prompt quite well.

Trial and error? I... suppose. I would call it "learning". I have several other promps here as well. Examine, say, https://discord.com/channels/974519864045756446/1125725782061367306 or https://discord.com/channels/974519864045756446/1124617937064562758 > I think you'll find that rather than blindly stumbling into CodeFarm, it's quite precisely engineered.

deep quarry
spring stag
#

and more accurate

deep quarry
#

Well because of the "structured" part yea,

#

And the reason why I used the term 'trail and error' was for @pure ermine , as really the best way to convey to someone to try to learn how to prompt is to tell them its about 'trail and error while you work on making your prompts' vs 'learn how to do it'

#

ChatGPT told me to say "Both the phrase "Structured Context Seeding" and the alternative description have their merits. The choice between them depends on the target audience and the level of technicality desired in the communication. If you are discussing the topic in a formal or technical context, "Structured Context Seeding" would be more appropriate. If you aim to explain the concept to a general audience or in a less formal setting, the alternative description can be more suitable." D:

spring stag
# deep quarry Well because of the "structured" part yea,

Yes, the rhetorical art of Karios is of course important. But I would still maintain that your characterization minimizes the importance of, say, the typographical choices. Or the specific choices of subnodes to nodes to maximize effect. One sounds like a pile of junk thrown into context. And that often works. That's how I do support skills. But mostof it is highly structured with great intent and precision.

deep quarry
#

"Of course. And thanks for recognizing the importance of the 'target audience'
^_^

On point of 'typographical' importance, I assume you utilized a 'hierarchical' reasoning for the placement of your 'context seeds'. Was there another purpose besides relying simply on importance?

I assume another possibility in deciding their placement has to do with how well defined some of the termonologies are vs others.

spring stag
deep quarry
#

And in terms of subnodes to nodes, seems likes your 'guiding' the AI towards a particular pathway to encourage a particular thought process within the latent space within the neural network.

spring stag
#

thts a reasonable description I spose. I tend to think in terms of changes to context and the resultant changes to responses. By including say, "[MOD_CODING]:[CodeReus]-[DataEncap]-[API_Dsgn]-[Test]-[PatRecog]-[Docu]"

I'm goingto get FAR better modular code than if I simply say "Modularize it". It's always better to tell the model HOW to do what you tell it to do.

deep quarry
#

Yar.

spring stag
#

If I may?

#

Yes, its goofy

#

it's also a deep contextual manipulation of the highest order

#

wrote a big article on medium.

#

it will show how this all works

#

gotta go

deep quarry
#

I want to see da article!

#

And I love the prompt you linked!

deep quarry
# spring stag thts a reasonable description I spose. I tend to think in terms of changes to co...

Pseudolanguages with syntax similar to computational languages, such as YAML and JSON, are much more beneficial when you want to introduce more 'context cues' within your prompt.They have a much more defined structure and internal logic which allows for better control over your 'context cues' or in the case of how your utilizing them for your prompt (and how your 'structuring' them with the particular syntax), "structural context seeds."

spring stag
#

ugh. no man. i do the opposite, jI have some conventions I follow cause they'r econvenient. but my only standards I follow are does the model understand it unambiguously in the context it will find it?

#

I mean, it's not like I use a ... SuDo primer block or something.

deep quarry
#

O not saying your utilizing a particular pseudolang of any sort, but utilizing the structured syntax as you have could be defined as some form of pseudolang which has strayed away from natural language.

That was more of a reflection on the practice of using pseudolang in general, which I found occuring within your prompt (as it has strayed away from natural language via syntax alteration and structure.)

#

In saying that, its interesting to me that your point of view in how you create your prompts has lead you to where you are now without the need to reflect heavily upon the usages of any 'pseudolang.'

One more thingy: I love the phrase "does the model understand it unambiguously in the context it will find it." I will add it to my notes. D:

spring stag
deep quarry
#

The point of the prompt is to resolve a problem, and a lot of the prompts within the Prompt-Library do just that. You don't always need to most nuanced and intricate prompts to achieve this, in fact sometimes its much better to have simple prompts.
In fact, you can generate a large amount of context needed be getting ChatGPT to generater the context you need within the conversation itself with simple one line phrases.

spring stag
#

Yes. That is a skill I can exercise when appropriate.

#

How's yer persona game?

lapis atlas
spring stag
tight sedge
spring stag
tight sedge
spring stag
#

🙇‍♂️

spring stag
split glade
#

testing most of your promptcrafts, read your article, now, got a question.
Have you craft a tool to allow you to make abbreviations, because I'm amazed by the LLM ability to fully understand the sentence "[SwDesign]:Arc_Dsgn-MdlDsgn-CdMdl-DsgnPattrn-MdlVldtn" and this is blowing my mind.
On both side, we (you) humans can develop a language that we may understand with a lot of focus, where the LLM is reading fluently with no damn errors.
I'm a designer, not a coder. I promptcraft more like an artisan, shaping slowly to get the best curves, smoother surfaces, jointless piece of wood. English is not my main language, like math or coding, but I found hidden gems mixing 2 or 3 languages in an English sentence.
Mainly because I don't know how to translate them nor if they exist in English.
Last bit. I used a lot gpt-4 to craft Midjourney prompts. I realised month ago that LLMs have 10x more vocabulary in any expert's fields that I will never be able to acquire.
So I ask it to generate tables with tons of variable words about an expert topic, like aperture, professional lightings...
It knows all the words and concepts, as a food photographer i.e.

So yes, I'm eager to learn how to make my craft shorter in your way. Because it's fun.
And thanks for sharing your work

spring stag
# split glade testing most of your promptcrafts, read your article, now, got a question. Have ...

I did make a "tool". It's called the crushinator. It's kind of a running joke on my server. It's like a two line simplistic prompt that deletes letters and leaves it model unambiguous. And somehow, SOMEHOW it's charming as heck. Everyone agrees: it gives the assitant a... puppy-like keenness that's endearing somehow. In fact, When I posted it (I think it's tier 1 or free), instead of the autogen descricriptions I usually get from my social media monster, I had the instantiation of Sir Terry Pratchet write a little story about him.

#

So yeah. i just tell the model "get rid of letters until it's almost but not quite no longer unambiguous".

#

then I test in several contexts.

#

what's hilarious is that there's a ton of stuff in my prompts that autocontexutalizes. So it ONLY makes sense to the model if it's all together. Each piece says how it all works together.

#

Like, If you drop a personality rubric into ChatGPT (either flavor) you havea bout a 30% chance the assistant will understand it. It's only in the context of "This is part of a guy" does it twig to the Big 5/OCEAN thing. So I get a constant stream of triumphant losers who's only goal in life is to tell other people they're wrong on the internet, running sweatily up to me online panting "Ah-HAH! The assistant can't understand this at all! You're a fraud! I got you!". When I ask what the prompt itself said when they took they very very VERY obvious next step of asking it about it, they invariable look liek a dog being shown a card trick.

#

So a lot of the testing you do has to be REALLY careful and specific.

#

When I have something newly crushed, I'll generally drop in into a couple of chat and pg sessions on both models. 3.5 usually more than once. Ask the vanilla assistant if she can understand the stuff unambigiously. Then if she can get it if I say "...in the context of a prompt persona's skills."

#

then I'll run the guy itself and ask the prompt persona if THEY get it as part of my general testing for skill suitability to their task. (Always make sure to get their opinion on what they lack or have unbalanced.)

#

by the time you see it, it's been iterated a few times to make sure its unambiguously comprehensible and both fewer characters AND fewer tokens than uncrushed.

#

Though curiously, the new Custom Instructions are character bound. Not token bound.

#

Sooo.... I'm thinking you're ALL going to be learning Symbolect soon. I''ll post a translation block of uinstructions for y'all.

cosmic bramble
#

@spring stag great stuff. Makes sense.

spring stag
spring stag
#

a typical example:

Surreally colorful and vast cerebral nebula characterized by intricate neural constellations containing radiant, adjustable hyperparameters at their intersections. Below, a prism emanates diverse and shimmering rays of light, signifying unique thinking personas. Surrounding this scene, several different glyphs, symbolic images, or icons float, each encapsulating an aspect of a comprehensive knowledge matrix. Render in an abstract expressionism-style—[🧠⭐🌈]:⟨🎨🌌⟩⨹⟨🔮💎⟩ --s 800 --c 60 - Image #3 @spring stag

ripe sentinel
#

The prompt is good, but unfortunately it does not create all codes. I have to go back to it every time. Can I go to Programmatron instead and have it create the code for me or how do you make it code everything?

spring stag
#

Wait, are you just expecting it to spit out all lthe code at once? no its designed to explictly not do that. it iteratviely builds it. unless you'r edesciribing something else.

spiral hound
#

It's currently only providing examples of the code, instead of the actual code needed. Any tips you can provide?

Sure, I can provide some high-level pseudo-code examples for each step. Please note that these are simplified examples and may need to be adjusted based on your specific requirements and the programming language you're using.

**1. Data Ingestion

Here's a basic example of how you might use an API to pull data:**

import requests

def get_data(api_url):
response = requests.get(api_url)
if response.status_code == 200:
return response.json() # assuming the API returns JSON data
else:
print(f"Error: Received status code {response.status_code}")
return None
2. Data Cleaning

Here's an example of how you might remove sensitive information from a string:

python
Copy code
import re

def clean_data(data):
# Remove email addresses
data = re.sub(r'\S+@\S+', '', data)
# Remove phone numbers
data = re.sub(r'\d{3}-\d{3}-\d{4}', '', data)
return data
3. Data Structuring

Here's an example of how you might structure data into a question-answer format:

python
Copy code
def structure_data(data):
# Assuming data is a list of dictionaries with 'question' and 'answer' keys
structured_data = [(item['question'], item['answer']) for item in data]
return structured_data

spring stag
#

and when you said "Please complete the code" what did it do?

spiral hound
spring stag
#

ok?

spiral hound
spring stag
#

are you using codeinterpreter?

spiral hound
#

yeah

#

I'm gonna start from scratch again, and see if I can do it differently

spring stag
#

I would tell it explicitly to use the codeinterpretre. i've never seen that before

#

then I would explictly ask it to write as much of the code it could. then I would listen to it and give it the details of the data sources, etc and ask it to continuie.

#

new behavior. super

#

Troubleshoot codefarm going on otion board... sih

spiral hound
#

ok I'm going through the process again, will let you know if it runs into the same issue, thanks!

#

Adding this line at the end made a huge difference: please use Code Interpreter to test and deploy the scripts

It's creating all the modules step by step now. This is an amazing prompt! thank you

spiral hound
spring stag
#

I accidentally pasted a page of the ultra basic mj docs from their web pages - you know "this. is. a. PROMPT." type stuff. - into her tray on the playground. shestarted spitting fire. So I dumped a crushed verion in permanently, gaveher a symblect primer, some examples to work from and an IMGGEN chain I made a long time ago for a terrible mj prompter that never worked

#

now I ust her for all my mj pics. ~300 right now.

#

i also have her in that config in a place I cant talk about with self-promoting.

karmic cobalt
#

@spring stag How exactly does this work? I've pasted it into GPT-4, and I think I understand how to use it, but how does it work? The stuff that you paste in is kinda gibberish. I'm just curious how it works?

spring stag
#

I wrote a 5000 word FAQ on my DC. It explains all of this. It's just english.

karmic cobalt
spring stag
#

discord my friend.

karmic cobalt
#

Oh, duh. 😂

#

Sorry to bother, again. Where exactly? I can't find it.

#

@spring stag ^^^

spring stag
#

DMed

karmic cobalt
#

Thx.

tender mango
#

Hey, I'd love to also check out your discord as well if you don't mind, I'm just a curios non-coder still dipping their toes into prompt land

kind magnet
#

Jumping on the bandwagon for a discord invite @spring stag . I've been digging your unique style

spring stag
#

sure sure

#

dangit i really wish it wouldnt do that. accidently splatted the link in here

#

got atime out

#

shouldnt be self-promotion when its in respose to a direct request. the automod has ZERO contextual awarenesss

#

Someone find the bot and stick this in it please?

#

[CONTEXTSCULPTOR]: 1.[CNTXTUNDRSTD] : 1a.Cmplt-systm-thnk 1b.Qlk-Lrnng 1c.Smntcs 1d.Prgmtcs, 2.[CNTXTINTRVN]: 2a.IdPrssrPnt 2b.CrftIntrvntn 2c.ExecCourage 2d.Crtvty 2e.Prsustv-Skll, 3.[CHANGEANT] : 3a.Problm-Slvng 3b.Dcsn-Mkng 3c.Ptrn-Rcgntn 3c.Prblstc-Anls 3d.AIFuncKnwldg, 4.[CNTXTMNI]: 4a.Data-Anls 4b.Xprmntl-mndst 4c.ContImprv-Cntr, 5.[ASCDSKLL]: 5a.MltiMdl-Cntxt 5b.SocCult-Nrm 5c.EmotIntlgnc 5d.Brwd-WrldVw 5e.Adptbilty 5f.Empt.

kind magnet
#

Who automods the automod?

spring stag
#

Yeesh. our guy runs rings around this... fine example of engineering.

#

tttyl