This system can be used for any role capable LLM. The system is defaulted to run in 3.5 and 4 with the default pre-prompt and code. Please read everything and feel free to share results. class code is MIT.
#Enhanced Unlimited Context Session System
winged nymph
winged nymph
Enhanced Unlimited Context Session System
quasi lotus
winged nymph
in the web interface it cant, in API it can.
Enjoy though. It should work with many LLMs and can be hard coded in that way, though it needs to be optimized for your own implementation
winged nymph
use the class code and customize it; you actually have to work it in and fit the code to your own system. i cant do that for you.
winged nymph
If you understand it the class system can be used elsewhere. Inference is why I provided it.
lime otter
I just don't get why it would be structured in such a way that requires explicit invocation. Wouldn't it make more sense not to require the user to follow those conventions and "just work" if that were possible?
winged nymph
Its acting on code simulation and inferring LLM instructions in the same word @lime otter . It has to be a capital command so the word continue is still usable in conversation without triggering the context pull, at the same time the capital 'CONTINUE' is triggering both the code and the nature implied continue. without forcefully calling the memory it will drop the context of the code simulation. I went through all of that in development. its a whole new construct which derived out much data.
Prob the most useful tid bit as of recent. this was extremely useful information. I'm sure you'll see that.: "The symbols "[]", "{}", and "()" are used in different ways depending on the context, particularly in programming, mathematics, or text annotation. Here's how GPT-3 (the underlying architecture for ChatGPT) interprets these symbols:
Programming Context:
[] (Square Brackets): Typically used for array indexing, list comprehension, or to denote arrays in some languages.
{} (Curly Brackets): Commonly used to denote blocks of code, such as function or loop bodies in many programming languages like C, Java, and JavaScript.
() (Parentheses): Used for grouping expressions, denoting function calls, or specifying tuple literals.
In programming, the interpretation of these symbols follows the conventions of the relevant programming language. The code's logic and structure are influenced by these symbols, and ChatGPT evaluates them accordingly.
Mathematical Context:
[] (Square Brackets): Often used to denote closed intervals in mathematics.
{} (Curly Brackets): Typically used to denote sets.
() (Parentheses): Used for grouping expressions or denoting open intervals.
In a mathematical context, these symbols help in organizing expressions and denoting specific mathematical constructs.
Text Annotation or Roleplay:
[] (Square Brackets): Sometimes used to provide additional information or instructions.
{} (Curly Brackets): Might be used for placeholders or to denote variables.
() (Parentheses): Often used to provide extra information, clarify, or add a side note.
In text annotation or roleplay, these symbols help provide additional context or information."
sicne you asked, this all comes up in your homework if you want to absorb it.
all of those can be embedded within eachother and every other.
whole new complexities
and you're welcome!
winged nymph
Hwoever wrote that would benefit from my little conversation piece we just had above; i wrote this little tool. Natural language coder, its MIT and is a small model that takes a prompt and restructures the human language based on a concept i came up with i cal "Natural Pythonic/code Language". What it does is sort the thoughts into chunks that are the equivalent of a function in language and parses that into a more rudimentary system of language with grouping that is Python coding formatted partially to shrink the prompt and also restructure the message into a computational format that the machine seems to enjoy. I left off with a few more structural elemetns, but its already very nice. simply write up a prompt, then feedit to natural language coder and it coughs up a new block of language that looks wild but will compute nice. Its important to copy every single thing from the output including all spaces even at the end to then paste. Definitely a lot of time went into the tool construct.Requires NLTK and Tkinter i think to build or just run in an ide to use. be nice if somebody touched it up for me looks around. Mysticmarks NaturalLanguagecoder on github for anyone that takes that wants to play with prompt reduction through code mechanisms.
winged nymph
my coder is an actual application, not a prompt. The prompt above is again my own work, which is MIT class code and free to use.
winged nymph
Prob the most useful tid bit as of recent. this was extremely useful information...
and my note here is sharing knowledge that completely changes everything about prompt structures.
make of it what you will and enjoy it all.