Skip to main content
22 votes
Accepted

How do AI developers for LLM fix bugs when the LLM misbehaves, given that they cannot control what the LLM learns from the data?

machine learning software behaviour is unpredictable since the developer cannot control what the software learns from the data. If you control the model, the data, the input, and the random, then the ...
candied_orange's user avatar
17 votes

How do AI developers for LLM fix bugs when the LLM misbehaves, given that they cannot control what the LLM learns from the data?

The answers so far have not mentioned reinforcement learning from human feedback (RLHF). I'm not an expert, but my understanding is that at least in the case of ChatGPT, this is the main way in which ...
N. Virgo's user avatar
  • 894
15 votes

What programming paradigm do you think would work best for the AI aspect of a chess engine?

I wish to write a chess AI which simulates the way I think over the board, using C++. Awesome! My focus is on writing the algorithms for choosing moves (decision making), not defining the board and ...
candied_orange's user avatar
11 votes

Machine learning without programming?

First of all, it seems to me you are missing (or perhaps misunderstanding) the "explicitly" bit in "without being explicitly programmed" (from the quote in the question). It doesn't mean that no ...
Iker's user avatar
  • 865
10 votes

Is it possible to predict last 3 digits of a closing stock price?

AI isn't magic. Machine learning algorithms work by detecting patterns in a particular set of data, then using those patterns to classify new samples, predict future members of the pattern, etc. If ...
IMSoP's user avatar
  • 5,877
9 votes

Is it OK to commit code written by a large model such as ChatGPT or Github Copilot?

At this time, I couldn't recommend that anyone use AI code generation tools in a professional context. Although it's the domain of lawyers, there are still open questions about copyright, licensing ...
Thomas Owens's user avatar
  • 83.3k
8 votes
Accepted

Distinction between AI, ML, Neural Networks, Deep learning and Data mining

You are correct about AI which includes ML which includes DL. NN can indeed be included in ML, be it inside or outside of the DL context. An example for the latter is when neuronal nets are used in ...
Christophe's user avatar
  • 78.4k
6 votes
Accepted

What is identity mapping in neural networks

I'm not an neural network expert but I understand that identity mapping ensures that the output of some multilayer neural net is ensured to be equal to its input. Such a net is also called a ...
Christophe's user avatar
  • 78.4k
6 votes
Accepted

What do I not understand about Alpha-Beta-Pruning in Chess?

Min-Max search (with or without ABP) is not supposed to "build the complete search tree" in memory. It is usually implemented as Depth First Search tree traversal, which traverses the complete search ...
Doc Brown's user avatar
  • 209k
5 votes

What programming paradigm do you think would work best for the AI aspect of a chess engine?

To my knowledge, most chess programs written to date are focused on taking advantage of the computer's calculating powers (aka brute force method). My program will be different in that the focus is ...
Paul92's user avatar
  • 2,591
5 votes

What is identity mapping in neural networks

It's the same as it is in Algebra. An identity map or identity function gives out exactly what it got. When they say: h(xl) = xl They mean h is an identity mapping / function. If you give it xl it ...
candied_orange's user avatar
5 votes
Accepted

Improved design for a turn-based multiplayer game using AI?

I would try to keep the TicTacToeGame completely UI agnostic. No observer, no publisher-subscriber inside that class. Only "business logic" (or call it "game-logic") inside that class, no mixed ...
Doc Brown's user avatar
  • 209k
5 votes

API for a statefull AI server

Here's a rough answer. "Good Design" for web apis is based around the assumption that you are running a web site. Web sites generally want to serve lots of users at the same time To serve ...
Ewan's user avatar
  • 77k
5 votes

How do AI developers for LLM fix bugs when the LLM misbehaves, given that they cannot control what the LLM learns from the data?

The natural way would be tuning the model, which is likely a long-term solution as it can take quite a bit to reparse input data (this time cleaned from a few unwanted sources). This is what ...
Frank Hopkins's user avatar
4 votes
Accepted

Getting started with machine learning a zero sum game?

This is a good candidate for Differential Evolution. DE is a very simple (but powerful) population based, stochastic function minimizer/maximizer. A key point for integrating DE in your scheme is ...
manlio's user avatar
  • 4,216
4 votes

How to design an AI For Sorry?

You have good start. Each AI needs to be able to figure out "value" of current board. But it is often not enough. This is because calculating good "value" heuristic might be complicated and often ...
Euphoric's user avatar
  • 37.5k
4 votes

Is manually tagging content an ideal start for eventual machine learning?

You're building a training set. This is used to teach the AI what you want. The important thing is to be careful that the set doesn't contain false tells like a red and white checkered table cloth ...
candied_orange's user avatar
4 votes

Best-first search on a graph

Node e has a cost value of 12 (actual cost 2 + 4, + heuristic value 6). As there are no other nodes from node c, the next node would be node e This depends a bit on what type of 'best first' search ...
JonasH's user avatar
  • 5,669
4 votes

Identifying the minimal set of columns that can be used to define a unique instance in a database table

Avoid over-engineering your application with the intent of generating primary keys for row-by-row table comparison. Here's why: Immutability Verification: Confirming the immutability of columns is ...
Newbie Guest's user avatar
3 votes

(AI) algorithm to optimize input parameters

I would recommend to try different standard algorithms for optimization of non-differentiable functions and see how well it works. Out of my head, in increasing complexity: hill climbing threshold ...
Doc Brown's user avatar
  • 209k
3 votes

Choosing a sequence of animations for an NPC - use a behaviour tree?

Generally you need to split your cats logic from its animations. Now first you need to write cats logic. One good approach I have found is to split logic into layers. Needs Cat can have some state ...
Kromster's user avatar
  • 511
3 votes
Accepted

Neural networks: No solutions for test data possible?

The first approach is called supervised learning, and the second is called reinforcement learning. There are two ways you can use a neural network with reinforcement learning for chess: as a policy ...
Don Reba's user avatar
  • 228
3 votes

Machine learning without programming?

You use programming to build a machine that can learn without you having to do any more programming. You use programming to build the machines mind. The mind goes on to learn things that were not ...
candied_orange's user avatar
3 votes
Accepted

Fitness Function altenatives in Genetic Algorithms for game AI

Disclaimer: I've got no practical experience with genetic algorithms but a bit of an interest... So your strategy is to pair individuals and have them compete against each other. Eliminate the losers ...
Kempeth's user avatar
  • 220
3 votes

Improved design for a turn-based multiplayer game using AI?

I would go with the strategy pattern. class Player { async getNextMove() { throw new Error('not implemented'); }; } class AiPlayer extends Player { async getNextMove() { /...
Peter's user avatar
  • 139
3 votes
Accepted

How to guess which pieces of text belong to which lines in a scanned document?

The ideas of the two commenters is by and large your best bet, because it relies on some kind of normalization, which is the typical way such problems are treated. Because of the "heuristic" nature of ...
Vector Zita's user avatar
  • 2,462
2 votes
Accepted

How to analyse user input and determine if it (kind of) matches an answer

It may be that the user misses out a few words, writes in short hand, supplies a partial answer, or something along those lines. This leads to a natural language processor. Since it looks like you ...
Arseni Mourzenko's user avatar
2 votes
Accepted

Building a simple AI for Noughts & Crosses game

My naïve approach would be something like a priority order. You can detect winning moves by pretty simple pattern matching, ie check each row and column and diagonal for two of the same and an empty ...
HamHamJ's user avatar
  • 497
2 votes

Building a simple AI for Noughts & Crosses game

Is there a smarter way of creating a noughts & crosses AI than simply outlining all the possible winning moves? Well, the 'smarter' the AI gets, the requisite information becomes the outlining of ...
Mars01's user avatar
  • 21
2 votes
Accepted

How to run 2-layer perceptron to solve XOR

You seem to be attempting to train your second layer's single perceptron to produce an XOR of its inputs. This isn't possible; a single perceptron can only learn to classify inputs that are linearly ...
Periata Breatta's user avatar

Only top scored, non community-wiki answers of a minimum length are eligible