Most Popular
1500 questions
6
votes
1 answer
Why did the openai's gym website close?
Openai's gym website redirects to the GitHub repository. Why did the openai's gym website close?
Franck Dernoncourt
- 3,004
- 1
- 19
- 34
6
votes
1 answer
Why are biases (typically) not used in attention mechanism?
Watching this video implementing attention in a transformer. He set query, key, and value biases to False and said "Typically, people don't use biases for these".
Even in official PyTorch code the default bias is False:
add_bias_kv: If specified,…
Peyman
- 564
- 1
- 3
- 11
6
votes
2 answers
How does GPT-based language model like ChatGPT determine the n-th letter of a word?
I understand that GPT models process input text by converting words into tokens and then embedding vectors and do not process them letter by letter. Given this approach, I am curious to know how a model like ChatGPT can identify the first (or n-th)…
Peyman
- 564
- 1
- 3
- 11
6
votes
1 answer
How should the racing agent take into account the velocity of the vehicle, given the images with a speedometer?
I'm developing a game AI, which tries to master racing simulations. I already trained a CNN (AlexNet) on in-game footage of me playing the game and the pressed keys as the target. As the CNN is only making predictions on a frame-to-frame basis, and…
TheJD
- 103
- 5
6
votes
1 answer
AI efficiency KPI
There should be some Key Performance Indicators designed for measuring AI performance. For example, the number of entities examples you have to feed it in order to obtain single task on a testing entity with repeatable 97% accuracy.
Is there any of…
kakaz
- 271
- 1
- 6
6
votes
5 answers
How is GPT 4 able to solve math?
How can GPT 4 solve complex calculus and other math problems. I believe these problems require analytical reasoning and ability to compute numbers. Does it still use a LLM to complete this process or does it add on to this?
Here is the link to the…
desert_ranger
- 650
- 1
- 5
- 20
6
votes
1 answer
Will a CNN that is Group Equivariant always be better than a regular CNN?
I am reading this paper about Group Equivariant Convolutional Networks. Basically, it is a CNN whose construction makes the network naturally equivariant to Group transformations (e.g. rotations) of the input.
This is, a GE-CNN trained with the…
puradrogasincortar
- 199
- 6
6
votes
2 answers
Can I use AI to interpret XML documents?
I think about a system which gets XML documents in various structures but with essentially the same data structure in it. For the example, let's assume each document contains data about one or more persons. So the AI would recognize a name.…
DBX12
- 171
- 2
- 7
6
votes
2 answers
Would self-hosting ChatGPT be feasible, w.r.t. computation costs?
Suppose the pre-trained, current date (2023-02-04) ChatGPT model was released open source, would it be feasible for regular users to interact with the model on a self-hosted computer?
Assumptions
I assume getting output based on some input is, at…
a.t.
- 243
- 1
- 7
6
votes
1 answer
Why do knowledge-based agents only add a sentence to the knowledge base when it is 100% sure the sentence is true?
According to Russell and Norvig, a knowledge-based agent will only add a sentence to its knowledge base if it follows logically from what it previously knows, or directly observes. To follow logically essentially means that if the premises are true,…
Peter
- 71
- 1
6
votes
4 answers
Is the play of strong Chess AI easily distinguishable from human play?
I don't play nearly enough Chess to be able to answer. For context, AlphaGo is stronger than the current strongest human player, but AlphaGo's game play has been cast as "inhuman" in the sense that it doesn't resemble human play. (In Go, this can…
DukeZhou
- 6,227
- 5
- 25
- 53
6
votes
3 answers
How is ChatGPT able to repeat random numbers?
From what I understand, ChatGPT is just a fancy neural network, operating like a sophisticated Markov Chain generator. As such, it should only be able to generate tokens that are in its training dataset. One thing it should not be able to generate…
yters
- 387
- 3
- 11
6
votes
1 answer
Why does ChatGPT create fake code?
ChatGPT has been a big thing lately. It also makes a lot of mistakes. For example, it creates fake functions of a package and tells it as it works for real. I was wondering how that works. Why is it creating fake functions of code and not just…
Quinten
- 185
- 7
6
votes
1 answer
How was ChatGPT trained?
I know that large language models like GPT-3 are trained simply to continue pieces of text that have been scraped from the web. But how was ChatGPT trained, which, while also having a good understanding of language, is not directly a language model,…
HelloGoodbye
- 313
- 1
- 11
6
votes
1 answer
How much of the ChatGPT output is copied from its training set (vs. being abstractively generated)?
One of the main concerns of using ChatGPT answers on Stack Exchange is that it may copy verbatim or almost verbatim some text from its training set, which may infringe the source text's license. This makes me wonder how much of the ChatGPT output is…
Franck Dernoncourt
- 3,004
- 1
- 19
- 34