jonath_laurent

7 ms·

1.
by jonath_laurent 4y ago
I agree with you. What I am taking issue with is the article presenting a multitude of (unnecessarily) complex refutations (as if one wasn't enough) and suggesting that there is no consensus on which one should be believed. This is not
2.
by jonath_laurent 4y ago
I am in fact disputing this empirical fact, or at least the interpretation of it that is suggested in the article. There is no such thing in mathematics such as an unresolved dispute over whether or not a five line proof with elementary con
3.
by jonath_laurent 4y ago
This is a good comment and I believe your justification is equally valid. The reason for the apparent disagreement is that in order to point out a logical flaw in an argument, one must make the argument formal and explicit enough first. The
4.
by jonath_laurent 4y ago
I would argue that this Wikipedia article is misleading and that it confuses more than it clarifies when it comes to resolving the paradox. In particular, I am disputing the fact that "no proposed solution is widely accepted as definit
5.
by jonath_laurent 4y ago
I do not like this explanation. In my opinion, the fancy mathematical argument involving infinite series is an unnecessary distraction from a much more fundamental and mundane mistake (i.e. incorrectly using a variable out of its scope). I
6.
by jonath_laurent 4y ago
This comment explains the fundamental reason why the reasoning is incorrect. I offer the same perspective in a different comment ( https://news.ycombinator.com/item?id=31569991 ). In my opinion, the Wikipedia article is makin
7.
by jonath_laurent 4y ago
Some excellent critiques have been provided in this thread already but I would like to offer a different perspective. In programming terms, the switching argument (see original link) is incorrect because it does not typecheck. And it does n
8.
by jonath_laurent 5y ago
As a researcher in machine learning, I wanted to explore applications of Deepmind’s AlphaZero algorithm beyond board games (such as in automated theorem proving or chemical synthesis). However, I noticed that existing open-source implementa
9.
by jonath_laurent 6y ago
I think that MuZero is a fascinating algorithm, but that a lot of news articles are misleading when they present it as a new, superior substitute for AlphaZero. MuZero is solving a harder problem, in which the learning agent does not have a
10.
by jonath_laurent 6y ago
I am wondering if your idea of using GBDTs in combination with AlphaZero might not be most influential in areas where no neural network architecture is known to provide the right inductive bias for the problem at hand. I think neural models
11.
by jonath_laurent 6y ago
Thanks! You can easily find my email address from my github. I am not aware of a canonical discord channel for discussing AlphaZero but the Lc0 community has a discord channel on which I had some interesting conversations.
12.
by jonath_laurent 6y ago
This needs clarification indeed. As I explain in the documentation, the aim of AlphaZero.jl is not to compete with hyper-specialized and hyper-optimized implementations such as LC0 or ELF OpenGO. These implementations are written in C++ wit
13.
by jonath_laurent 6y ago
This is interesting, thanks! Is there anything else you can tell me about the results of your experiments with small networks? I am really interested in this. For example: did you notice than increasing or decreasing network size required s
14.
by jonath_laurent 6y ago
And what were the results of these experiments? What error rate can you reach with the smallest network architecture you tried for example?
15.
by jonath_laurent 6y ago
Actually, I found your blog article when I was reading about AlphaZero and I found it useful!
16.
by jonath_laurent 6y ago
What do you mean by WLD output head? So far, the main idea I have pulled from the Lc0 crowd is to have a prior temperature indeed. The next thing I am planning to add is the possibility to batch inference requests across game simulations in
17.
by jonath_laurent 6y ago
Your series of blog articles has been an important source of inspiration in writing AlphaZero.jl and I cite it frequently in the documentation. Thanks to you and your team!
18.
by jonath_laurent 6y ago
Thanks for your kind message.
19.
by jonath_laurent 6y ago
Multiple GPUs support definitely belongs to the TODO list. However, I am currently limited by the state of CUDA.jl on this, as it does not have a device-aware memory pool yet. I am also looking forward to CUDA.jl supporting f16 and int8 com
20.
by jonath_laurent 6y ago
I agree with the quoted numbers. As I mentioned in another comment, you have to keep in mind that AlphaZero is an extremely sample-inefficient learning technique, even for simple problems. However, it has two major strengths: 1) it is prett
21.
by jonath_laurent 6y ago
Yes. Go 19x19 would be completely intractable on a single machine (one comment is citing a $25 million cost estimate in computing power to train AlphaGo Zero). A more reasonable target would be Go 9x9 but even this would be an extreme chall
22.
by jonath_laurent 6y ago
Such an evaluation is available in the tutorial: https://jonathan-laurent.github.io/AlphaZero.jl/dev/tutorial... Admittedly, the connect four agent is still far from perfect but there is a lot of margin for improv
23.
by jonath_laurent 6y ago
Yes, the agent is trained without access to the deterministic solution.
24.
by jonath_laurent 6y ago
I completely agree with you. Let me just add two remarks. First, although picking 9x9 boards makes connect-four intractable for bruteforce search indeed, I would be suprised if it made it much more difficult for AlphaZero, which relies on t
25.
by jonath_laurent 6y ago
I suspect FLux/Knet are still slightly slower and less memory efficient than PyTorch/TF, although things are moving very fast here! This is not relevant in understanding AlphaZero.jl speed though. The reason it is much faster than
26.
by jonath_laurent 6y ago
Author here. AlphaZero.jl supports both Flux and Knet indeed and users can choose whatever framework they want to use. As far as I understand, Flux and Knet have different strengths. I think Knet is a bit more stable and mature for large-sc
27.
by jonath_laurent 6y ago
This is very interesting! If your experiments work out, I would be interested in adding "Gradient Boosted Decision Trees" support to AlphaZero.jl. I saw the work on KataGo and implementing "Playout Cap Randomization" is
28.
by jonath_laurent 6y ago
Author here: I am happy to answer any question you may have about AlphaZero.jl. :-)