Jonathan Laurent

I personally have trained small games with up to 40 iterations but number of iterations is not a very interesting metric on its own. There is a tradeoff between doing...

I personally do not have experience with Kaggle notebooks so I cannot help here. @Snimm Do you want to leave the issue open in case someone else can help?

Such an error is a bug in CUDA.jl or Julia itself. Internal errors like those aren't supposed to happen, whatever sins AlphaZero.jl may be committing. I would encourage you to...

The more relevant number to measure game-complexity when it comes to AlphaZero is the size of the action space and not the total number of positions. Also, strictly speaking, Chess...

@smart-fr This is a very good example of using game-specific knowledge to make AlphaZero tractable when a naive application wouldn't be. Thanks!

Is this still an issue? I cannot replicate on my own machine. If this is still a problem, can you please give me all standard details (Julia and package versions,...

I did not manage to replicate this on my own MacOS machine. Are you running the `#master` version of AlphaZero.jl on a fresh environment? One thing that may be causing...

@LaurentMazare This is interesting, thanks. To be clear though, do you think Async has something to do with the problems you observed or do you think they arise more generally?

I concur with this thread. A solution with `getname` that would not require to add a `subtypekey` field to every subtype seems preferable to me.

I like the solution above but I am wondering: are there some standard utilities one can use to avoid writing all the associated boilerplate by hand?