TalkChess.com

I intend to reuse this thread in the future for all Winter releases, as other authors have done similarly. This first post is the Winter 0.9 release post. Before I get into what has changed since version 0.9, I think it is interesting to note what has not changed. Winter is still using the exact sam...

Ethereal 12.50 vs Stockfish 12 CCRL FRC testing has this chunk of games right at the very end of the 300 games they played : [redacted for brevity]1 1 1 [redacted for brevity] Overall results ..... [redacted for brevity] Wow! I knew Ethereal progressed a lot, but a perfect score against SF12 is ama...

It would not be to difficult to add a lot of ideas from Stockfish and add NNUE evaluation to pimp it up 100 or 200 Elo, but I simply don't want to do this because it would give me the feeling that it is not my own engine anymore. There is a word for this. I believe it is called "integrity." This is...

Awesome! I haven't actually gotten around to properly reading the paper, so I what a rough deadline for submitting questions could be?

A tester does not need 5000 games, unless the goal is only to maximize the Elo precision. A useful tool for engine developers. What are you trying to answer. 1. What is the exact elo deference between A vs B. 2. Who is better only between A vs B, and this usually takes much fewer games to prove sta...

Nice data! It is interesting to me that Winter seems inversed. In your second post Winter goes 31->24->22->19, progressively improving ranking the later the game goes.

I am not completely sure how to interpret the data however, perhaps Winter is just slow at closing out games?

I think given the information you have provided we cannot say for certain what is going on. It should be noted however, that there might not be anything wrong at all. I you have a fail-hard framework, then moves will not return a static score below your lower bound. This means, if you search the bes...

1) If I want the NN to return an evaluation, which could be used in a traditional alphabeta search engine, how is this achieved most commonly? Do I have a last hidden layer that feeds into a single output neuron which uses e.g. a tanh activation function so -1 would be a loss and +1 would be a win?...

Oh wow! Thank you! I honestly only followed the first half of the tournament and mostly followed the gauntlet after that. After a very bad start I was just happy Winter was no longer last, but I guess with how close the field was Winter was able to make a fairly extreme comeback. Thanks for organizi...

Thank you! I have an i7-9700k with 2080ti. Which version should I use? popcount, bmi2, or avx2? The GPU is irrelevant for SF. Use whichever on happens to run fastest on your system, it's not like you pay money per binary. Just run them each from a couple positions. Most likely BMI2 version will be ...