This was updated a while ago, but I didn’t really update the blog due to being just… lazy. \ud83d\ude42 Yes, cbaduk.net plays handicapped games, and it was playing handicapped games for the last three months! There was a lot of changes internally regarding what I was doing – sometimes it got worse, sometimes it was doing blatantly stupid moves, and now it seems that I have something.<\/p>\n\n\n\n
The basic idea is to have a third head, as I wrote previously on https:\/\/github.com\/leela-zero\/leela-zero\/issues\/2331<\/a><\/p>\n\n\n\n That is, winrate is calculated by the multiple of expected score and uncertainty – that is, the engine will prefer playing a chaotic game rather than giving the opponent clear territory.<\/p>\n\n\n\n The idea seemed to be quite good, except it seems to need to gobble up a terribly large amount of compute. Leela zero already spend more than millions of dollars worth of compute resources to get to the current situation, and all I have is three NVIDIA GTX1080s. Spending compute resources for doing all that was just going nowhere. Progress was slow, and even after months it was playing somewhat sensibly and then… playing blatantly stupid moves. It seemed that I can’t really afford to do all that training.<\/p>\n\n\n\n Instead, I tried a different approach… what if, I just collected Leela Zero self-play data and recreate the endstate plane? To elaborate,<\/p>\n\n\n\n This strategy resulted in processing 100K games in roughly 40 hours. Yes it isn’t perfect since the original plays were for a komi of 7.5, but it probably is much better than any amount of compute I can afford.<\/p>\n\n\n\n Give it a try! I played a couple of six-stone handicap games (obviously I was playing black) and it seems to make aggressive attacks trying to capture black’s stones. Most of the games I ended up getting at least one dragon being slaughtered and then lose by a dozen points. \ud83d\ude42<\/p>\n","protected":false},"excerpt":{"rendered":" This was updated a while ago, but I didn’t really update the blog due to being just… lazy. \ud83d\ude42 Yes, cbaduk.net plays handicapped games, and it was playing handicapped games for the last three months! There was a lot of changes internally regarding what I was doing – sometimes it got worse, sometimes it was … Continue reading “Handicap Games”<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"_links":{"self":[{"href":"https:\/\/cbaduk.net\/w\/index.php?rest_route=\/wp\/v2\/posts\/144"}],"collection":[{"href":"https:\/\/cbaduk.net\/w\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/cbaduk.net\/w\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/cbaduk.net\/w\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/cbaduk.net\/w\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=144"}],"version-history":[{"count":1,"href":"https:\/\/cbaduk.net\/w\/index.php?rest_route=\/wp\/v2\/posts\/144\/revisions"}],"predecessor-version":[{"id":145,"href":"https:\/\/cbaduk.net\/w\/index.php?rest_route=\/wp\/v2\/posts\/144\/revisions\/145"}],"wp:attachment":[{"href":"https:\/\/cbaduk.net\/w\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=144"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/cbaduk.net\/w\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=144"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/cbaduk.net\/w\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=144"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}} # endstate head\n conv_st = self.conv_block(flow, filter_size=1,\n input_channels=self.RESIDUAL_FILTERS,\n output_channels=2,\n name=\"endstate_head\")\n h_conv_st_flat = tf.reshape(conv_st, [-1, 2 * 19 * 19])\n W_fc4 = weight_variable(\"w_fc_4\", [2 * 19 * 19, (19 * 19) * 2])\n b_fc4 = bias_variable(\"b_fc_4\", [(19 * 19) * 2])\n self.add_weights(W_fc4)\n self.add_weights(b_fc4)\n h_fc4 = tf.add(tf.matmul(h_conv_st_flat, W_fc4), b_fc4)\n<\/code><\/pre>\n\n\n\n endstate_winrate = tanh( avg_delta * confidence \/ 10.0 )\n delta = sum (number_of_my_stone - number_of_opponent_stone + komi_bias)\n confidence = average ( (v - 0.5) * ( v - 0.5) for v in endstate_plane )\n<\/code><\/pre>\n\n\n\nWell…<\/h3>\n\n\n\n
So, how does it play?<\/h3>\n\n\n\n