search

Tencent / PhoenixGo

Go AI program which implements the AlphaGo Zero paper
Other
2.87k stars 576 forks source link

[Solved] How to show the PV ? (winrate and continuation of moves in variation) #82

Closed wonderingabout closed 5 years ago

wonderingabout commented 5 years ago

currently, when using phoenixgo, unless i didnt find the option, we cant see the variations of the next moves (PV)

full options list here : https://github.com/wonderingabout/PhoenixGo/blob/faqv2-bazel-master/docs/mcts-main-help.md

pv1

but if using

with leela zero engine, you can see the PV, which allows to review with tools such as go review partner https://github.com/pnprog/goreviewpartner (submitted a PR here https://github.com/pnprog/goreviewpartner/pull/96 ) or simply display variations on gtp2ogs on OGS https://github.com/online-go/gtp2ogs

pv2

@wodesuck @guikarist @funionguo @fiskerhuang

wonderingabout commented 5 years ago

actually it seems there is a PV but i dont see it on my window (probably not using all the needed arguments) https://github.com/pnprog/goreviewpartner/pull/96

wonderingabout commented 5 years ago

@pnprog found how to do it

in fact it was --logtostderr --v=1 when you use it, the PV appears :

see : https://github.com/pnprog/goreviewpartner/pull/96

now the PV is here : 

Jan 16 15:55:09 !!!!! [5489] stderr: I0116 15:55:09.910559  5489 mcts_debugger.cc:43] ========== debug info for 27th move(b) begin ==========
Jan 16 15:55:09 !!!!! [5489] stderr: I0116 15:55:09.910596  5489 mcts_debugger.cc:44] main move path: mg(6482,-0.12,0.89,-0.09),lf(6349,0.12,0.88,0.08),lg(3006,-0.12,0.53,-0.12),kf(2654,0.14,0.63,-0.01),qh(941,-0.11,0.16,-0.11),qi(785,0.12,0.77,0.13),rh(784,-0.12,0.98,-0.13),pi(538,0.13,0.52,0.09),rg(445,-0.12,0.73,-0.18),og(365,0.13,0.76,0.15),pf(363,-0.13,0.98,-0.19),rd(159,0.10,0.52,0.14),ql(47,-0.06,0.21,-0.18),oq(19,0.04,0.36,-0.03),nq(12,-0.05,0.58,-0.07),pq(11,0.05,0.97,0.03),np(10,-0.05,0.79,-0.12),qn(9,0.04,0.84,0.03),ol(8,-0.05,0.82,-0.05),fc(6,0.07,0.60,0.06),kg(4,-0.06,0.42,-0.13),jf(2,-0.02,0.53,0.02),jg(1,0.05,0.45,0.05),cg(0,-nan,0.00,nan)
Jan 16 15:55:09 !!!!! [5489] stderr: I0116 15:55:09.910604  5489 mcts_debugger.cc:139] mg: N=6482, W=-757.077, Q=-0.116797, p=0.887549, v=-0.0932888
Jan 16 15:55:09 !!!!! [5489] stderr: I0116 15:55:09.910609  5489 mcts_debugger.cc:139] oh: N=262, W=-38.6935, Q=-0.147685, p=0.0710239, v=-0.137658
Jan 16 15:55:09 !!!!! [5489] stderr: I0116 15:55:09.910614  5489 mcts_debugger.cc:139] qh: N=38, W=-6.34157, Q=-0.166883, p=0.0139782, v=-0.144544
Jan 16 15:55:09 !!!!! [5489] stderr: I0116 15:55:09.910617  5489 mcts_debugger.cc:139] nb: N=1, W=-0.294998, Q=-0.294998, p=0.000126838, v=-0.295011
Jan 16 15:55:09 !!!!! [5489] stderr: I0116 15:55:09.910621  5489 mcts_debugger.cc:139] lc: N=1, W=-0.354553, Q=-0.354553, p=0.000164667, v=-0.354567
Jan 16 15:55:09 !!!!! [5489] stderr: I0116 15:55:09.910626  5489 mcts_debugger.cc:139] ob: N=1, W=-0.28775, Q=-0.28775, p=0.000373717, v=-0.287757
Jan 16 15:55:09 !!!!! [5489] stderr: I0116 15:55:09.910630  5489 mcts_debugger.cc:139] oj: N=1, W=-0.359818, Q=-0.359818, p=0.000128213, v=-0.359819
Jan 16 15:55:09 !!!!! [5489] stderr: I0116 15:55:09.910634  5489 mcts_debugger.cc:139] rc: N=1, W=-0.30574, Q=-0.30574, p=0.000222438, v=-0.305748
Jan 16 15:55:09 !!!!! [5489] stderr: I0116 15:55:09.910639  5489 mcts_debugger.cc:139] oc: N=1, W=-0.281403, Q=-0.281403, p=0.000368299, v=-0.281408
Jan 16 15:55:09 !!!!! [5489] stderr: I0116 15:55:09.910642  5489 mcts_debugger.cc:139] qd: N=1, W=-0.288712, Q=-0.288712, p=0.00012808, v=-0.288724
Jan 16 15:55:09 !!!!! [5489] stderr: I0116 15:55:09.910646  5489 mcts_debugger.cc:48] model global step: 639200
Jan 16 15:55:09 !!!!! [5489] stderr: I0116 15:55:09.910648  5489 mcts_debugger.cc:49] ========== debug info for 27th move(b) end   ==========

pv phoenixgo

wonderingabout commented 5 years ago

Solved !