HCha/Stockfish
1
0
Fork 0
mirror of https://github.com/HChaZZY/Stockfish.git synced 2025-12-24 02:57:11 +08:00
Code Issues Packages Projects Releases Wiki Activity

Add docs for auto_lr_drop

Browse Source
This commit is contained in:
Tomasz Sobczyk
2020-10-12 11:48:57 +02:00
committed by nodchip
parent 4a2bf16b30
commit 4a340ad3b2
1 changed files with 2 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/learn.md
View File

@@ -64,6 +64,8 @@ Currently the following options are available:
`newbob_num_trials` - determines after how many subsequent rejected nets the training process will be terminated. Default: 4.
`auto_lr_drop` - every time this many positions are processed the learning rate is multiplied by `newbob_decay`. In other words this value specifies for how many positions a single learning rate stage lasts. If 0 then doesn't have any effect. Default: 0.
`nn_options` - if you're reading this you don't use it. It passes messages directly to the network evaluation. I don't know what it can do either.
`eval_save_interval` - every `eval_save_interval` positions the network will be saved and either accepted or rejected (in which case an LR drop follows). Default: 100000000 (100M). (generally people use values in 10M-100M range)