ScaleQ - Updated implementation of LogitQ idea (WIP - not for merge) #1408

AlexisOlson · 2020-08-03T20:23:25Z

The ScaleQ parameter is a tunable parameter that goes from 0 (current behavior) to 1 (logit of win probability).

Catch up to master

* Appended files. * Compiles. * Compiles again. * Make smart pruning use smoothed nps. * Seems to be fully implemented. * Mistype. * One more bug. * Found discrepancy with documentaiton. * Bugfixes. * Don't smooth nps during the first move. * Too large default for timeuse decay. * Bugfix. * Fix build. * Relax defaults a bit. Add fixed to logging. * Remove "smooth" to "smooth-experimental" for now.

* Add M effect logic to output section * Fix missing prefixes and semicolons * Some fixes. * Slight format improvement? Co-authored-by: Tilps <[email protected]>

* Added TempDecayStartMove for starting temp decay only after a given number of moves. This allows keeping initial game up for a few moves and still use decay. * Doesn't allow temperature to fall below endgame temp during temp decay. Still allows initial temp to be below endgame temp. * Doesn't allow temperature to fall below endgame temp during temp decay. Still allows initial temp to be below endgame temp. * Hide temp options * renamed TempDecayStartMove to TempDecayDelayMoves Co-authored-by: Alexis Olson <[email protected]>

* Changelog for 0.25.0-rc2. * Add one more PR to the changelog.

* custom winograd convolution for cuda backends * custom winograd fixes - fix a bug to make it work for non-SE networks - enable by default only with fp32. * address review comments * remove random line in comment * remove unused constants - W,H are hardcoded to 8 - because there are assumptions in the code based on that. No point in defining constants.

* cuda winograd fixes - don't typecast directly to half datatype in CPU side code as older CUDA runtime doesn't support that. - don't use gemmEx version on GPUs older than Maxwell generation (not supported). - modify the check to enable custom_winograd setting. It should be faster in most cases - except presently on RTX GPUs when using fp16.

Default to white to move, no castling, no en passant, 0 rule50ply, 1 total move. Also convert other string to std::string and removing using.

…nce_movestart_ms == 0 (LeelaChessZero#1243)

* memory optimization for cudnn custom_winograd - don't save untransformed weights - print warning message when low memory is detected. * address review comments * fix warning message * fix total weight size calculation 2 layers per residual block!

Use printing lambdas for parts of the verbose output to share between the newly outputted node and its children.

…ero#1284) * Allow smart pruning to terminate search if win is known. * Minor tweak, better safe than sorry.

…ro#1286) * Fix bug where pv might not update. * Fix...

* Time management refactoring (LeelaChessZero#1195) * Appended files. * Compiles. * Compiles again. * Make smart pruning use smoothed nps. * Seems to be fully implemented. * Mistype. * One more bug. * Found discrepancy with documentaiton. * Bugfixes. * Don't smooth nps during the first move. * Too large default for timeuse decay. * Bugfix. * Fix build. * Relax defaults a bit. Add fixed to logging. * Remove "smooth" to "smooth-experimental" for now. * MLH verbose stats - Issue 1200 (LeelaChessZero#1230) * Add M effect logic to output section * Fix missing prefixes and semicolons * Some fixes. * Slight format improvement? Co-authored-by: Tilps <[email protected]> * Start TempDecay only after a given number of moves (LeelaChessZero#1212) * Added TempDecayStartMove for starting temp decay only after a given number of moves. This allows keeping initial game up for a few moves and still use decay. * Doesn't allow temperature to fall below endgame temp during temp decay. Still allows initial temp to be below endgame temp. * Doesn't allow temperature to fall below endgame temp during temp decay. Still allows initial temp to be below endgame temp. * Hide temp options * renamed TempDecayStartMove to TempDecayDelayMoves Co-authored-by: Alexis Olson <[email protected]> * Changelog for 0.25.0-rc2. (LeelaChessZero#1233) * Changelog for 0.25.0-rc2. * Add one more PR to the changelog. * Cuda winograd (LeelaChessZero#1228) * custom winograd convolution for cuda backends * custom winograd fixes - fix a bug to make it work for non-SE networks - enable by default only with fp32. * address review comments * remove random line in comment * remove unused constants - W,H are hardcoded to 8 - because there are assumptions in the code based on that. No point in defining constants. * cuda winograd fixes (LeelaChessZero#1238) * cuda winograd fixes - don't typecast directly to half datatype in CPU side code as older CUDA runtime doesn't support that. - don't use gemmEx version on GPUs older than Maxwell generation (not supported). - modify the check to enable custom_winograd setting. It should be faster in most cases - except presently on RTX GPUs when using fp16. * Allow most parts of fen to be optional. (LeelaChessZero#1234) Default to white to move, no castling, no en passant, 0 rule50ply, 1 total move. Also convert other string to std::string and removing using. * Fix UpdateNps to actually smooth the nps and correctly handle time_since_movestart_ms == 0 (LeelaChessZero#1243) * Update changelog for 0.25.0 final release. (LeelaChessZero#1244) * Always report at least 1 depth. (LeelaChessZero#1247) * Fix un-intended regression for GTX GPUs (LeelaChessZero#1246) * memory optimization for cudnn custom_winograd (LeelaChessZero#1250) * memory optimization for cudnn custom_winograd - don't save untransformed weights - print warning message when low memory is detected. * address review comments * fix warning message * fix total weight size calculation 2 layers per residual block! * keep pdb files only for release builds (LeelaChessZero#1256) * doc update (LeelaChessZero#1267) * Include verbose stats for the node. (LeelaChessZero#1268) Use printing lambdas for parts of the verbose output to share between the newly outputted node and its children. * add alphazero time manager (LeelaChessZero#1201) * Updated FLAGS.md with logfile flag (LeelaChessZero#1275) * Fixed a typo in CONTRIBUTING.md (LeelaChessZero#1274) * Update Readme about using git (LeelaChessZero#1265) * Make `wl_` double. (LeelaChessZero#1280) * Move move filter population to a constructor. (LeelaChessZero#1281) * Filter out illegal searchmoves to avoid crashing. (LeelaChessZero#1282) * Clear policy for terminal loss. (LeelaChessZero#1285) * Allow smart pruning to terminate search if win is known. (LeelaChessZero#1284) * Allow smart pruning to terminate search if win is known. * Minor tweak, better safe than sorry. * Fix bug where pv might not update for best move change. (LeelaChessZero#1286) * Fix bug where pv might not update. * Fix... Co-authored-by: Alexander Lyashuk <[email protected]> Co-authored-by: Tilps <[email protected]> Co-authored-by: Naphthalin <[email protected]> Co-authored-by: Ankan Banerjee <[email protected]> Co-authored-by: Ed Lee <[email protected]> Co-authored-by: borg323 <[email protected]> Co-authored-by: Hace <[email protected]> Co-authored-by: Kip Hamiltons <[email protected]> Co-authored-by: nguyenpham <[email protected]>

Catch up to master

Catch up to LCZ master

leedavid · 2020-08-10T09:41:03Z

how is about to lower the Q value if being inchecked or checking, so can simulate checking extension?

AlexisOlson · 2020-08-10T17:02:22Z

Preliminary Tuning Results:

Constant: CpuctFactor = 0.0

Constant: ScaleQ = 20.0, CPuctFactor = 0.0

AlexisOlson · 2020-08-20T17:14:48Z

After trying a bunch of different tuned parameters, I always lose strength.

It could be that my tunes were bad but I'm archiving this since no progress is apparent.

AlexisOlson · 2020-09-08T14:12:49Z

I'm going to try tuning again with less exploitative acquisition functions.

Naphthalin · 2022-11-02T14:22:22Z

As logitQ was removed again, this likely can be closed as well.

Naphthalin · 2023-06-06T11:47:58Z

Closing this as incompatible with the current implementation.

Furthermore, after having worked on #1791, I'm convinced that the idea of logitQ based on the expected score under training conditions is in fact flawed, and the actual value we were looking for back when we tested logitQ is actually the WDL_mu quantity in #1791.

AlexisOlson and others added 30 commits April 18, 2020 13:50

Merge pull request #1 from LeelaChessZero/master

58d97b0

Catch up to master

Merge pull request #3 from LeelaChessZero/master

3667427

Catch up to master

MLH verbose stats - Issue 1200 (LeelaChessZero#1230)

6ff9bd0

* Add M effect logic to output section * Fix missing prefixes and semicolons * Some fixes. * Slight format improvement? Co-authored-by: Tilps <[email protected]>

Changelog for 0.25.0-rc2. (LeelaChessZero#1233)

5ce0f35

* Changelog for 0.25.0-rc2. * Add one more PR to the changelog.

Allow most parts of fen to be optional. (LeelaChessZero#1234)

1f81b79

Default to white to move, no castling, no en passant, 0 rule50ply, 1 total move. Also convert other string to std::string and removing using.

Fix UpdateNps to actually smooth the nps and correctly handle time_si…

312d0f5

…nce_movestart_ms == 0 (LeelaChessZero#1243)

Update changelog for 0.25.0 final release. (LeelaChessZero#1244)

af5be39

Always report at least 1 depth. (LeelaChessZero#1247)

558a378

Fix un-intended regression for GTX GPUs (LeelaChessZero#1246)

2942289

keep pdb files only for release builds (LeelaChessZero#1256)

3634dad

doc update (LeelaChessZero#1267)

64884f5

Include verbose stats for the node. (LeelaChessZero#1268)

ac6d920

Use printing lambdas for parts of the verbose output to share between the newly outputted node and its children.

add alphazero time manager (LeelaChessZero#1201)

d98eceb

Updated FLAGS.md with logfile flag (LeelaChessZero#1275)

0123e8e

Fixed a typo in CONTRIBUTING.md (LeelaChessZero#1274)

eac826c

Update Readme about using git (LeelaChessZero#1265)

5304394

Make wl_ double. (LeelaChessZero#1280)

9a1ca2a

Move move filter population to a constructor. (LeelaChessZero#1281)

d5c9328

Filter out illegal searchmoves to avoid crashing. (LeelaChessZero#1282)

d18ba0f

Clear policy for terminal loss. (LeelaChessZero#1285)

9563aba

Allow smart pruning to terminate search if win is known. (LeelaChessZ…

8573f15

…ero#1284) * Allow smart pruning to terminate search if win is known. * Minor tweak, better safe than sorry.

Fix bug where pv might not update for best move change. (LeelaChessZe…

d7c5d5e

…ro#1286) * Fix bug where pv might not update. * Fix...

Merge pull request #7 from LeelaChessZero/master

d086d97

Catch up to master

Merge pull request #9 from LeelaChessZero/master

f5783d0

Catch up to LCZ master

AlexisOlson added 14 commits August 3, 2020 10:30

Add FastAtanh

ef47d65

Define ScaleQ

84c8662

Apply ScaleQ to GetQ, GetWL, GetD

7c96e4b

Add ScaleQ parameter

8336399

Add ScaleQ

2b773a0

Remove GetScaleQ bool parameter

2bc17ce

Update GetQ

c86cc1c

Revert GetWL and GetD changes

6905cb3

Use unscaled Q in m_evaluator

2620382

Remove .GetID()

918084c

Include GetScaleQ (?)

555a596

Add "params_."

ed7efa3

Revert 555a596

5db8522

Apply ScaleQ inside GetFpu

2355425

AlexisOlson marked this pull request as ready for review August 5, 2020 15:11

AlexisOlson added 3 commits August 5, 2020 13:52

Tweak parameterization for tuning purposes.

f7010ec

Missing colon

1eacdce

Include cmath for std::pow

0ffd1ec

AlexisOlson closed this Aug 20, 2020

AlexisOlson reopened this Sep 8, 2020

Naphthalin added not for merge Experimental code which is not intended to be merged into the master stale Outdated PR, might be closed due to merge conflicts or inactivity labels Nov 2, 2022

Naphthalin added the enhancement New feature or request label Nov 3, 2022

Naphthalin closed this Jun 6, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ScaleQ - Updated implementation of LogitQ idea (WIP - not for merge) #1408

ScaleQ - Updated implementation of LogitQ idea (WIP - not for merge) #1408

AlexisOlson commented Aug 3, 2020 •

edited

Loading

leedavid commented Aug 10, 2020

AlexisOlson commented Aug 10, 2020 •

edited

Loading

AlexisOlson commented Aug 20, 2020

AlexisOlson commented Sep 8, 2020

Naphthalin commented Nov 2, 2022

Naphthalin commented Jun 6, 2023

ScaleQ - Updated implementation of LogitQ idea (WIP - not for merge) #1408

ScaleQ - Updated implementation of LogitQ idea (WIP - not for merge) #1408

Conversation

AlexisOlson commented Aug 3, 2020 • edited Loading

leedavid commented Aug 10, 2020

AlexisOlson commented Aug 10, 2020 • edited Loading

AlexisOlson commented Aug 20, 2020

AlexisOlson commented Sep 8, 2020

Naphthalin commented Nov 2, 2022

Naphthalin commented Jun 6, 2023

AlexisOlson commented Aug 3, 2020 •

edited

Loading

AlexisOlson commented Aug 10, 2020 •

edited

Loading