tensorflow/minigo issues and pull requests

#100 - Consider active learning approach

Issue - State: open - Opened by brilee almost 7 years ago - 12 comments

#99 - update RESULTS.md with links to our 9x9 files

Pull Request - State: closed - Opened by amj almost 7 years ago
Labels: size/M

#98 - Does the global_step go back to 0 every time we train a new model?

Issue - State: closed - Opened by weedwind almost 7 years ago - 1 comment

#97 - If the SGF says W+R (or B+R), who wins?

Issue - State: closed - Opened by weedwind almost 7 years ago - 1 comment

#96 - selfplay function only plays 1 game, can we make it play many games in one call

Issue - State: closed - Opened by weedwind almost 7 years ago - 7 comments

#95 - Add Komi Feature

Issue - State: open - Opened by artasparks almost 7 years ago - 2 comments

#94 - Where to see who wins when playing with gnu go

Issue - State: closed - Opened by weedwind almost 7 years ago - 1 comment

#93 - main.evaluate needs updating for vlosses, currently does not run.

Issue - State: closed - Opened by weedwind almost 7 years ago - 9 comments

#92 - Adds validate call to rl_runner. Fixes #43

Pull Request - State: closed - Opened by amj almost 7 years ago - 1 comment
Labels: lgtm, size/S

#91 - Comment or code problem.

Issue - State: closed - Opened by marcocalignano almost 7 years ago - 7 comments

#90 - Create a Google Group for Discussions

Issue - State: closed - Opened by artasparks almost 7 years ago - 2 comments

#89 - Move python code to subdirectory

Issue - State: open - Opened by artasparks almost 7 years ago

#88 - Upload SGFs without debug comments

Issue - State: closed - Opened by artasparks almost 7 years ago - 8 comments

#87 - Use dynamic parent Q for children's prior

Pull Request - State: closed - Opened by killerducky almost 7 years ago - 8 comments
Labels: size/S

#86 - Re-add Set gpu_options.allow_growth = True

Pull Request - State: closed - Opened by killerducky almost 7 years ago - 4 comments
Labels: size/XS

#85 - Fix shbang

Pull Request - State: closed - Opened by artasparks almost 7 years ago - 2 comments
Labels: size/XS

#84 - Test for "symmetry divergence" during training

Issue - State: open - Opened by brilee almost 7 years ago - 2 comments

#83 - Test out the harness.

Pull Request - State: closed - Opened by artasparks almost 7 years ago - 3 comments
Labels: size/L

#82 - Fix some lint errors.

Pull Request - State: closed - Opened by artasparks almost 7 years ago
Labels: size/XS

#81 - Improve coords.py

Pull Request - State: closed - Opened by nealwu almost 7 years ago - 4 comments
Labels: lgtm, size/XL

#80 - Add `validate` to DualNetworkTrainer. Add command to main.py

Pull Request - State: closed - Opened by amj almost 7 years ago - 9 comments
Labels: size/L

#79 - Add a V2 prow-bootstrap.py harness config.

Pull Request - State: closed - Opened by artasparks almost 7 years ago - 3 comments
Labels: size/L, lgtm

#78 - Test the harness

Pull Request - State: closed - Opened by artasparks almost 7 years ago - 6 comments
Labels: size/XS

#77 - Is it possible to install minigo on windows 10?

Issue - State: open - Opened by zxo102 almost 7 years ago - 4 comments

#76 - Change the name of the test-harness image to make its purpose clear.

Pull Request - State: closed - Opened by artasparks about 7 years ago

#75 - Pausable worker

Pull Request - State: closed - Opened by amj about 7 years ago

#74 - Make training job 'pause' if there aren't enough new games.

Pull Request - State: closed - Opened by amj about 7 years ago

#73 - How to automatically adjust the resign threshold?

Issue - State: closed - Opened by amj about 7 years ago - 4 comments

#72 - Use only a requirements.txt

Pull Request - State: closed - Opened by artasparks about 7 years ago
Labels: enhancement

#71 - hotfixes from 9s cluster

Pull Request - State: closed - Opened by amj about 7 years ago

#70 - Update the Bucket ACLs so that the service account can read the models

Pull Request - State: closed - Opened by artasparks about 7 years ago

#69 - Small changes to dockerfiles

Pull Request - State: closed - Opened by artasparks about 7 years ago

#68 - update Dockerfile to build tensorflow from source, optimized for extended CPU instructions

Pull Request - State: closed - Opened by amj about 7 years ago

#67 - Add boardsize to bucket.

Pull Request - State: closed - Opened by artasparks about 7 years ago

#66 - Lots more GPU / CPU K8S cluster updates

Pull Request - State: closed - Opened by artasparks about 7 years ago

#65 - import sys for sys.exit() on line 38

Pull Request - State: closed - Opened by cclauss about 7 years ago - 1 comment

#64 - from mcts import MAX_DEPTH for line 193

Pull Request - State: closed - Opened by cclauss about 7 years ago

#63 - Implement gradient norm logging

Pull Request - State: closed - Opened by brilee about 7 years ago - 3 comments

#62 - Fix a comment in cluster-up-gpu

Pull Request - State: closed - Opened by artasparks about 7 years ago

#61 - A better solution for hyperparams

Issue - State: closed - Opened by brilee about 7 years ago - 15 comments

#60 - Pass around saved/exported models everywhere

Issue - State: closed - Opened by amj about 7 years ago - 7 comments

#59 - Hotfixes from running-head

Pull Request - State: closed - Opened by amj about 7 years ago - 4 comments

#58 - Add custom compiling of TensorFlow for CPU Image

Issue - State: closed - Opened by artasparks about 7 years ago - 3 comments

#57 - update selfplay to write 5% of games to a holdout directory

Pull Request - State: closed - Opened by amj about 7 years ago

#56 - Add names to the output tensors for graph exports.

Pull Request - State: closed - Opened by amj about 7 years ago

#55 - Forgot to import sys

Issue - State: closed - Opened by FrankRouter about 7 years ago - 1 comment

#54 - Upgrade a bunch of Deps

Pull Request - State: closed - Opened by artasparks about 7 years ago - 11 comments

#53 - Decide on learning decay rate and target games/generation numbers for 9s

Issue - State: closed - Opened by amj about 7 years ago - 1 comment

#52 - Fix bad result not respecting resign

Pull Request - State: closed - Opened by brilee about 7 years ago

#51 - python3 not python and also not sure how to do it locally

Issue - State: closed - Opened by kwccoin about 7 years ago - 3 comments

#50 - Reduce history length requirement

Issue - State: open - Opened by brilee about 7 years ago - 2 comments

#49 - Clean up evaluation.py and add a bunch of comments.

Pull Request - State: closed - Opened by artasparks about 7 years ago

#48 - Clean up the cluster scripts

Pull Request - State: closed - Opened by artasparks about 7 years ago - 1 comment

#47 - Terminate at depth

Pull Request - State: closed - Opened by amj about 7 years ago - 1 comment

#46 - Support more GTP commands

Issue - State: open - Opened by amj about 7 years ago - 3 comments
Labels: enhancement, good first issue

#45 - faster opening statistics in C

Pull Request - State: closed - Opened by marcocalignano about 7 years ago - 7 comments

#44 - Create evaluation server

Issue - State: closed - Opened by brilee about 7 years ago - 4 comments

#43 - Create validation set (self-play and professional)

Issue - State: closed - Opened by brilee about 7 years ago - 5 comments

#42 - Add monitoring of gradient magnitude

Issue - State: closed - Opened by brilee about 7 years ago - 1 comment

#41 - Implement progressive widening.

Pull Request - State: closed - Opened by killerducky about 7 years ago - 5 comments

#40 - Create a prow test harness

Pull Request - State: closed - Opened by artasparks about 7 years ago - 2 comments

#39 - Make sure the directory has been created in make file

Pull Request - State: closed - Opened by artasparks about 7 years ago

#38 - Several python cleanups

Pull Request - State: closed - Opened by artasparks about 7 years ago - 1 comment

#37 - softmax_cross_entropy_with_logits deprecated

Issue - State: closed - Opened by artasparks about 7 years ago - 7 comments

#36 - TensorFlow: DeprecationWarnings

Issue - State: closed - Opened by artasparks about 7 years ago - 1 comment

#35 - Add some automatic python linting

Issue - State: closed - Opened by artasparks about 7 years ago - 5 comments

#34 - Add some notes about code formatting.

Pull Request - State: closed - Opened by artasparks about 7 years ago

#33 - Set gpu_options.allow_growth = True

Pull Request - State: closed - Opened by killerducky about 7 years ago - 1 comment

#32 - Move Python files into a py subdir

Pull Request - State: closed - Opened by artasparks about 7 years ago - 1 comment

#31 - Script for looking at model accuracy on pro positions

Pull Request - State: closed - Opened by jmgilmer about 7 years ago

#30 - Autoformat the Python

Issue - State: closed - Opened by artasparks about 7 years ago

#29 - Autofmt all python

Pull Request - State: closed - Opened by artasparks about 7 years ago - 7 comments

#28 - Using too much GPU memory

Issue - State: closed - Opened by killerducky about 7 years ago

#27 - V2 running head

Pull Request - State: closed - Opened by amj about 7 years ago - 1 comment

#26 - Improve worker/training balance.

Issue - State: closed - Opened by amj about 7 years ago - 6 comments

#25 - Decide on learning rate decay schedule.

Issue - State: closed - Opened by amj about 7 years ago - 4 comments

#24 - Integrate with Automatic/Continuous Testing

Issue - State: closed - Opened by artasparks about 7 years ago - 5 comments

#23 - Change some errors to go to stderr.

Pull Request - State: closed - Opened by killerducky about 7 years ago - 1 comment

#22 - Minor README fixes

Pull Request - State: closed - Opened by killerducky about 7 years ago - 3 comments

#21 - Redo selfplay termination after 450 moves

Issue - State: closed - Opened by amj about 7 years ago

#20 - Integrate with Cloud TPU

Issue - State: closed - Opened by artasparks about 7 years ago - 13 comments

#19 - Investigate/Monitor preemption rate

Issue - State: closed - Opened by artasparks about 7 years ago - 2 comments

#18 - Hook up Google CLA bot.

Issue - State: closed - Opened by artasparks about 7 years ago - 1 comment

#17 - Add code of conduct.

Pull Request - State: closed - Opened by artasparks about 7 years ago - 1 comment

#16 - Code of Conduct

Issue - State: closed - Opened by artasparks about 7 years ago

#15 - Choose a python code formatter and format the code

Issue - State: closed - Opened by artasparks about 7 years ago - 2 comments

#14 - Rewrite self-play in more performant language

Issue - State: closed - Opened by artasparks about 7 years ago - 7 comments

#13 - Need better story for shared images/data

Issue - State: closed - Opened by artasparks about 7 years ago - 1 comment

#12 - Have a way to validate accuracy with a test set of pro games

Issue - State: closed - Opened by amj about 7 years ago - 5 comments

#11 - Feature: nice gogui/GTP extensions

Issue - State: closed - Opened by amj about 7 years ago - 3 comments
Labels: good first issue

#10 - CGOS play improvements

Issue - State: closed - Opened by amj about 7 years ago
Labels: good first issue

#9 - SGF Analysis tools

Issue - State: closed - Opened by amj about 7 years ago - 2 comments
Labels: enhancement

#8 - Implement superko

Issue - State: closed - Opened by brilee about 7 years ago - 8 comments

#7 - Clean up naming in coords.py

Issue - State: closed - Opened by brilee about 7 years ago
Labels: good first issue

GitHub / tensorflow/minigo issues and pull requests