This is a pure Python implementation of a neural-network based Go AI, using TensorFlow. While inspired by DeepMind's AlphaGo algorithm, this project is not a DeepMind project nor is it affiliated with the official AlphaGo project.

Repeat, this is not the official AlphaGo program by DeepMind. This is an independent effort by Go enthusiasts to replicate the results of the AlphaGo Zero paper ("Mastering the Game of Go without Human Knowledge," Nature), with some resources generously made available by Google.

Minigo is based off of Brian Lee's "MuGo" -- a pure Python implementation of the first AlphaGo paper "Mastering the Game of Go with Deep Neural Networks and Tree Search" published in Nature. This implementation adds features and architecture changes present in the more recent AlphaGo Zero paper, "Mastering the Game of Go without Human Knowledge". More recently, this architecture was extended for Chess and Shogi in "Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm". These papers will often be abridged in Minigo documentation as AG (for AlphaGo), AGZ (for AlphaGo Zero), and AZ (for AlphaZero) respectively.

1. Provide a clear set of learning examples using Tensorflow, Kubernetes, and Google Cloud Platform for establishing Reinforcement Learning pipelines on various hardware accelerators.

2. Reproduce the methods of the original DeepMind AlphaGo papers as faithfully as possible, through an open-source implementation and open-source pipeline tools.

3. Provide our data, results, and discoveries in the open to benefit the Go, machine learning, and Kubernetes communities.

An explicit non-goal of the project is to produce a competitive Go program that establishes itself as the top Go AI. Instead, we strive for a readable, understandable implementation that can benefit the community, even if that means our implementation is not as fast or efficient as possible.

While this product might produce such a strong model, we hope to focus on the process. Remember, getting there is half the fun. :)

We hope this project is an accessible way for interested developers to have access to a strong Go model with an easy-to-understand platform of python code available for extension, adaptation, etc.

If you'd like to read about our experiences training models, see RESULTS.md.

To see our guidelines for contributing, see CONTRIBUTING.md.

This project assumes you have the following:

• virtualenv / virtualenvwrapper
• Python 3.5+
• Docker

The Hitchhiker's guide to python has a good intro to python development and virtualenv usage. The instructions after this point haven't been tested in environments that are not using virtualenv.

pip3 install virtualenv
pip3 install virtualenvwrapper

First set up and enter your virtualenv. Then start by installing TensorFlow and the dependencies:

pip3 install -r requirements.txt

The requirements.txt file assumes you'll use a GPU; if you wish to run on GPU you must install CUDA 8.0 or later (see TensorFlow documentation).

If you don't want to run on GPU or don't have one, you can downgrade:

pip3 uninstall tensorflow-gpu
pip3 install tensorflow

Or just install the CPU requirements:

pip3 install -r requirements-cpu.txt

You may want to use a cloud project for resources. If so set:

PROJECT=foo-project

Then, running

source cluster/common.sh

will set up other environment variables defaults.

BOARD_SIZE=9 python3 -m unittest discover tests

All commands are compatible with either Google Cloud Storage as a remote file system, or your local file system. The examples here use GCS, but local file paths will work just as well.

To use GCS, set the BUCKET_NAME variable and authenticate via gcloud login. Otherwise, all commands fetching files from GCS will hang.

For instance, this would set a bucket, authenticate, and then look for the most recent model.

export BUCKET_NAME=your_bucket;
gcloud auth application-default login
gsutil ls gs://minigo/models | tail -3

Which might look like:

gs://$BUCKET_NAME/models/000193-trusty.data-00000-of-00001
gs://$BUCKET_NAME/models/000193-trusty.index
gs://$BUCKET_NAME/models/000193-trusty.meta

These three files comprise the model, and commands that take a model as an argument usually need the path to the model basename, e.g. gs://$BUCKET_NAME/models/000193-trusty

You'll need to copy them to your local disk. This fragment copies the latest model to the directory specified by MINIGO_MODELS

MINIGO_MODELS=$HOME/minigo-models
mkdir -p $MINIGO_MODELS
gsutil ls gs://minigo/models | tail -3 | xargs -I{} gsutil cp "{}" $MINIGO_MODELS

To watch Minigo play a game, you need to specify a model. Here's an example to play using the latest model in your bucket

python rl_loop.py selfplay --readouts=$READOUTS -v 2

where READOUTS is how many searches to make per move. Timing information and statistics will be printed at each move. Setting verbosity (-v) to 3 or higher will print a board at each move.

Minigo uses the [http://www.lysator.liu.se/~gunnar/gtp/gtp2-spec-draft2/gtp2-spec.html](GTP protocol), and you can use any gtp-compliant program with it.

# Latest model should look like: /path/to/models/000123-something
LATEST_MODEL=$(ls -d $MINIGO_MODELS/* | tail -1 | cut -f 1 -d '.')
python3 main.py gtp $LATEST_MODEL -r $READOUTS -v 3

(If no model is provided, it will initialize one with random values)

After some loading messages, it will display GTP engine ready, at which point it can receive commands. GTP cheatsheet:

genmove [color]             # Asks the engine to generate a move for a side
play [color] [coordinate]   # Tells the engine that a move should be played for `color` at `coordinate`
showboard                   # Asks the engine to print the board.

One way to play via GTP is to use gogui-display (which implements a UI that speaks GTP.) You can download the gogui set of tools at http://gogui.sourceforge.net/. See also documentation on interesting ways to use GTP.

gogui-twogtp -black 'python3 main.py gtp gs://$BUCKET_NAME/models/000000-bootstrap' -white 'gogui-display' -size 19 -komi 7.5 -verbose -auto

Another way to play via GTP is to watch it play against GnuGo, while spectating the games

BLACK="gnugo --mode gtp"
WHITE="python3 main.py gtp path/to/model"
TWOGTP="gogui-twogtp -black \"$BLACK\" -white \"$WHITE\" -games 10 \
  -size 19 -alternate -sgffile gnugo"
gogui -size 19 -program "$TWOGTP" -computer-both -auto

The following sequence of commands will allow you to do one iteration of reinforcement learning on 9x9. These are the basic commands used to produce the models and games referenced above.

The commands are

• bootstrap: initializes a random model
• selfplay: plays games with the latest model, producing data used for training
• gather: groups games played with the same model into larger files of tfexamples.
• train: trains a new model with the selfplay results from the most recent N generations.

This command creates a random model, which appears at . gs://$BUCKET_NAME/models/$MODEL_NAME(.index|.meta|.data-00000-of-00001)

export MODEL_NAME=000000-bootstrap
python3 main.py bootstrap gs://$BUCKET_NAME/models/$MODEL_NAME

This command starts self-playing, outputting its raw game data in a tensorflow-compatible format as well as in SGF form in the directories

gs://$BUCKET_NAME/data/selfplay/$MODEL_NAME/local_worker/*.tfrecord.zz
gs://$BUCKET_NAME/sgf/$MODEL_NAME/local_worker/*.sgf

python3 main.py selfplay gs://$BUCKET_NAME/models/$MODEL_NAME \
  --readouts 10 \
  -v 3 \
  --output-dir=gs://$BUCKET_NAME/data/selfplay/$MODEL_NAME/local_worker \
  --output-sgf=gs://$BUCKET_NAME/sgf/$MODEL_NAME/local_worker

python3 main.py gather

This command takes multiple tfrecord.zz files (which will probably be KBs in size) and shuffles them into tfrecord.zz files that are ~100 MB in size.

Gathering is done according to model numbers, so that games generated by one model stay together. By default, rl_loop.py will use directories specified by the environment variable BUCKET_NAME, set at the top of rl_loop.py.

gs://$BUCKET_NAME/data/training_chunks/$MODEL_NAME-{chunk_number}.tfrecord.zz

The file gs://$BUCKET_NAME/data/training_chunks/meta.txtis used to keep track of which games have been processed so far. (more about this needed)

python3 main.py gather \
  --input-directory=gs://$BUCKET_NAME/data/selfplay \
  --output-directory=gs://$BUCKET_NAME/data/training_chunks

This command finds the most recent 50 models' training chunks and trains a new model, starting from the latest model weights.

Run the training job:

python3 main.py train gs://$BUCKET_NAME/data/training_chunks \
    gs://$BUCKET_NAME/models/000001-somename \
    --load-file=gs://$BUCKET_NAME/models/000000-bootstrap \
    --generation-num=1 \
    --logdir=path/to/tensorboard/logs \

The updated model weights will be saved at the end. (TODO: implement some sort of local checkpointing based on global_step that will resume appropriately.)

Additionally, you can follow along with the training progress with TensorBoard - if you give each run a different name (logs/my_training_run, logs/my_training_run2), you can overlay the runs on top of each other.

tensorboard --logdir=path/to/tensorboard/logs/

As you might notice, playing games is fairly slow. One way to speed up playing games is to run Minigo on many computers simultaneously. Minigo was originally trained by containerizing these worker jobs and running them on a Kubernetes cluster, hosted on the Google Cloud Platform.

NOTE These commands will result in VMs being created and will result in charges to your GCP account! Proceed with care!

You'll want to install the following command line tools

• gcloud
• gsutil (via gcloud components install gsutil)
• kubectl (via gcloud components install kubectl)
• docker

In order for each step to work, you'll have to have the following permissions:

• storage.bucket.(create, get, setIamPolicy) ("Storage Admin")
• storage.objects.(create, delete, get, list, update) ("Storage Object Admin")
• iam.serviceAccounts.create ("Service Account Admin")
• iam.serviceAccountKeys.create ("Service Account Key Admin")
• iam.serviceAccounts.actAs ("Service Account User")
• resourcemanager.projects.setIamPolicy ("Project IAM Admin")
• container.clusters.create ("Kubernetes Engine Cluster Admin")
• container.secrets.create ("Kubernetes Engine Developer")

You'll also want to activate Kubernetes Cluster on your GCP account (just visit the Kubernetes Engine page and it will automatically activate.)

A Kubernetes cluster instantiates nodes on a node pool, which specifies what types of host machines are available. Jobs can be run on the cluster by specifying details such as what containers to run, how many are needed, what arguments they take, etc. Pods are created to run individual containers -- the pods themselves run on the nodes.

In our case, we won't let kubernetes resize of our node pool dynamically, we'll manually specify how many machines we want: this means kubernetes will leave machines running even if they're not doing anything! So be sure to clean up your clusters...

The main way these jobs interact is through GCS, a distributed webservice intended to behave like a filesystem.

The selfplay jobs will find the newest model in the GCS directory of models and play games with it, writing the games out to a different directory in the bucket.

The training job will collect games from that directory and turn it into chunks, which it will use to train a new model, adding it to the directory of models, and completing the circle.

0. Switch to the cluster directory

1. Set the common environment variables in common.sh corresponding to your GCP project and bucket names.

2. Run deploy, which will: a. Create a bucket b. Create a service account c. Grant permissions d. Fetch the keys. If any of the above have already been done, the script will fail. At a minimum, run step 'd' to create the keyfile.

3. Run cluster-up or (cluster-up-gpu), which will: a. Create a Google Container Engine cluster with some number of VMs b. Load its credentials locally c. Load those credentials into our kubectl environment, which will let us control the cluster from the command line.

   Creating the cluster might take a while... Once its done, you should be able to see something like this:

   $ kubectl get nodes
   NAME                                  STATUS    ROLES     AGE       VERSION
   gke-minigo-default-pool-b09dcf70-08rp   Ready     <none>    5m        v1.7.8-gke.0
   gke-minigo-default-pool-b09dcf70-0q5w   Ready     <none>    5m        v1.7.8-gke.0
   gke-minigo-default-pool-b09dcf70-1zmm   Ready     <none>    5m        v1.7.8-gke.0
   gke-minigo-default-pool-b09dcf70-50vm   Ready     <none>    5m        v1.7.8-gke.0
   

4. (Optional, GPU only). If you've set up a GPU enabled cluster, you'll need to install the NVIDIA drivers on each of the nodes in your cluster that will have GPU workers. This is accomplished by running:

kubectl apply -f gpu-provision-daemonset.yaml

5. Resizing your cluster. Note that the cluster will not use autoscaling by default, so it's possible to have a lot of idle containers running if you're not careful!

gcloud alpha container clusters resize $CLUSTER_NAME --zone=$ZONE --size=8

You will need a Docker image in order to initialize the pods.

If you would like to override the GCR Project or image tag, you can set:

export PROJECT=my-project
export VERSION=0.1234

Then make will produce and push the image!

CPU worker:

make image
make push

GPU worker:

make gpu-image
make gpu-push

Now that our cluster is set up, lets check some values on our player job before deploying a bunch of copies:

In cluster/player.yaml (or cluster/gpu-player.yaml) check the 'parallelism' field. This is how many pods will try to run at once: if you can run multiple pods per node (CPU nodes), you can set it accordingly. For GPUs, which don't share as well, limit the parallelism to the number of nodes available.

Now launch the job via the launcher. (it just subs in the environment variable for the bucket name, neat!)

source common.sh
envsubst < player.yaml | kubectl apply -f -

Once you've done this, you can verify they're running via

kubectl get jobs

and get a list of pods with

kubectl get pods

Tail the logs of an instance:

kubectl logs -f <name of pod>

To kill the job,

envsubst < player.yaml | kubectl delete -f -

• Check your gcloud -- authorized? Correct default zone settings?

• Check the project name, cluster name, & bucket name variables in the cluster/common.sh script. Did you change things?

  • If Yes: Grep for the original string. Depending on what you changed, you may need to change the yaml files for the selfplay workers.

• Create the service account and bucket, if needed, by running cluster/deploy, or the relevant lines therein.

• Check the number of machines and machine types in the cluster/cluster-up script.

• Set up the cluster as above and start the nvidia driver installation daemonset

• While the nvidia drivers are getting installed on the fleet, check the various hyperparameters and operating parameters:

  • dual_net.py, check the get_default_hyperparams function
  • player_wrapper.sh, the invocation of rl_loop.py selfplay has the readout depth, game parallelism, resign threshold, etc.
  • strategies.py, check the move threshold for move 'temperature' (affects deterministic play), and the max game depth.
  • mcts.py, check the noise density and the tree branching factor (lol good luck)

• Seed the model directory with a randomly initialized model. (python3 rl_loop.py bootstrap /path/to/where/you/want/new/model)

• If you're getting various tensorflow RestoreOp shape mismatches, this is often caused by mixing up 9x9 vs. 19x19 in the various system parts.

• Build your docker images with the latest version of the code, optionally bumping the version number in the Makefile.

• Don't forget to push the images!

• Now you can launch your job on the cluster -- check the parallelism in the spec! -- per the instructions above. You should let the selfplay cluster finish up a bunch of games before you need to start running the training job, so now's a good time to make sure things are going well.

• Getting a list of the selfplay games ordered by start time.

  kubectl get po --sort-by=.status.startTime
  

• Attaching to a running pod (to check e.g. cpu utilization, what actual code is in your container, etc)

  kubectlc exec -it <pod id> /bin/bash
  

• Monitoring how long it's taking the daemonset to install the nvidia driver on your nodes

  kubectl get no -w -o yaml | grep -E 'hostname:|nvidia-gpu'
  

If you've run rsync to collect a set of SGF files (cheatsheet: gsutil -m cp -r gs://$BUCKET_NAME/sgf/$MODEL_NAME sgf/), here are some handy bash fragments to run on them:

• Find the proportion of games won by one color:

  grep -m 1 "B+" **/*.sgf | wc -l
  

  or e.g. "B+R", etc to search for how many by resign etc.

• A histogram of game lengths (uses the 'ministat' package)

  find . -name "*.sgf" -exec /bin/sh -c 'tr -cd \; < {} | wc -c' \; | ministats
  

• Get output of the most frequent first moves

  grep -oh -m 1 '^;B\[[a-s]*\]' **/*.sgf | sort | uniq -c | sort -n
  

• Distribution of game-winning margin (ministat, again):

  find . -name "*.sgf" -exec /bin/sh -c 'grep -o -m 1 "W+[[:digit:]]*" < {} | cut -c3-'
  \; | ministat
  

Also check the 'oneoffs' directory for interesting scripts to analyze e.g. the resignation threshold.