Barry G Becker

Error running pip install dlgo

8

I'm trying to follow the instructions to install dlgo. When I run `pip install dlgo` I get ``` Collecting dlgo Using cached https://files.pythonhosted.org/packages/05/c7/cce2a35bb0a2c1c33c2c222e868055fe6f8bc584d73b81d22c5c5bcf9809/dlgo-0.2-py3-none-any.whl Collecting future (from dlgo) Using cached https://files.pythonhosted.org/packages/90/52/e20466b85000a181e1e144fd8305caf2cf475e2f9674e797b222f8105f5f/future-0.17.1.tar.gz...

I noticed that the way that the game state is sent to the input of the NN (see ZeroEncoder) is different from what is described [here](https://applied-data.science/static/main/res/alpha_go_zero_cheat_sheet.png). In the cheat sheet,...

Can weights from leela-zero training be used to bootstrap the ScalphaGoZero network?

3

#### Question I assume that ScalphaGoZero will not play very well without a whole lot of training first. Leela-zero has been training its network for over a year now. Both...

spark-admin link is broken in doc

2

0.8.1-SNAPSHOT on master This is very minor, but the "spark-admin" hyperlink near the bottom of the README.md is broken. It points to https://github.com/adatao/spark-admin - which is not found. Where should...

The job server UI does not show request json if called with sync=true

6

We make two sorts of json requests to the job-server: synchronous (sync=true) and asynchronous. For both sorts of requests, the initial job-id link in the Web UI gives the json...

Passing a large json request to jobServer causes it to fail (whether using JobFileDao or JobSqlDao)

9

spark 2.1.1 and also spark 2.3.0 spark 0.7.0, 0.8.0, 0.8.1-SNAPSHOT client mode Different (but similar) wrong behavior depending on whether you are using JobFileDao or JobSqlDao when issueing a large...

JobStatus sometimes changes from RUNNING to something else, then back to RUNNING

6

spark 2.3.0 tip of sjs 0.8.1 Spark Standalone JobSqlDao using MySql (really mariadb). This is not a problem when using JobFileDao. I have some cucumber bdd tests that I run...

Instead of an infinite loop, there should be an exception thrown. Granted this is an edge case, but probably should be handled.

If the DendrogramFrame were split out into a separate class, it would be easier to re-use. If the constructor took a Cluster parameter, then you could do something like `new...

I don't think globalIndex should be static in ClusterPair. It means that it just keeps getting incremented for every clustering that you do in a single instantiation of a JVM....