This page is not created by, affiliated with, or supported by Slack Technologies, Inc.
2017-12-20
Channels
- # adventofcode (140)
- # beginners (107)
- # boot (120)
- # boot-dev (1)
- # clara (13)
- # cljs-dev (9)
- # clojure (107)
- # clojure-argentina (5)
- # clojure-art (16)
- # clojure-dev (23)
- # clojure-greece (19)
- # clojure-italy (5)
- # clojure-russia (2)
- # clojure-serbia (3)
- # clojure-spec (27)
- # clojure-sweden (1)
- # clojure-uk (15)
- # clojurescript (134)
- # cursive (5)
- # data-science (10)
- # datomic (23)
- # duct (28)
- # fulcro (48)
- # garden (5)
- # hoplon (2)
- # instaparse (1)
- # klipse (7)
- # leiningen (8)
- # lumo (36)
- # off-topic (72)
- # om (4)
- # onyx (37)
- # perun (4)
- # re-frame (64)
- # reagent (86)
- # remote-jobs (1)
- # shadow-cljs (59)
- # spacemacs (16)
- # sql (1)
- # uncomplicate (6)
- # unrepl (90)
@danielglauser http://pdfdata.io/ is a product written by @cemerick which you might find interesting
@U09MR0T5Y killin it with the community shout outs! Hows the starcraft AI coming along?
Starcraft AI is coming along fine, no neural nets used yet though.. Blizzard recently added a render interface for the raw RGB values. I am now working on a jupyter notebook that runs on my "server", an ubuntu pc with decent specs. It can run a clojure agent and then renders a video of what it did, right there in the notebook! That should make it easier to demonstrate progress, of which the most interesting stuff recently is that I put the game info into a datascript "database". So you can query static information about the units in the game and even have a function to turn the raw observation data into datascript, so you can write rule-based agents.
Thanks @U09MR0T5Y!
Starcraft AI is coming along fine, no neural nets used yet though.. Blizzard recently added a render interface for the raw RGB values. I am now working on a jupyter notebook that runs on my "server", an ubuntu pc with decent specs. It can run a clojure agent and then renders a video of what it did, right there in the notebook! That should make it easier to demonstrate progress, of which the most interesting stuff recently is that I put the game info into a datascript "database". So you can query static information about the units in the game and even have a function to turn the raw observation data into datascript, so you can write rule-based agents.
I should do a write-up of my workflow on clojureverse: http://clojureverse.org/t/share-the-nitty-gritty-details-of-your-clojure-workflow/1208 But haven't even found the time to introduce myself there.
well, the idea of vega is definitely great in form of the grammar of graphics, but can you really compare it to ggplot2, matplotlib or plotly? in my experience there is a huge feature gap with it. also i think if a declarative approach along the grammar of graphics is taken, a react compatible js frontend version would be ideal, otherwise it will not behave well in frontends (we had this problem with vega in a cljs dashboard). so far plotly is sufficient to compete with matplotlib for static plots for me.
i also prefer not to have a wrapper around it, just a way to interact with it directly through the two libraries above, because then i can use the official documentation directly, which is awesome, e.g.: https://plot.ly/javascript/2d-density-plots/
Plotly is a great library, I didn't even consider using it from clojurescript. The benefits of using a hosted language are still sinking in for me.
actually this is the overview: https://plot.ly/javascript/