Current Events > I would legit give someone like a steam game if they could get this tensorflow

(+) Yeah! (+)

environment working for me.

I can't get it to work. Anyone want to help for free though?

https://github.com/georgwiese/2048-rl

I got to the last step and it doesn't work

Once that's done, you can simple run python py_2048_rl/learning.py <train_dir>.

And it gives me this error:

am I a dummy?

I can do it. I've done tensorflow stuff before.

What game?
---

it's using an old tensorflow

All the commands are depreciated

I will legit pay someone 20 bucks if they get this to run on their ubuntu

I upgraded every single script and it's still not running.

it's giving me a save_path error now. I want to legit kill this dumb fucker

Replace calls of tf.histogram_summary(...) to tf.summary.histogram(...)

You'll need to do similar things with other similar calls (e.g. tf.scalar_summary(...) -> tf.summary.scalar(...))

This is the new API: https://www.tensorflow.org/api_docs/python/tf/summary

edit: whoops, was a little late on this. Give me a sec.
---

I got far in it and it still doesn't work. I already upgraded all the syntax

and it runs, then it gives me weird results (it loses 19% of the time) and errors. it doesn't show a gui anything

idk what this shit is lmao

I give up

Regarding the save_path issue, it's because tf.train.latest_checkpoint(...) keeps returning None for some stupid reason. If you delete the save_path then it won't try resuming from a checkpoint and thus you won't get that error.

But yeah, I've been frustrated when going through ML material. I'm slowly trying to learn.
---