10 comments

[ 4.7 ms ] story [ 36.4 ms ] thread
Has anyone tried running this? I wonder how long does the training take.
I hope this will not combine safety of Bash with clarity of natural language through reliability of machine learning.
I wonder why it doesn't use Webpack.
I really wish there was a Web service that would let me point it to a GitHub repo and then get an ephemeral URL running Jupyter or something hooked up to a GPU. That would be cool.

I want to poke this for 4.89 seconds, see how bad it is, take a couple of screenshots and move on.

Can we turn this into a NN that explains bash commands in natural language?
“I see you are confused about when to use single square brackets vs doubled based on the number of times you’ve tried both. And failed I might add”
OK Bash: remove objects, recursively, starting at the root directory. Oh, and don't preserve root. Yes, just do it. Thank you!
I didn't look at the raw data, but no awk in the top 50?
It was mentioned in the paper (https://arxiv.org/abs/1802.08979): commands that correspond to non-Bash language interpreters were filtered out from the dataset and `awk` and `sed` were filtered out because they are the interpreters of string processing languages.