The Nineteenth Byte

Anonymous

00:23

$ python3 quartata.py < onion.txt
Using TensorFlow backend.
Illegal instruction (core dumped)

Anonymous

@quartata what do

quartata

00:34

what CPU do you have

I suspect the prebuilt version you got is trying to use something like AVX

Anonymous

@quartata gist.github.com/Mego/266b85ad9346c655d50e265c0c5fd7f1

quartata

yep

OK uhh

you have two options

build Tensorflow from source

or downgrade to 1.5 (?) I think those weren't built with AVX

Anonymous

That's not hard

Downgoat

@quartata oh god this is the worst

Anonymous

I'll work on that tonight

Downgoat

00:38

bazel is also the worst

@Mego idk if I've mentioned but google colab lets you run a python notebook and comes pre-installed with python/tensorflow/keras and all that fun stuff: colab.research.google.com

Anonymous

04:04

@quartata So it's complaining because it wants libmpfr4 and I have libmpfr6... This is the worst

Anonymous

Oh cool it's just failing in general

quartata

Torch does too, go ahead and make a symlink

Anonymous

This is why ML people drink

quartata

suffering an identity crisis rn

Anonymous

this rule is missing dependency declarations for the following files included by 'external/protobuf_archive/src/google/protobuf/compiler/ruby/ruby_generator.cc':

Anonymous

04:15

Can you translate that into penguin?

Anonymous

(followed by basically the entirety of GCC's include)

Anonymous

Literally all --verbose_failures is doing is removing the line that suggests using --verbose_failures...

Anonymous

@quartata I went ahead and just used 1.5.0 from pip. Should I be concerned about this?

venv/lib/python3.6/site-packages/keras/engine/saving.py:304: UserWarning: Error in loading the saved optimizer state. As a result, your model is starting with a freshly initialized optimizer.

quartata

04:32

Nope

Anonymous

Cool

Anonymous

I posted hidden test scores for you and the goat

quartata

1.5 probably doesn't have Adam which is the optimizer I used in training

The paper only came out in 2015 I think?

But since you're not training it it doesn't matter

He got 84.5%! Yikes!

Time for me to get off my ass

Anonymous

Lol

quartata

I probably made the activation threshold too aggressive

It helped on one of my sets

But might not generalize as well

Anonymous

04:35

I should probably figure out why my Linux box doesn't want to run gcc

quartata

At any rate something is wrong with my model I just haven't figured it out yet

One of us will crack 90

I'm confident

New Main Posts

05:20

0

Q: How to use slice and hstack in python

Is there an easy way to vectorize this? Thanks def expand(X): X_expanded = np.zeros((X.shape[0], 6)) for i in range(X.shape[0]): X_expanded[i,0]=X[i,0] X_expanded[i,1]=X[i,1] X_expanded[i,2]=X[i,0]**2 X_expanded[i,3]=X[i,1]**2 X_expanded[i,4]=X[i,...

New Main Posts

09:50

0

Q: Approximate the value of e using continued fraction (recursion)

This is what I had in mind, but it only works when input = 2 . Please help me point out the mistake in the for loop. Thanks! n = int(input("Enter your approximation: ")) denom = 0 num = 0 denom_2 = 1 for i in range ( 1, n ): denom_2 += ((i) + (i + 1)/(i + 2)) num = i / denom_2 f...

code-challenge python programming-puzzle

Downgoat

14:04

@quartata oh wow dang

quartata

Probably some data there that was in your training set (same for me for that matter)

but it's all good because I figured out what's wrong with my model

Downgoat

14:20

@quartata What was it?

quartata

14:30

actually two things, one of which was punctuation

Downgoat

14:49

@quartata you added or removed that?

Conor O'Brien

18:38

@EriktheOutgolfer can the language be adjusted slightly after that date?

Erik the Outgolfer

why not

I can't see any term which states that "your submission is immutable" or something like that

the deadline is in 2 days though! (not that anybody wins, especially since the only entry is Mr. Xcoder's RRange)

Conor O'Brien

oh right

@EriktheOutgolfer Tidy is my submission for your challenge

(readme is still incomplete)

Erik the Outgolfer

languages are usually based on implementation here, so, if you modify your language, your submission doesn't change (but that doesn't mean it gets any worse :P)

Conor O'Brien

kk

Erik the Outgolfer

wow, 528 lines already for the main file

Zacharý

19:10

@ConorO'Brien what happened to the D rewrite f Attache?

quartata

attached

please pay me royalties if you use thta name.

Conor O'Brien

19:30

@Zacharý I wanted more experience with writing and dealing with tagged unions in D. specifically I want to figure out a nice, concise way of writing overloaded, typed functions.

@quartata Datache

Zacharý

@ConorO'Brien Ah, so you don't having to write a wrapper around an Algebraic in order to get operators to work?

Conor O'Brien

@Zacharý I wasn't planning on it. I suppose I could read up on the specifics of Algebraic and see if it fit my purposes. I've also decided I should really "close" Attache before I start rewriting it. seeing as how attache isn't yet done, I don't see much point in progressing

Zacharý

@ConorO'Brien I was referring to the hack that I had to do for RAD. I couldn't find a way to overload all the operators at once, so I resorted to a wrapper struct.

Conor O'Brien

have you tried using mixin?

like this?

@Zacharý if all you're writing is a wrapper type, using mixin + opBinary overloads should be enough

Zacharý

... I FORGOT ABOUT MIXIN!

Conor O'Brien

19:44

mixin is a truly wonderful thing :) I kinda forgot about it too

Zacharý

Anyways, that still might not work, because I'd have to do .get! and .peek! trickery, which I don't think will work with CTFE.

Conor O'Brien

all mixin does is insert the operator into your code literally. if you can write code for operator + and - the same way, you can use mixin

Zacharý

What I'd be looking for is a ⍟ b => a.get!A ⍟ b.get!B, where A and B are the current types held in a and b. mixin can probably still clean up some of the code though.

Conor O'Brien

assuming a and b are algebraics? or are they variants

Zacharý

@ConorO'Brien I tried variants, and that issue with scoping happened. So I switched to algebraics.

I think I might be able to get something working...

@ConorO'Brien Do you know about alias ... this

Conor O'Brien

19:52

@Zacharý you mean somethingl ike alias AtomValue = Algebraic!(int, string); ?

Zacharý

@ConorO'Brien No: like this

Conor O'Brien

wtf that looks amazing

Zacharý

Yeah, RAD abuses it (but not in a good way, since Algebraics don't do operator overloading)

Conor O'Brien

small proof of concept?

TLDR you can operator overload with algebraics

Downgoat

zaidalyafeai.github.io/pix2pix/cats.html

Zacharý

19:57

Yeah, one would have to type out every combination of types though.

Conor O'Brien

I think this is the most meaningful way to declare how operators work with different types

@Zacharý this is better

Zacharý

@ConorO'Brien I can't do static if on a call to peek and/or get

tio.run/##rVNNT8MwDL33V3g9TImYEIwb1Q5w4A8wcZl2yLpsteiSKU0HCJW/…

Conor O'Brien

21:49

@Zacharý I think the problem there is that you're trying to precompile int + int[] which doesn't work

Zacharý

I know that's the issue, I'm saying that's why I can't use mixin for my purposes.

Conor O'Brien

what you need to do is overload that type of operation to throw a runtime error

Zacharý

22:03

@ConorO'Brien If I'm using a mixin, then how so?

Conor O'Brien

the best approach I can think of is to define addition manually... I'll give it some more thought

Zacharý

@ConorO'Brien thus every operation. I've already implemented something like that for RAD

WTF do I keep referring to that atrocity instead of RAD?!

@ConorO'Brien What you could do is implement every operation into groups, depending on which operators error for what type combos.

Or maybe do some trait-trickery.

Conor O'Brien

traits are compile time, still wouldn't help you

Zacharý

Sorry, I meant to say: or using traits to group them, if that's possible for operators.

Conor O'Brien

maybe using some combination of functions+a type hash? that's my approach for stacked, it seems feasible given that D can compile the functions before execution

e.g. make a type typeIndicator then make an aa for typeIndicator and functions which work with an instance

Zacharý

22:15

something like this is what I'm talking about with "grouping".

@ConorO'Brien Maybe.

Conor O'Brien

btw op.among!("+", "-", "/", "*") might be clearer (std.algorithm.comparison : among)

also don't forget the __compiles trait, you could use some trickery with that maybe

Zacharý

22:31

Isn't it __traits(compiles, ...)?

tio.run/##tVTPS8MwFL73r3jrYSQ4BnotHhQFD4OBDD2WtI3dwzQtSbopo/…

...if only D had macros :-(

Mr. Xcoder

23:00

> This account is temporarily suspended network-wide. The suspension period ends on Mar 18 '92 at 16:28.

3

A 74-year suspension? Holy frick, SE!

New Main Posts

0

Q: find coordinates in GPS tracker hexadecimal payload

I bought a GPS tracker and set the server to my local one and I'm getting this payload from the tracker (just a TCP dump). I tried to find some coordinate in the payload but I don't know how to convert/decode it. 7878 0D01 0353 6022 0000 7249 0007 1831 0D0A 7878 0D01 0353 6022 0000 7249 0008 E0...

hexadecimal reverse-engineering

Conor O'Brien

@Zacharý I feel like there's a clever way to use templates here...

Anonymous

@Mr.Xcoder That's typically reserved for repeatedly bad behavior across multiple sites, with no signs of improving from previous suspensions. CMs rarely do that, so they must have had a really good reason.

quartata

23:30

It's usually only 30 years though

Anonymous

Normally, when mods issue more-or-less-permanent suspensions, we use 9999 days (~27 years), which (I think) is the maximum number of days a mod can suspend a user. Since this was applied network-wide, it came from SE, who don't have the same restrictions.

New Sandboxed Posts

0

A: Sandbox for Proposed Challenges

Polite Near-Sighted Drunk Bot on a Minefield As the title may suggest, this problem is semi-inspired by the Polite Near-Sighted Drunk Bot by @N.P. Our poor bot is placed on a cartesian grid at the origin, and after each minute, it moves 1 unit in one of four directions (Up, Down, Left, Right)....

New Sandboxed Posts

23:54

0

A: Sandbox for Proposed Challenges

Should I index like this or like that? code-golf integer decision-problem Challenge Let's take a 0-indexed array of length \$l\$. You can index into it with \$i\$, where \$-l\le i<l\$, and the returned element is the element in position \$[i<0]\times l+i\$ (\$[\dots]\$ is the Iverson bracket). ...

Transcript for