00:00:01 <moon_> its used to end the s command
00:00:06 <moon_> it can be other characters
00:00:22 <oerjan> moon_: hm i'm not sure about your addition to #esoteric. my original seemed funny just _because_ it was out of place, so adding more of the same kind diminishes it.
00:01:27 -!- hppavilion[1] has set topic: The interdisciplinary hub of solidity matrices, esoteric programming language design, and font forging | Ingredients: Antimatter | BOOM! | logs: http://codu.org/logs/_esoteric/ http://tunes.org/~nef/logs/esoteric/?C=M;O=D | http://esolangs.org/ | https://dl.dropboxusercontent.com/u/2023808/wisdom.pdf.
00:03:24 <fizzie> moon_: It can't be "other characters" when the previous two were slashes.
00:03:43 <fizzie> `sed s/prescriptivist/descriptivist_
00:03:45 <HackEgo> sed: -e expression #1, char 31: unterminated `s' command
00:04:00 <fungot> hppavilion[1]: they say that a bag of, sack: summer passed all too quickly. on the road between hurworth and neasham, but of adjudicating in the kidnappings and raids the inhabitants of the babylonian god of boundaries, guardian of the skin first appears greenish, then the handles at each end, all delicately wrought in the sense that he is married to mehut, a cloven hoof, the samurai. it was the ruthless manner in which they ha
00:04:21 <fungot> hppavilion[1]: does your boss know what you're doing right now. while he was nearly burnt out, and in later times for wicked goblin spirits, as near as might be, eighteen or twenty feet across, and write at cos. the chest was divided into three compartments. in " a wizard can do it, is partially absorbed by the hazards since he has sucker feet and is generally more picturesque than useful.
00:04:32 <fungot> hppavilion[1]: meeting your own ghost!
00:04:37 <fizzie> hppavilion[1]: Well, you know what they say: a bag of, sack.
00:05:16 <fungot> Available: agora alice c64 ct darwin discworld enron europarl ff7 fisher fungot homestuck ic irc iwcs jargon lovecraft nethack* oots pa qwantz sms speeches ss wp youtube
00:06:16 <fizzie> There's a lot of stuff in the data.base.
00:06:28 <HackEgo> #esoteric is the only channel that exists. After monqy left it became slightly off-centër. It's about 30 m (100 ft) across. oerjan seems to be making a lawn in the northern part. May contain crude drawings of nuts. Contains only free ranging moons.
00:06:33 <hppavilion[1]> Oh, well I got fizzie's message, so I'm still here
00:06:46 <fizzie> hppavilion[1]: There's a limit of four consecutive replies to the same person.
00:07:01 <fizzie> "The Wumpus, by the way, is not bothered by the hazards since he has sucker feet and is too big for a bat to lift."
00:07:30 <hppavilion[1]> fizzie: ...please tell me you had to look that up. As in, you don't just have NetHack memorized
00:07:36 <oerjan> `` sed -i 's/part/part, but it keeps getting trampled by free ranging moons./;s/ Cont.*//' wisdom/'#esoteric'
00:07:46 <HackEgo> #esoteric is the only channel that exists. After monqy left it became slightly off-centër. It's about 30 m (100 ft) across. oerjan seems to be making a lawn in the northern part, but it keeps getting trampled by free ranging moons.. May contain crude drawings of nuts.
00:08:08 <oerjan> `` sed -i 's/\.\./\./' wisdom/'#esoteric'
00:08:17 <HackEgo> #esoteric is the only channel that exists. After monqy left it became slightly off-centër. It's about 30 m (100 ft) across. oerjan seems to be making a lawn in the northern part, but it keeps getting trampled by free ranging moons. May contain crude drawings of nuts.
00:09:26 <oerjan> <hppavilion[1]> "sucker feet" made me think lovecraft <-- i'd expect nethack to steal from everywhere...
00:10:27 <oerjan> <hppavilion[1]> oerjan: Is HackEgo on a new server now? <-- don't ask me
00:10:38 -!- Phantom_Hoover has quit (Remote host closed the connection).
00:11:16 <moon_> lawn should be garden
00:11:22 <moon_> you cant trample lawns
00:11:29 <fizzie> And no, it's not on a new server.
00:11:34 <oerjan> moon_: no, i can't tell people to get off my garden.
00:11:54 <oerjan> moon_: i'm pretty sure you can.
00:11:55 <fizzie> Usually it's a little bit faster when it's actively being used.
00:12:06 -!- hppavilion[1] has quit (Ping timeout: 244 seconds).
00:15:42 <moon_> `med rr bin/med 1 3
00:15:54 <moon_> `` med rr bin/med 1 3
00:15:57 <HackEgo> #Rewritten by hand \ #info on lines 2 4 9 15 20 26.. continued on line 2 \ #32 38
00:16:10 <oerjan> `` sed -i 's/trampled/digged up/' wisdom/#esoteric
00:16:20 <HackEgo> #esoteric is the only channel that exists. After monqy left it became slightly off-centër. It's about 30 m (100 ft) across. oerjan seems to be making a lawn in the northern part, but it keeps getting digged up by free ranging moons. May contain crude drawings of nuts.
00:17:40 <moon_> `learn_append #esoteric There is a mountain made of shell scripts in the southwestern part.
00:17:43 <HackEgo> Learned '#esoteric': #esoteric is the only channel that exists. After monqy left it became slightly off-centër. It's about 30 m (100 ft) across. oerjan seems to be making a lawn in the northern part, but it keeps getting digged up by free ranging moons. May contain crude drawings of nuts. There is a mountain made of shell scripts in the southweste
00:18:39 <oerjan> i think we've exceeded the length limit.
00:19:03 <HackEgo> rm: cannot remove `/home/hackbot/hackbot.hg/multibot_cmds/env/.hg/store/data/canary.orig': Is a directory \ Done.
00:19:15 <HackEgo> #esoteric is the only channel that exists. After monqy left it became slightly off-centër. It's about 30 m (100 ft) across. oerjan seems to be making a lawn in the northern part, but it keeps getting digged up by free ranging moons. May contain crude drawings of nuts.
00:19:30 <oerjan> `` sed -i 's/digged/dug/' wisdom/#esoteric
00:19:46 <moon_> `learn_append #esoteric There is a mountain made of shell scripts in the south. exit 0
00:19:49 <HackEgo> Learned '#esoteric': #esoteric is the only channel that exists. After monqy left it became slightly off-centër. It's about 30 m (100 ft) across. oerjan seems to be making a lawn in the northern part, but it keeps getting dug up by free ranging moons. May contain crude drawings of nuts. There is a mountain made of shell scripts in the south. exit 0
00:21:06 * oerjan gives up and sidles back to retiring.
00:21:18 <oerjan> moon_: also, get off my lawn!
00:23:04 * oerjan waves his cane O======
00:31:35 <zzo38> On my computer design I have written about "four level mode" which is the intention of working high resolution monochrome mode without having to change all of the clock rates
00:43:50 <\oren\> digged up means that it went uo the page on digg
00:44:15 <\oren\> by the way, is everyone aware that neopets still exists?
00:49:32 -!- lambda-11235 has joined.
00:49:38 <\oren\> @tell hppavilion[1] I'm overprescriptive myself
00:50:02 <zzo38> I did not expect that it was gone
00:50:19 <zzo38> Therefore, the default position I believe was still exists
00:50:49 -!- hwpplayer1 has left.
00:51:24 <oerjan> \oren\: digg still exists?
00:51:31 -!- t0dd has joined.
00:52:04 -!- t0dd has quit (Read error: Connection reset by peer).
00:52:06 <HackEgo> t0dd: Welcome to the international hub for esoteric programming language design and deployment! For more information, check out our wiki: <http://esolangs.org/>. (For the other kind of esoterica, try #esoteric on EFnet or DALnet.)
00:52:53 <oerjan> HackEgo: you were fast, but not fast enough
01:01:30 -!- Kaynato has quit (Read error: Connection reset by peer).
01:02:21 -!- Kaynato has joined.
01:05:19 <zzo38> Is there anything you think you need Internet Quiz Engine to do that it does not currently have? I already have conditional questions, multiple-answers questions, arithmetic operations, multiple variables, and others.
01:06:06 <shachaf> I think it should support free-form answers.
01:07:49 <zzo38> Yes, I did think of that too, although am unsure of how to implement it properly. I did think of a few ideas though.
01:08:01 <HackEgo> /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: exec: `digg: not found
01:08:06 <HackEgo> /hackenv/bin/`: line 4: digg: command not found
01:08:33 <zzo38> Clearly there need to be some way to be able to parse free-form answers.
01:11:36 <shachaf> zzo38: Maybe use Amazon Mechanical Turk.
01:13:04 <zzo38> No, it should do by itself.
01:14:02 <zzo38> For example, what kind of calculations does it need to be able to do with the free-form answers?
01:14:56 <shachaf> Number of words, number of letters.
01:16:34 <zzo38> Comparing text with constants also seem one thing that should be done
01:17:38 -!- moon_ has quit (Ping timeout: 250 seconds).
01:20:44 -!- augur has joined.
01:25:45 -!- augur has quit (Ping timeout: 260 seconds).
01:28:36 -!- ybden has quit (Quit: SIGTUNA).
01:48:38 -!- bender has joined.
02:01:35 -!- bender has quit (Ping timeout: 258 seconds).
02:14:04 -!- erdic has quit (Ping timeout: 260 seconds).
02:15:33 -!- erdic has joined.
03:15:48 -!- augur has joined.
03:50:26 -!- hppavilion[1] has joined.
04:25:09 -!- proderpH has joined.
04:31:07 -!- hppavilion[1] has quit (Ping timeout: 252 seconds).
04:34:02 -!- hppavilion[1] has joined.
04:43:59 -!- proderpH has quit (Remote host closed the connection).
04:45:42 -!- xkapastel has quit (Quit: Connection closed for inactivity).
05:07:24 -!- lambda-11235 has quit (Ping timeout: 276 seconds).
05:08:18 -!- lambda-11235 has joined.
05:14:41 -!- lambda-11235 has quit (Max SendQ exceeded).
05:15:15 -!- lambda-11235 has joined.
05:20:37 -!- hppavilion[1] has quit (Ping timeout: 252 seconds).
05:25:41 -!- Kaynato has quit (Ping timeout: 240 seconds).
05:54:41 -!- hppavilion[1] has joined.
05:55:04 <zzo38> Can you show example of how a quiz that is using free-form answers might do?
05:55:38 <shachaf> A question might be "Can you show example of how a quiz that is using free-form answers might do?"
05:56:40 <zzo38> It is unclear how the computer is supposed to score such a question though.
05:58:35 <hppavilion[1]> Is there any equivalent to the DSM-n that documents "psychological disorders" picked up by AIs?
05:58:44 <lambdabot> \oren\ said 5h 9m 5s ago: I'm overprescriptive myself
06:00:31 <zzo38> What would you think of the following: {?} Creature - ? (1/1) ;; Last strike, Lifelink ;; ~ gets +1/+1 for each point of damage marked on it up to eleven.
06:02:27 <shachaf> zzo38: What would happen when that creature combated a 3/3?
06:03:40 <zzo38> Now it has three points of damage marked on it but 4 toughness, and also 4 power so during the last strike step it does enough damage to destroy the 3/3 creature and you gain 4 life.
06:04:09 <zzo38> If the 3/3 creature it is blocking or blocked by has infect or wither then this does not happen, since it won't mark any damage on it.
06:07:05 <zzo38> If the 3/3 creature also has last strike, then this creature is destroyed and you gain 1 life, and the 3/3 creature now has one point of damage marked on it and is not destroyed.
06:08:46 <shachaf> Why is this creature destroyed if the 3/3 has last strike, and not if it has normal strike?
06:09:12 <zzo38> O no wait it is not destroyed; I made a mistake.
06:09:35 <zzo38> However then the 3/3 creature is not destroyed either, and you still gain only 1 life.
06:10:01 <zzo38> It does survive and is now 4/4.
06:10:22 <shachaf> Maybe it should have normal strike instead of last strike.
06:11:13 <zzo38> There is no "normal strike" ability; if it doesn't have first strike or double strike or last strike then it assigns damage during the normal strike step.
06:11:42 <zzo38> Without last strike it is a bit less powerful in most cases I think.
06:13:10 <shachaf> That's what I meant by normal strike.
06:13:24 <shachaf> It's less powerful, but it's still good vs. first-strike creatures.
06:14:10 <zzo38> Yes, I know, it is that; I did think of that, but decided to give it last strike.
06:15:23 <shachaf> But last strike isn't in the rules.
06:15:39 <zzo38> I know; I made up that rule!
06:15:54 <zzo38> (And did write how it interacts with first strike and double strike on the same permanent.)
06:16:53 <zzo38> Another idea is a card that says: State-based actions cannot destroy creatures during combat damage steps unless a card is on the stack.
06:35:28 -!- xfix has quit (Ping timeout: 250 seconds).
06:37:44 -!- xfix has joined.
06:51:42 <hppavilion[1]> I wonder if I could bloat the GoL into something terrible and sell it on steam xD
06:56:25 -!- jaboja has joined.
07:04:04 -!- jaboja has quit (Ping timeout: 250 seconds).
07:15:01 -!- hppavilion[1] has quit (Ping timeout: 252 seconds).
07:17:02 -!- jaboja has joined.
07:22:14 -!- augur has quit (Remote host closed the connection).
07:26:50 -!- Cale has quit (Ping timeout: 260 seconds).
07:27:40 -!- jaboja has quit (Remote host closed the connection).
07:31:28 -!- oerjan has quit (Quit: Nite).
07:35:08 <zzo38> Should the blitter include the ability to load a translation table and to use it? I am not sure if it is worth to include such thing, at all.
07:39:53 -!- Cale has joined.
07:40:57 -!- Cale has quit (Read error: Connection reset by peer).
07:41:08 -!- Cale has joined.
07:52:05 <shachaf> zzo38: Is that a triggered or activated ability?
08:05:16 -!- hppavilion[1] has joined.
08:06:55 -!- lambda-11235 has quit (Quit: Bye).
08:29:05 -!- Qund has joined.
08:31:28 -!- hppavilion[1] has quit (Ping timeout: 252 seconds).
08:37:50 -!- hppavilion[1] has joined.
08:41:03 -!- Qund has left.
09:23:22 -!- augur has joined.
09:28:14 -!- augur has quit (Ping timeout: 272 seconds).
10:11:06 <rntz> zzo38: yeah, the timing of that card doesn't work...
10:12:23 <rntz> oh, no, wait, I misread
10:12:47 <rntz> "gets +1/+1 for each point of damage" is a state-based thing, not a triggered ability. ok. that's extremely peculiar.
10:15:03 -!- heroux has quit (Ping timeout: 240 seconds).
10:31:11 -!- heroux has joined.
11:43:34 -!- Phantom_Hoover has joined.
11:44:38 -!- Phantom__Hoover has joined.
11:48:01 -!- Phantom_Hoover has quit (Ping timeout: 240 seconds).
11:52:55 <hppavilion[1]> I just asked someone online who wasn't making much sense if they're fungot. I await their utter confusion.
12:04:50 -!- bender__ has joined.
12:35:59 -!- hwpplayer1 has joined.
12:42:49 -!- hppavilion[1] has quit (Ping timeout: 252 seconds).
12:58:54 -!- ybden has joined.
13:10:23 -!- bender__ has changed nick to bender|.
13:10:35 -!- bender| has quit (Changing host).
13:10:36 -!- bender| has joined.
13:25:03 -!- augur has joined.
13:29:28 -!- augur has quit (Ping timeout: 244 seconds).
13:39:59 -!- hwpplayer1 has quit (Read error: No route to host).
13:40:03 -!- hwpplayer11 has joined.
13:40:54 -!- hwpplayer11 has quit (Remote host closed the connection).
13:46:06 -!- hwpplayer1 has joined.
13:51:13 -!- bender| has changed nick to bender.
13:57:36 -!- beoswind has joined.
13:58:53 -!- beoswind has quit (Client Quit).
15:09:18 -!- gamemanj has joined.
15:12:36 -!- idris-bot has joined.
15:14:42 -!- augur has joined.
15:27:17 -!- ais523 has joined.
15:35:41 -!- augur has quit (Remote host closed the connection).
15:45:32 -!- augur has joined.
15:54:48 -!- hwpplayer1 has quit (Ping timeout: 276 seconds).
15:56:39 -!- shikhin has changed nick to freud.
15:58:41 -!- freud has changed nick to freudi.
15:58:43 -!- Kaynato has joined.
15:59:51 -!- bender has changed nick to sid123.
16:01:01 -!- freudi has changed nick to shikhin.
16:02:21 -!- sid123 has changed nick to RespublikBender.
16:02:22 -!- shikhin has changed nick to freud.
16:03:15 -!- RespublikBender has changed nick to bender.
16:04:37 <HackEgo> [wiki] [[VL]] https://esolangs.org/w/index.php?diff=47077&oldid=47076 * 31.185.252.84 * (+51)
16:06:50 -!- Kaynato has quit (Quit: Leaving).
16:20:47 -!- hwpplayer1 has joined.
16:27:28 -!- Kaynato has joined.
16:27:50 -!- freud has changed nick to shik.
16:28:10 -!- shik has changed nick to shikhin.
16:33:02 -!- gamemanj has quit (Ping timeout: 250 seconds).
16:34:50 -!- lambda-11235 has joined.
16:38:04 -!- hwpplayer1 has quit (Quit: Leaving.).
16:42:48 -!- ais523 has quit.
16:45:15 -!- gamemanj has joined.
16:51:51 -!- Kaynato has quit (Ping timeout: 244 seconds).
16:53:49 <fizzie> fungot: Were you the person who wasn't making much sense?
16:53:49 <fungot> fizzie: human, chieftain, guard, ninja, nurse, page, ronin, shopkeeper, you get promoted to demi-god.
16:54:01 <fizzie> That's a nice career path.
17:11:49 -!- hwpplayer1 has joined.
17:14:39 -!- bender__ has joined.
17:14:41 -!- bender has quit (Ping timeout: 258 seconds).
17:14:46 -!- bender__ has changed nick to bender.
17:23:35 -!- bender has quit (Ping timeout: 260 seconds).
17:27:14 -!- Koen_ has joined.
17:28:07 <\oren\> someone threw a chocolate cake in the german parliament
17:28:33 <\oren\> i decided not to inquire for more details
17:30:02 <tswett> b'ahstunls tt hdoto is ldpossthc stttd,ls cavscts s bs j id tesod ap Wh ieihdthdid I thils tf ar thp '
17:31:50 <tswett> b"Th m nn oonsehron eom pran tr L misrlw phund winr ches in nofs of nhns'hon nun nolpofrim tuos of ndr"
17:34:11 <gamemanj> Presumably they thought it would taste better than the usual comedic thrown foodstuff.
17:39:31 <tswett> b'Th tts toes6thtubrrtesspoothived i mouthumetsisatht, tiatioss whtr tyiniuloretapedtithrics. Itzrensu'
17:43:49 <tswett> It doesn't mean anything.
17:44:13 <gamemanj> ...it has ". ", it has a sentence structure, there's a particular use of capital letters. Are you sure?
17:45:59 <gamemanj> Though I do admit a toes6thtubrrtesspoothived is a bit of a long word.
17:48:48 <tswett> And it has a numeral in it, which real words generally don't.
17:49:18 <tswett> Here, let's have a better one.
17:49:19 <tswett> b'To pooitas apitaie aul Weakiome aioui goatiagionauateon. Chiter tinmeoucioveany oi meo, orined Mouna'
18:01:01 -!- ais523 has joined.
18:02:01 <tswett> b'Thated top el to te a cter of Wis to tonts Aexper. It ted (oved itto te than nxlo tovt cter Norton J'
18:04:39 <tswett> Ooh, it's starting to get some really good words in there.
18:04:40 <tswett> b'Tex in icwuites the toll, ovel whevexpected to the kingdom s Ituresp. the Kizyingerne, haves Mit te '
18:07:37 <tswett> b'The Therexpectuitee Fist.avhas oo decfuiteoacteretooee tollboothetteralomeations an Wistomes full of'
18:07:47 <tswett> I wonder what tollboothetteralomeations are.
18:08:35 <tswett> Output from a neural net trained to output character strings.
18:08:39 <tswett> b"The to the book the to the book is a lo e chev noppeon the 'ook was boothoed iat, Tany eventorestes "
18:08:52 <tswett> The training data is the first paragraph of the Wikipedia article about the book "The Phantom Tollbooth".
18:10:28 <tswett> Ooh, I think this is the first time it successfully reproduced the first couple of words:
18:10:29 <tswett> b'The Phantom Tollbooth one afternoon and, hovhan attoec Mted) of Oeastrens Thed Moto ctor now pons Al'
18:10:32 <gamemanj> once upon a time someone was going down a road, which had no tollbooth... then a tollbooth appeared and charged them money
18:13:40 <ais523> tswett: that's not a lot of training data
18:13:52 <tswett> Indeed, that's an extremely small amount of training data.
18:14:08 <tswett> Though, the first training data I ever successfully used was:
18:14:26 <tswett> b'The Phere is ted into many The Kell tr the konderful Wirl, it tells The Woo sel, it ral to the kio s'
18:14:35 <tswett> It's like... it's like it's trying to say something.
18:15:36 <tswett> b"Thetoovres stortofo to Concl to m. Itur's into Thar,s rovhoust to thansports him toy Poop and has so"
18:15:43 <tswett> It's not very nice to thansport someone toy Poop.
18:16:07 <tswett> b'The Will and gas wold onte the o dd Fdlatin ixs and gextextextextextextexpected inte aptor nov itve'
18:16:36 <tswett> That's not the word I extextextextextextexpected it to use.
18:19:31 -!- augur has quit (Remote host closed the connection).
18:20:40 <tswett> Ooh, it managed to reproduce a pretty long section verbatim.
18:20:41 <tswett> b'The Phe text is full of puns and wordplay; many events, such as when Milo unintenn wrer whine aftern'
18:21:15 <tswett> Most of that is a verbatim quote: "[T]he text is full of puns and wordplay; many events, such as when Milo uninten[tionally ...]"
18:22:43 <tswett> Here it almost reproduced the first 100 bytes of the whole thing:
18:22:44 <tswett> b"The Phantom Tollbooth is a 1961 children's adventure novel by American writer Norton Juster Nills an"
18:23:10 <tswett> Correct quote: "The Phantom Tollbooth is a 1961 children's adventure novel by American writer Norton Juster [with illustrations ...]"
18:25:43 -!- Koen_ has quit (Remote host closed the connection).
18:38:51 <HackEgo> [wiki] [[Evil]] https://esolangs.org/w/index.php?diff=47078&oldid=46809 * 90.255.15.157 * (+1) Fixed spelling mistake - 'wave function' -> 'weave function'
18:55:34 <fizzie> If your goal is to learn English in general, I fear you might be overfitting a bit.
18:56:12 <fizzie> http://karpathy.github.io/2015/05/21/rnn-effectiveness/ did a bit of Wikipedia markup.
19:13:38 -!- lambda-11235 has quit (Quit: Bye).
19:17:42 -!- xkapastel has joined.
19:26:22 <tswett> fizzie: but if my goal is to learn this exact paragraph, I'm not overfitting at all!
19:28:53 <tswett> I think it's fair to say that the neural net has successfully memorized the entire paragraph.
19:29:03 <tswett> Which is slightly impressive.
19:30:11 -!- Opodeldoc has joined.
19:31:03 <tswett> Lemme see. The paragraph is 1114 bytes, 51 distinct. This means that, naively encoded, the number of bits of information there is...
19:31:09 <tswett> > 1114 * log 51 / log 2
19:33:38 <fizzie> And how many parameters are there in your model?
19:34:17 <tswett> The neural net has an input layer with effectively 51 neurons (no parameters); a hidden layer with 50 neurons (effectively (51+1)*50 parameters); another hidden layer with 50 neurons ((50+1)*50 parameters); and finally an output layer with effectively 51 neurons (effectively (50+1)*51 parameters).
19:34:30 <tswett> Actually, I'm not sure if the output layer has biases.
19:34:48 <tswett> So the number of effective parameters is...
19:35:01 <tswett> > (51+1)*50 + (50+1)*50 + (50+1)*51
19:35:09 <shachaf> Is there a word other than "neuron" to describe the sizes of the vectors?
19:35:30 <shachaf> n-dimensional hidden layer?
19:36:11 <gamemanj> and now for your complex term of the day: n-dimensional mathematical numpy-supported matrices supported on a large hardware platform.
19:36:12 <shachaf> Well, I'm not a fan of the word "hidden" either, but at least it's not forcing a biological analogy.
19:37:06 <tswett> It is, in fact, accurate to say that these are "n-dimensional mathematical numpy-supported matrices".
19:37:27 <tswett> And they are, in a sense, "supported on a large hardware platform", because this is running on my laptop, and my laptop is physically large.
19:38:20 <shachaf> I asked the other day: How much justification is there to Tensorflow et al.'s use of the word "tensor"?
19:38:46 <shachaf> Are these things tensors in the usual sense, with covariance/contravariance, multilinear maps, etc.?
19:40:00 <fizzie> shachaf: RNN elements are conventionally called "cells".
19:41:07 <tswett> shachaf: they can be used that way.
19:41:21 <HackEgo> [wiki] [[Emmental]] https://esolangs.org/w/index.php?diff=47079&oldid=35639 * 96.48.108.186 * (+231) /* Hello, World! */ Added easier to understand example
19:41:30 <tswett> A "tensor" in Tensorflow is the tensor product of some copies of R^n, for varying n.
19:42:04 <gamemanj> tswett: I like to be correct while also being stupidly verbose
19:42:16 <HackEgo> [wiki] [[Joke language list]] https://esolangs.org/w/index.php?diff=47080&oldid=47071 * Ais523 * (-13) Undo revision 47051 by [[Special:Contributions/Moon|Moon]] ([[User talk:Moon|talk]]); as mentioned in the intro, redlinks are removed from this list if they stay red too long
19:42:28 <tswett> In other words, it's a multidimensional array.
19:42:43 <shachaf> I don't think a tensor is just a multidimensional array.
19:45:05 <fizzie> Also, if the "et al." is large enough, there's no need for justification.
19:48:49 <fizzie> If you don't want to use the word "neuron", it's possible to just say "the size of the layer is X".
19:48:57 <fizzie> I saw TensorBoard for the first time the other day.
19:50:12 <shachaf> fizzie: I don't think it's too large.
19:51:22 <shachaf> And if all of these libraries used "vector" to mean "scalar", I would still want justification, because their usage is in the context of the rest of the world.
19:52:16 <fizzie> Yes, well, I was including rest of the world in the et al., possibly.
19:52:18 <tswett> A tensor is certainly not just a multidimensional array.
19:57:07 <tswett> > map (\n -> 2 * 10 ** (1/n)) [0..6]
19:57:09 <lambdabot> [Infinity,20.0,6.324555320336759,4.308869380063768,3.5565588200778455,3.1697...
19:57:26 <tswett> > map (\n -> 2 * 10 ** (n/6)) [0..6]
19:57:28 <lambdabot> [2.0,2.935598535244139,4.308869380063768,6.324555320336759,9.283177667225557...
19:58:04 <tswett> > map (\n -> round $ 100 * 10 ** (n/6)) [0..6]
19:59:25 <tswett> > map (\n -> round $ 96 * 10 ** (n/6)) [0..6]
19:59:45 <tswett> > map (\n -> round $ 98 * 10 ** (n/6)) [0..6]
20:02:29 <tswett> > let blah = [10.0, 15, 20, 30, 50, 70, 100, 150, 200, 300, 500, 700, 1000] in zipWith (/) (tail blah) blah
20:02:31 <lambdabot> [1.5,1.3333333333333333,1.5,1.6666666666666667,1.4,1.4285714285714286,1.5,1....
20:06:17 -!- puckipedia has quit (Ping timeout: 244 seconds).
20:07:02 -!- ybden_ has joined.
20:07:14 -!- ybden_ has quit (Client Quit).
20:11:36 -!- puckipedia has joined.
20:36:40 -!- shikhin has quit (Quit: Alas.).
20:36:40 -!- hydraz has quit (Quit: Bai.).
20:37:34 -!- shikhin has joined.
20:37:54 -!- hydraz has joined.
20:37:54 -!- hydraz has quit (Changing host).
20:37:54 -!- hydraz has joined.
20:41:57 <zzo38> I do not yet know what my new kind of computer design is called.
20:42:33 <zzo38> To know what it is called can help with discussion elsewhere, with naming programs that do emulation or hardware implementation of parts of it, etc.
20:42:48 <ais523> do you think it already has a name and you don't know what it is? or is it something that doesn't have a naem yet and you haven't named it?
20:43:34 <ais523> gamemanj: I didn't even realise that wasn't part of this conversation :-P
20:43:56 <int-e> fungot: kitten kitten kitten?
20:43:56 <fungot> int-e: they say that if you start at the man fled, clutching his bulging guts, finally to fall from his throat; the drunk vomited lumps of human flesh. ( the fellowship of the english puritans regarded all fairies as devils. the couatl are very sharp; watch you don't usually mind if you need a mirror to notice a mimic in an ideal existence, drinking and eating without restraint and fighting over again the battles in which he ca
20:44:04 <fungot> Available: agora alice c64 ct darwin discworld enron europarl ff7 fisher fungot homestuck ic irc iwcs jargon lovecraft nethack* oots pa qwantz sms speeches ss wp youtube
20:44:24 <zzo38> ais523: I think that it does not yet have a proper name.
20:44:39 <zzo38> That is why I obviously do not know what the name is!
20:44:44 <zzo38> Because there isn't one!
20:49:06 <int-e> the free kitten refuses to be bound to any particular channel
20:49:25 <gamemanj> yes, let's worship the great kitten
20:56:45 <zzo38> I had some ideas about the main processor's microcode which are a bit similar to some of the features of VAX microcode, such as huge VLIW, every instruction contains a jump (actually two in the design I have thought of), and XFC.
21:00:12 <zzo38> The microcode would include ROM and RAM areas, and you can define your own instructions in the RAM area.
21:04:33 <zzo38> It is a kind of horizontal microcode format, and some of the pieces of a microcode instruction might be: register numbers for the Muxcomp inputs/outputs, ALU code, external memory I/O code, condition code, jump if true, jump if false.
21:05:59 <zzo38> If it is muxcomp32 then seven register must be specify for muxcomp, and some registers might be immediate values, constants, and external memory access.
21:10:46 -!- Moon_ has joined.
21:15:22 -!- Moon_ has quit (Ping timeout: 250 seconds).
21:17:42 -!- Moon_ has joined.
21:18:23 <Moon_> > zip [1,4..] [1,6..]
21:18:25 <lambdabot> [(1,1),(4,6),(7,11),(10,16),(13,21),(16,26),(19,31),(22,36),(25,41),(28,46),...
21:19:16 <Moon_> `echo > zip [1,4..] [1,6..]
21:19:24 <Moon_> `echo '> zip [1,4..] [1,6..]'
21:21:08 <Moon_> > zip "Im being ziped!" "!depiz gnieb mI"
21:21:10 <lambdabot> [('I','!'),('m','d'),(' ','e'),('b','p'),('e','i'),('i','z'),('n',' '),('g',...
21:22:24 <shachaf> please use /msg unless you're showing something to others
21:30:26 -!- augur has joined.
21:33:06 -!- Kaynato has joined.
21:36:15 -!- jaboja has joined.
21:36:26 <HackEgo> /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: exec: apt: not found
21:36:57 <HackEgo> [1;24r[0;10m[4l[?7h[39;49m[?1000h[?25l[?1c[39;49m[0;10m[H[J[24d[0;10;1m[37m[41m[J[H[37m[44m Actions Undo Package Resolver Search Options Views Help[K
21:39:15 <Moon_> i wonder, some linux distros have the 'lolcat' command, does hackego have it?
21:39:21 <HackEgo> Today is Pungenday, the 2nd day of Confusion in the YOLD 3182
21:39:22 <Moon_> lets try that first
21:39:40 <HackEgo> usage: ddate [+format] [day month year]
21:39:50 <HackEgo> /hackenv/bin/`: line 4: lolcat: command not found
21:40:23 <gamemanj> wait a second... *opens up Twister*
21:40:28 <myname> you should learn about it
21:40:33 <myname> discordianism,is great
21:40:53 <gamemanj> attempting to identify @tasty...
21:41:33 <Moon_> lolcat rainbow colors its input, myname
21:41:56 <Moon_> ah well, time to make a irc version *jokes*
21:42:03 <HackEgo> Today is Pungenday, the 2nd day of Confusion in the YOLD 3182
21:42:20 <fizzie> Well, it's a bit fancier.
21:42:29 <Moon_> yea, lolcat is much fancier
21:42:36 <HackEgo> Moon_: Welcome to the international hub for esoteric programming language design and deployment! For more information, check out our wiki: <http://esolangs.org/>. (For the other kind of esoterica, try #esoteric on EFnet or DALnet.)
21:42:47 <Moon_> im already a known member -.-
21:42:49 <fizzie> Uses the 256-color control codes and all that.
21:43:13 <myname> why do you do that stuff, then?
21:43:30 <Moon_> because im bored :P
21:43:39 <Moon_> but i am a known member
21:44:29 <fizzie> I don't think "member" is the right word for anyone here.
21:45:47 <zzo38> I have written a Discordian calendar program on ifMUD too
21:49:54 <fizzie> Whether ddate should or shouldn't be part of util-linux on Debian was a long, long debate. From what I recall, it got removed at least once, and maybe reinstated or split to a separate package.
21:49:57 <zzo38> I think there is actually two versions of Discordian calendar, which are the common version (which is synchronized with the Gregorian calendar), and the uncommon version (which is synchronized with the Julian calendar); however there is currently no difference between them, I think they would start to differ in 2100 maybe.
21:53:55 <fizzie> Is that just because Principia Discordia says "every four years"?
21:56:08 <nortti> but seeing as it is otherwise synchronised one-to-one with the gregorian calendar, discordian-with-gregorian-leps
22:01:22 -!- sebbu has quit (Ping timeout: 260 seconds).
22:03:23 <zzo38> I think ddate probably should not be the part of the core package but probably should be another package.
22:04:49 -!- sebbu has joined.
22:06:46 <ais523> zzo38: it isn't part of the Ubuntu package it used to be in (which contained fsck and similar programs, IIRC) any more
22:07:03 <ais523> gamemanj: it was in a package of important Linux utilities, many of which most systems will use
22:07:08 <ais523> I think fsck was the most famous
22:07:10 <gamemanj> bash: ddate: command not found
22:07:13 <ais523> not sure, there might be an even more famous one
22:09:07 <gamemanj> it apparently used to be in util-linux,
22:09:11 <gamemanj> but isn't built anymore by default.
22:13:44 -!- Koen_ has joined.
22:17:08 -!- Koen_ has quit (Client Quit).
22:21:10 -!- Frooxius has quit (Quit: *bubbles away*).
22:21:11 <fizzie> That's what I was referring to.
22:23:25 <fizzie> 1998: Bug#30015: util-linux: please include ddate
22:23:36 <fizzie> 2010: Bug#583503: ddate: this useless joke program has no place in a base debian install.
22:23:50 <fizzie> 2012: Bug#583503: ddate, upstream, and debian.
22:23:56 <fizzie> 2014: Bug#583503: ddate no longer shipped in upstream release v2.24.2.
22:24:14 <gamemanj> a quick look at wikipedia reveals something interesting...
22:24:54 <fizzie> Those weren't even all the titles.
22:25:00 <gamemanj> ddate: removed from default build by the same people who brought you: "tunelp: parallel port printers are probably almost extinct devices"
22:26:08 <fizzie> I have a vague recollection they must've flip-flopped on its inclusion at least once, before it ended up in a separate 'ddate' package.
22:27:16 <gamemanj> I'm a tad worried that a package for various utilities is being committed to by someone who intends to strip down the commands... isn't that the point of busybox?
22:28:07 <fizzie> Well, the reasonable argument was that it's not a package of "various utilities", it's a package of things you absolutely need to have available.
22:29:17 <zzo38> I think ddate should be a separate package, such as "comprehensive date/time package" which may include that and others
22:29:39 <gamemanj> ...such as "wall" (good to have on mainframes maybe), "script" (eh could be useful) , "ul" (ok who needs this)
22:29:40 <zzo38> Alternatively, a package by itself, is another possibility
22:30:29 <gamemanj> all 3 of those, wall, script and ul have man pages on my system, and ul especially seems particularly useless. If the "things that are required to be available" is the point, then it would be a lot smaller
22:30:48 <fizzie> ul is not in util-linux on my system.
22:30:49 -!- chron_ has joined.
22:31:02 <fizzie> Neither is 'wall' or 'script'.
22:31:02 <int-e> script can be useful for documentation purposes
22:31:17 <zzo38> The program "wall" could be useful on multi-user systems, although for single-user system does not seem so useful.
22:31:26 <fizzie> script's part of the Debian upgrade process documentation.
22:31:41 <zzo38> (But still, someone might use wall for some purpose even in a single-user system)
22:31:41 -!- ais523 has quit (Read error: Connection reset by peer).
22:32:07 <gamemanj> Ok, fair enough, but tunelp presumably has a similar basis.
22:32:10 <int-e> oh yeah, good old wall/write/talk/mesg
22:32:22 <gamemanj> The point is: where to draw the line?
22:32:28 <zzo38> Yes, on multi-user systems, those programs can be useful.
22:32:32 <fizzie> gamemanj: Somewhere south of ddate, probably.
22:32:33 -!- ais523 has joined.
22:32:33 -!- ais523 has quit (Read error: Connection reset by peer).
22:32:47 <int-e> gamemanj: in the case of ul, right under the text...
22:32:51 <tswett> Now I'm training my neural net on #esoteric logs again.
22:33:03 <zzo38> A separate package for programs that are useful for multi-user could be added if someone wants this, I suppose.
22:33:05 <gamemanj> Well, I guess it doesn't make any sense to include ddate but not date.
22:33:26 <tswett> Here's a quote from it:
22:33:27 <tswett> hd cu#8rin84cde3!i aNenei r s
22:33:28 <zzo38> I think date is supposed to be core
22:33:33 <int-e> I've never used ddate... I didn't even know that it exists.
22:33:45 <fizzie> I used to have a ddate login script, I think.
22:33:47 <int-e> so... "objectively" the latter is not needed.
22:33:47 -!- ais523 has joined.
22:33:48 -!- ais523 has quit (Read error: Connection reset by peer).
22:33:49 <fizzie> It's been a while, though.
22:34:16 <int-e> it's hard to draw hard lines
22:34:32 <fizzie> int-e: Yes, I think Debian should switch to a probabilistic package manager.
22:34:35 <gamemanj> To me, it seems that ddate is just another date format and makes more sense as a part of date.
22:34:46 <int-e> but I think date and ddate are rahter obvious choices ;)
22:34:58 <int-e> fizzie: you scare me
22:35:00 -!- ais523 has joined.
22:35:27 <gamemanj> As for date, it's not even part of util-linux (it's part of coreutils), so ddate, being another date format, should be there.
22:35:38 * int-e learned last week that there are several flavors of netcat.
22:35:45 <zzo38> I think it is rather a different calendar rather than just a different output format
22:36:24 <int-e> so that's another dimensions: if there are several variants of a tool, which one will you include?
22:36:38 <fizzie> The netcat variants are a big annoyance.
22:36:44 -!- ais523 has quit (Read error: Connection reset by peer).
22:36:48 <int-e> hmm, dates... I should buy some dried dates...
22:36:51 <zzo38> Are you going to include Mayan calendar too, and also Muslim, and whatever others?
22:36:57 <myname> the bsd one, the gnu one, the others
22:36:58 <fizzie> At least there's only one socat. (Which hasn't been enough of a draw to learn the socat syntax.)
22:37:51 <int-e> gamemanj: https://en.wikipedia.org/wiki/Date_%28fruit%29
22:37:53 -!- ais523 has joined.
22:38:09 <myname> for a moment i thought socat is some useless thing to interact with stackoverflow
22:38:24 <tswett> Dates are dates? That's not true at all.
22:38:29 * int-e should look into socat too
22:38:44 <tswett> "Datetime" means something different in every programming language.
22:38:52 <fizzie> The thing that spews out those wiki updates is almost all socat.
22:39:20 -!- ais523 has quit (Read error: Connection reset by peer).
22:39:21 <int-e> Oh is "socat" a pun? ... meh.
22:39:35 <int-e> ess-oh-cat... socket
22:40:14 <fizzie> "SOcket CAT", it just says.
22:42:17 <int-e> fizzie: I guess my question is how one is supposed to pronounce it.
22:42:33 <fizzie> I don't rightly know, I think I've always just written it.
22:42:38 <fizzie> Bleh. rdiff-backup ran out of memory again. For some reason it just does that; then the system starts swapping and the backup takes two days or something ridiculous.
22:46:19 <fizzie> It's a relatively extensive set of files (3582083 used inodes on the filesystem, and it's a large chunck of that), but from what I gather, it's still supposed to only take a reasonable amount of memory, and seems to do just that for most people. Someone else was complaining of excessive memory use as well, but I don't think that thread ever resolved.
22:48:10 <fizzie> "If rdiff-backup seems to be leaking memory, it is probably because it is using an early version of librsync. librsync 0.9.5 leaks lots of memory. Later versions should not leak and are available from the librsync homepage."
22:48:44 <fizzie> But this is 0.9.7, and shouldn't have such problems, so who knows.
22:49:00 <fizzie> "The amount of memory rdiff-backup uses should not depend much on the size of directories being processed. Keeping track of hard links may use up memory, so if you have, say, hundreds of thousands of files hard linked together, rdiff-backup may need tens of MB."
22:49:39 <fizzie> Shouldn't be many of those, and it's at three hundred megs (out of ~400-ish or so).
22:50:03 <tswett> 0.:32:26: -!- iacht5 ion1nhndonul7gviin ao1ar ridl
22:51:00 <fizzie> Hmm. It's got a built-in _librsync.so.
22:51:09 <fizzie> (rdiff-backup itself is mostly Python.)
22:53:20 <fizzie> I think that must be just some Python glue to call into the real one, because the package is listed as depending on librsync1.
22:53:52 <int-e> ldd .../_librsync.so ?
22:55:34 <fizzie> So it shouldn't be due to that.
22:56:40 <fizzie> Don't suppose there's any way to ask a running Python process for a memory use summary or something.
22:57:51 <tswett> Ooh, the neural net has learned a nick... kinda.
22:57:52 <tswett> c2 cl:12:22: <his52 > am1yms c a,gi t7on an2sn rucarc n niraao h gota> hom c l'itnannor3a
22:58:52 <fizzie> Manages to match the tone of the channel pretty well, too.
22:59:20 <gamemanj> I thought only I used C2/CL time!
23:02:33 -!- Kaynato has quit (Ping timeout: 240 seconds).
23:10:25 <fizzie> http://www.metoffice.gov.uk/ looks like tomorrow's weather is very XMLy.
23:11:31 <fizzie> http://sprunge.us/hiaT is how it appears to me.
23:11:52 <gamemanj> You're looking at an RSS feed, somehow.
23:12:03 <fizzie> Well, that's what I get as the front page.
23:12:03 <int-e> look at the first line: 2000<?xml version="1.0" encoding="UTF-8"?>
23:12:36 <gamemanj> fizzie: Which entry do you want to see?
23:12:41 <fizzie> http://sprunge.us/LfUa
23:12:56 <fizzie> I had my normal forecast link in the browser history, that worked just fine.
23:13:02 <fizzie> It's only the front page that appears strange.
23:13:18 <int-e> so akamai messed up?
23:13:56 <Phantom__Hoover> here, what do you call a program that unpacks itself into more programs
23:13:59 <int-e> www.metoffice.gov.uk. 842 IN CNAME www.metoffice.gov.uk.edgesuite.net.
23:14:10 <int-e> (and it looks fine from here)
23:14:19 <tswett> 20:44:18: -E- rodgort has qoin jokoet nesoq)af quine(k ne uorirdaie8:one natoEam9pat
23:14:41 <fizzie> 2.21.75.40, 2.21.75.10 what it eventually resolves as to me.
23:14:42 <int-e> gamemanj: you do know what a CDN is, don't you?
23:14:50 <fizzie> How about curl -s -H 'Host: www.metoffice.gov.uk' http://2.21.75.40/ | head =
23:15:19 <fizzie> (www.metoffice.gov.uk.edgesuite.net is an alias for a376.r.akamai.net, a376.r.akamai.net has address 2.21.75.40 / 2.21.75.10)
23:15:24 <int-e> fizzie: yeah that gives me the paste
23:15:33 <gamemanj> a CDN is a bunch of servers people pay lots of money to use, which have a lot of bandwidth available.
23:15:37 <int-e> including the initial 2000, which I still find rather odd
23:15:57 <gamemanj> They are generally meant for delivering vast amounts of static resources.
23:16:08 <fizzie> int-e: There's a bunch of numbers later on as well.
23:16:11 <int-e> gamemanj: unless you're somebody like Google or Amazon who's running their own CDN.
23:16:26 <fizzie> int-e: Right in the middle of the content. I suspect some sort of a length-delimited format.
23:17:10 <fizzie> Or possibly debugging output or something, who knows.
23:19:22 <int-e> so http://www.metoffice.gov.uk/datapoint/support/notifications/rss is the original link for that... no numbers though
23:20:32 <zzo38> I now made the JavaScript implementation of Internet Quiz Engine so that if you write #<INQUIZJS>BEGIN and #<INQUIZJS>SHUFFLE around some lines then those lines will be shuffled in the output. An implementation that doesn't support this will treat these lines as comments and display the text without shuffling; the #< indicates extensions which are safe to ignore.
23:20:37 <fizzie> int-e: The next "\r\n2000\r\n" is exactly 0x2000 bytes after the first one.
23:22:33 <fizzie> Actually, isn't that just https://en.wikipedia.org/wiki/Chunked_transfer_encoding
23:22:55 <fizzie> Except missing the \r\n after the initial chunk length indicator.
23:23:41 <fizzie> (No Transfer-Encoding header either.)
23:23:48 <pikhq> Yes, that's basically chunked transfer encoding getting botched.
23:24:07 -!- Frooxius has joined.
23:24:54 -!- chron_ has left ("Leaving").
23:29:56 -!- hppavilion[1] has joined.
23:37:53 <fizzie> Found something called "gdb-heap" for unplanned memory debugging -- https://fedorahosted.org/gdb-heap/ -- but I'm worried the memory debugging tool will run out of memory.
23:39:01 <fizzie> (It prints "Blocks retrieved 150000" etc. after every 10k blocks, and swaps a whole lot.)
23:39:26 -!- hwpplayer1 has left.
23:42:04 <fizzie> Up to 380000. That's a lotta allocations.
23:42:40 <gamemanj> why can't people just use static allocations...
23:44:14 <Phantom__Hoover> gamemanj, because sometimes you don't know how much memory you'll need at compile time??
23:44:37 <fizzie> Yeah, the gdb process is now using almost twice as much memory as the rdiff-backup.
23:44:52 <fizzie> It's also at 590000 blocks.
23:45:05 <pikhq> It's not unusual for static allocation to be insufficient.
23:45:27 <pikhq> That said, I also think a lot of people end up using dynamic allocation when static allocation would suffice.
23:45:34 <pikhq> Or overusing dynamic allocation.
23:51:19 <int-e> oh wow, programs compiled with ghc-8.0.1 have a virtual memory size of 1 TB on startup (x86_64 linux)... even when most of that memory is unused, that seems overly generous.
23:51:51 <tswett> All right, here's the first completely well-formed thing produced by my new neural net:
23:51:52 <tswett> 09:01:26: <FiCe> his fhrtunn pikGr
23:52:38 <fizzie> http://comments.gmane.org/gmane.comp.sysutils.backup.rdiff-backup.general/7878 -- I assume I'm hitting the same thing, but nobody's answered. :/
23:54:49 <fizzie> "FWIW, I used to use rdiff-backup but found it to be nonrobust on machines with limited (only a few GB) RAM and hundreds of GB of backup."
23:54:52 <fizzie> Should probably switch tools.