Logs: liberachat/#haskell
| 2021-07-29 17:40:17 | × | curiousgay quits (~curiousga@77-120-186-48.kha.volia.net) (Ping timeout: 252 seconds) |
| 2021-07-29 17:40:20 | <motle> | just for streaming i think, but they can output back into themselves, so i think thats how it does recursion too |
| 2021-07-29 17:41:04 | <motle> | eg you could just provide initial data and have it unfold data streamingly |
| 2021-07-29 17:41:47 | <motle> | or you could also stream data into it while it runs, in sync with the clock of the internal programs evaluation scheduling as specified on the graph as a bunch of simultanious opperations per cycle |
| 2021-07-29 17:42:42 | <motle> | i mean, its all just streaming through graphs for recursive programs |
| 2021-07-29 17:43:03 | <motle> | (with hopfiled nets as a type restriction!) |
| 2021-07-29 17:43:58 | × | burnsidesLlama quits (~burnsides@dhcp168-019.wadham.ox.ac.uk) (Remote host closed the connection) |
| 2021-07-29 17:44:15 | <motle> | not sure if its wrong (or realistic) to ask for help, but anyway... thats what it is |
| 2021-07-29 17:45:12 | <lechner> | Clint: thanks! i was already looking into Data.ByteString.Lazy.UTF8.fromString but your solution is much better! |
| 2021-07-29 17:45:22 | → | Atum_ joins (~IRC@user/atum/x-2392232) |
| 2021-07-29 17:47:05 | → | eggplantade joins (~Eggplanta@108-201-191-115.lightspeed.sntcca.sbcglobal.net) |
| 2021-07-29 17:47:54 | <motle> | sorry i was confusing the state which does the unfold to the states on the nodes of the graph scanning over it |
| 2021-07-29 17:48:50 | <motle> | thats why theres the clock function which just streams a list of (), to have a small shaped state fo the unfolding state and the rest of the unfold as a graph program scanning over the stream of () |
| 2021-07-29 17:49:16 | → | roboguy_ joins (~roboguy_@2605:a601:afe7:9f00:1557:b33d:ca8c:dfa4) |
| 2021-07-29 17:49:28 | <Drew[m]> | <motle> "not sure if its wrong (or..." <- I feel like I wouldn't expect many people to be able to follow your thoughts to know how to help. Clearly you've been thinking about it a lot but I think you'd need to condense it down to a more concrete question or series of questions. |
| 2021-07-29 17:49:48 | <motle> | i mean n terms of getting something implemented |
| 2021-07-29 17:50:15 | <Drew[m]> | What do |
| 2021-07-29 17:50:17 | <Drew[m]> | oops |
| 2021-07-29 17:50:23 | <Drew[m]> | What do you want to implement? |
| 2021-07-29 17:50:34 | × | fossdd quits (~fossdd@sourcehut/user/fossdd) (Ping timeout: 240 seconds) |
| 2021-07-29 17:50:58 | × | azeem quits (~azeem@dynamic-adsl-94-34-48-122.clienti.tiscali.it) (Ping timeout: 240 seconds) |
| 2021-07-29 17:51:15 | × | lavaman quits (~lavaman@98.38.249.169) (Remote host closed the connection) |
| 2021-07-29 17:51:22 | <motle> | i guess if i could implement a square grid as a cycle for the data in a covnet and express the net as a fold through layers that are graphs... something like that, it uses the convolutions to do both message parsing and data processing then |
| 2021-07-29 17:51:42 | → | fossdd joins (~fossdd@sourcehut/user/fossdd) |
| 2021-07-29 17:51:59 | <motle> | but that the same structure (the aim of the implementation) be used for both, the graph with the stencil api |
| 2021-07-29 17:52:29 | <motle> | the hetrogenaity and polymorphicity are difficult |
| 2021-07-29 17:53:08 | <motle> | ends up with type level programs specifying the shape and for the indexing etc |
| 2021-07-29 17:53:29 | → | azeem joins (~azeem@176.201.11.200) |
| 2021-07-29 17:53:45 | <zzz> | do imported but not used modules contribute to binary size? and i can ask the same for functions inside and outside those modules |
| 2021-07-29 17:53:52 | <motle> | i guess the strange thing about passing types over the edges is thinking that they could modify the types at the nodes |
| 2021-07-29 17:54:21 | <lechner> | Hi, why does Haskell always assume that with f g x i mean f(g, x) instead of f(g( x )) ? |
| 2021-07-29 17:54:24 | <zzz> | s/those/any |
| 2021-07-29 17:54:28 | <motle> | ie the types of the functions held in the program, a wierd result of the thing being modifiable i guess |
| 2021-07-29 17:54:34 | → | wroathe joins (~wroathe@c-68-54-25-135.hsd1.mn.comcast.net) |
| 2021-07-29 17:54:49 | <motle> | really wants to be used for genetic programming though |
| 2021-07-29 17:55:01 | <zzz> | lechner: f g x === (f g) x |
| 2021-07-29 17:55:17 | <motle> | have a pottential project there with elucidating slightly less common higher order programing patters |
| 2021-07-29 17:55:25 | <zzz> | prefix function notation is left associative by default |
| 2021-07-29 17:55:55 | <motle> | just cool to have a program that can be modifying itself while it runs i guess |
| 2021-07-29 17:56:03 | <lechner> | zzz: somehow that goes against my RPN pbrain |
| 2021-07-29 17:56:06 | <motle> | not sure how to handle runtime typechecking though... |
| 2021-07-29 17:56:35 | <zzz> | lechner: use `f $ g $ x` then |
| 2021-07-29 17:56:50 | <motle> | which would correspond to establishing or removing edges i guess, or changing types of arguments, such as the dimension of subgraph segment they were passing for insertion somewhere or something like that |
| 2021-07-29 17:57:20 | <lechner> | zzz: maybe the other way woudn't work. it just seems i need a lot of brackets |
| 2021-07-29 17:57:26 | <lechner> | or $ |
| 2021-07-29 17:57:27 | <roboguy_> | lechner: wouldn't RPN be x f g? |
| 2021-07-29 17:58:04 | <lechner> | roboguy_: you are right |
| 2021-07-29 17:58:17 | <lechner> | not sure what i am saying |
| 2021-07-29 17:58:20 | <zzz> | :t (&) |
| 2021-07-29 17:58:22 | <lambdabot> | a -> (a -> b) -> b |
| 2021-07-29 17:58:33 | <zzz> | & = flip ($) |
| 2021-07-29 17:59:40 | <motle> | i guess its more difficult than a regular machine learning problem. you cant just compare its outputing a stream of functions - you would want to "regularize" it directly, which would be complicated programs acting on the various types in the graph program, to try to get it to produce eg higher order functions in some order |
| 2021-07-29 17:59:59 | <lechner> | it just seems there are more functions with just one argument |
| 2021-07-29 18:00:03 | <roboguy_> | lechner: I don't think eta-reduction would work normally if "f g x" meant "f (g x)" |
| 2021-07-29 18:00:12 | <lechner> | i ynow |
| 2021-07-29 18:00:15 | <lechner> | know |
| 2021-07-29 18:00:29 | <lechner> | what is eta please? |
| 2021-07-29 18:00:37 | <motle> | like its not exactly straitforwards how to encourage a program to assemble itself in such a way |
| 2021-07-29 18:01:14 | <roboguy_> | lechner: for example, eta-reduction allows you to rewrite "f x = g 2 x" as "f = g 2" |
| 2021-07-29 18:01:56 | <roboguy_> | eta-expansion is the other direction, which is also fine |
| 2021-07-29 18:02:08 | <motle> | i guess if it had some really badass way of examining the functions it was supposed to be reproducing, it could kind of make a way to produce them properly |
| 2021-07-29 18:02:10 | <roboguy_> | (if the function takes enough arguments) |
| 2021-07-29 18:02:25 | <lechner> | motle: don't lose those thoughts; write them into a word processor instead |
| 2021-07-29 18:02:42 | <motle> | that wont make them real like you! |
| 2021-07-29 18:02:58 | <lechner> | paper is real |
| 2021-07-29 18:02:58 | × | azeem quits (~azeem@176.201.11.200) (Read error: Connection reset by peer) |
| 2021-07-29 18:03:12 | <motle> | why you want to limit the audience so? |
| 2021-07-29 18:03:23 | <motle> | i mean really impactful i guess |
| 2021-07-29 18:03:28 | → | merijn joins (~merijn@83-160-49-249.ip.xs4all.nl) |
| 2021-07-29 18:03:31 | <motle> | more real |
| 2021-07-29 18:03:44 | <motle> | no pressure |
| 2021-07-29 18:04:02 | <lechner> | on the contrary, i want to make sure your thoughts (ultimately) get the attention they deserve |
| 2021-07-29 18:04:12 | → | azeem joins (~azeem@dynamic-adsl-94-34-48-122.clienti.tiscali.it) |
| 2021-07-29 18:04:34 | <motle> | anyway dont we have tunes still here? |
| 2021-07-29 18:05:17 | <motle> | lechner: well i heard about the machine learning problem in my first hear classes so i guess similar projects have *some* attention? |
| 2021-07-29 18:05:26 | → | Obo joins (~roberto@70.pool90-171-81.dynamic.orange.es) |
| 2021-07-29 18:05:31 | <motle> | probably makes more sense now though... |
| 2021-07-29 18:05:44 | <motle> | not sure how much anyone exept me ever heard of those classes anyway |
| 2021-07-29 18:06:18 | <motle> | anyway, i feel like saying something |
| 2021-07-29 18:06:31 | <motle> | the point is im not sure i can write it! |
| 2021-07-29 18:06:42 | <motle> | this is the barrier, for the asking for help |
| 2021-07-29 18:06:45 | <lechner> | roboguy_: yeah i knew something would be off. maybe applying functions partially also breaks |
| 2021-07-29 18:07:06 | <lechner> | roboguy_: don't listen to me. i am new |
| 2021-07-29 18:07:15 | <motle> | anyway, machine learning problem asside, this graph at the language seems resounding |
| 2021-07-29 18:07:37 | <roboguy_> | lechner: no worries! We were all new at some point! |
| 2021-07-29 18:07:37 | <motle> | like enough to not just be a total waste of time for anyone to bother with |
| 2021-07-29 18:08:00 | <roboguy_> | motle: are you saying that you want to parse something using a neural network? |
| 2021-07-29 18:08:17 | <motle> | thats more of a use case for sake of example |
| 2021-07-29 18:08:30 | <motle> | the thing is like a net with type abstraction everywhere |
| 2021-07-29 18:08:42 | <motle> | ends up like a program representation |
| 2021-07-29 18:08:49 | <motle> | wierdly dynamic |
| 2021-07-29 18:08:52 | <lechner> | roboguy_: i love haskell though. where can i get my tattoo :) |
| 2021-07-29 18:09:12 | × | superstar64 quits (~superstar@2600:1700:ed80:50a0:d250:99ff:fe2c:53c4) (Quit: Leaving) |
| 2021-07-29 18:09:28 | <motle> | just for the graph really, it uses the stencil convolution for data and program evaluation ordering |
| 2021-07-29 18:09:35 | <roboguy_> | lechner: haha, not sure! |
| 2021-07-29 18:10:10 | × | mikoto-chan quits (~mikoto-ch@ip-193-121-10-50.dsl.scarlet.be) (Ping timeout: 256 seconds) |
| 2021-07-29 18:10:21 | <lechner> | roboguy_: it would be a giant "λ" ! |
| 2021-07-29 18:10:24 | <roboguy_> | motle: hmm, how familiar are you with traditional ways to implement those things? |
| 2021-07-29 18:10:34 | <motle> | this is why that example |
| 2021-07-29 18:10:44 | <motle> | graph, pointer, stencil? (classes) |
All times are in UTC.