!fXpAvneDgyJuYMZSwO:nixos.org

Nix Data Science

80 Members
28 Servers

Load older messages


SenderMessageTime
30 Jun 2022
@necrophcodr:matrix.orgnecrophcodrI'd prefer to use Nix for that though, if something like it exists. jupyterWith is not close to that behaviour, and wouldn't work on a self-hosted jupyter platform as far as I can tell, especially for the users of the platform (which is my core focus as a data scientist)08:15:59
@necrophcodr:matrix.orgnecrophcodrjupyterWith is great if what you want to do is setup a reproducible simple environment on your own system or for yourself, or for a general reproducible platform, but not for individually reproduced (even foreign) notebooks on a hosted platform.08:17:22
@necrophcodr:matrix.orgnecrophcodr
In reply to @schnecfk:ruhr-uni-bochum.de
There is a nix kernel in https://github.com/tweag/jupyterWith
The nix kernel might be very useful though, if it's possible to spin up other per-cell kernels in that
08:19:04
5 Jul 2022
@rgrunbla:matrix.orgRémy Grünblatt changed their display name from Reventlov to Rémy Grünblatt.12:36:43
@carlthome:matrix.orgcarlthomeWonder if anyone is thinking about seeing ML model training as a nix-build and adding the resulting model binaries to the nix store, similarly to how DVC does it: https://dvc.org/doc/user-guide/project-structure/internal-files#structure-of-the-cache-directory Crazy? Reproducible experiment tracking as a build process rather than application state?21:35:28
@jb:vk3.wtfjbedoYeah I do that a lot21:38:54
@jb:vk3.wtfjbedoAlso run bioinformatics pipelines similarly: https://github.com/papenfusslab/bionix21:40:03
6 Jul 2022
@jez:petrichor.meJez (he/him) 🌦️Makes sense that the outputs (including intermediate ones) should go in the nix store and be immutable06:39:46
@jez:petrichor.meJez (he/him) 🌦️Although how do you deal with nondeterminism? Would it work to consider the RNG seed to be as part of the build input to recovery reproducibility? 06:41:46
@jez:petrichor.meJez (he/him) 🌦️* Although how do you deal with nondeterminism? Would it work to consider the RNG seed to be as part of the build input to recover reproducibility? 06:48:39
@jb:vk3.wtfjbedoi set my seeds so it's deterministic 06:48:58
@jb:vk3.wtfjbedothere's that new non-determinism feature of nix that might be useful if you can't do that06:50:38
@jez:petrichor.meJez (he/him) 🌦️ooh, I hadn't heard about that06:53:32
@jb:vk3.wtfjbedohttps://github.com/NixOS/nix/pull/622706:55:00
@jb:vk3.wtfjbedoi haven't really thought about this in the context of data pipelines, just remember seeing it pass my notifications06:55:36
@cdepillabout:matrix.orgcdepillabout

While I don't have a huge amount of experience with this, one thing to keep in mind is that if you have huge datasets, they end up in the Nix store (unless you're doing something unusual) along with your output model. Your Nix store can really blow up in space. You may need to make sure you're garbage collecting frequently, but that may reduce the helpfulness of running training with Nix in the first place.

You'd also have to be careful with running builds on a machine that pushes results to a shared cache, since huge datasets and models could really clear out everything else in a cache.

You may also want to setup your Nix builders so they can see your GPUs. And fiddle with the max-jobs option so that two builds aren't trying to train in parallel.

07:45:28
@cdepillabout:matrix.orgcdepillaboutIf you're trying to productionize something, it may be easier to start with a system that already has answers for all this type of stuff (some MLOps service or something?), rather than rolling your own with Nix.07:55:13
9 Jul 2022
@betaboon:0x80.ninjabetaboon changed their profile picture.11:33:12
12 Jul 2022
@pederbs:pvv.ntnu.nopederbs joined the room.22:55:46
26 Jul 2022
@tinybronca:sibnsk.nettinybronca changed their display name from tinybronca to tailrec.14:39:42
@tinybronca:sibnsk.nettinybronca changed their display name from tailrec to tinybronca.15:40:54
@ipk:h0n3yb4dg3r.ems.hostIPK joined the room.18:32:31
27 Jul 2022
@collinarnett:matrix.orgCollin Arnett joined the room.20:39:42
1 Aug 2022
@better_sleeping:converser.eu@better_sleeping:converser.eu joined the room.09:06:29
@better_sleeping:converser.eu@better_sleeping:converser.eu left the room.09:06:44
@collinarnett:matrix.orgCollin ArnettHow do people deal with packages trying to download models at runtime to the nix store? This is more of a general question since I've noticed that a lot of data science libraries nowadays are just downloading dependencies at runtime.18:18:39
2 Aug 2022
@FRidh:matrix.orgFRidh
In reply to @collinarnett:matrix.org
How do people deal with packages trying to download models at runtime to the nix store? This is more of a general question since I've noticed that a lot of data science libraries nowadays are just downloading dependencies at runtime.
I think all you can do is contact upstream to fix where they install to. If they need models, that could be fine, but then they should probably go to ~/.cache
09:26:59
@collinarnett:matrix.orgCollin Arnett
In reply to @FRidh:matrix.org
I think all you can do is contact upstream to fix where they install to. If they need models, that could be fine, but then they should probably go to ~/.cache
Yeah I think this is probably the best solution. Thank you :)
17:16:04
6 Aug 2022
@yuu:matrix.orgyuu changed their display name from yuu to yuu[m].17:42:49
@yuu:matrix.orgyuu changed their display name from yuu[m] to yuu.20:35:43

There are no newer messages yet.


Back to Room ListRoom Version: 6