Newest at the top
| 2025-07-14 17:46:08 +0200 | Frostillicus | (~Frostilli@pool-71-174-119-69.bstnma.fios.verizon.net) |
| 2025-07-14 17:44:58 +0200 | Frostillicus | (~Frostilli@pool-71-174-119-69.bstnma.fios.verizon.net) (Read error: Connection reset by peer) |
| 2025-07-14 17:38:44 +0200 | weary-traveler | (~user@user/user363627) user363627 |
| 2025-07-14 17:30:22 +0200 | ChaiTRex | (~ChaiTRex@user/chaitrex) ChaiTRex |
| 2025-07-14 17:29:44 +0200 | Frostillicus | (~Frostilli@pool-71-174-119-69.bstnma.fios.verizon.net) |
| 2025-07-14 17:28:31 +0200 | weary-traveler | (~user@user/user363627) (Remote host closed the connection) |
| 2025-07-14 17:28:13 +0200 | ChaiTRex | (~ChaiTRex@user/chaitrex) (Ping timeout: 244 seconds) |
| 2025-07-14 17:27:53 +0200 | <tomsmeding> | that's the fun way :) |
| 2025-07-14 17:27:14 +0200 | <juri_> | I'm finding out all of the things the hard way. :) |
| 2025-07-14 17:25:39 +0200 | acidjnk | (~acidjnk@p200300d6e70b6600550565d8f561750a.dip0.t-ipconnect.de) (Ping timeout: 252 seconds) |
| 2025-07-14 17:24:45 +0200 | <tomsmeding> | iirc repa does not have a forward permutation primitive, but perhaps for machine learning you don't need that very much |
| 2025-07-14 17:23:42 +0200 | <tomsmeding> | sure |
| 2025-07-14 17:23:38 +0200 | <juri_> | I'm not tied-tied yet. it's just the first tool grabbed. |
| 2025-07-14 17:23:18 +0200 | <tomsmeding> | programming models are very similar |
| 2025-07-14 17:23:07 +0200 | <tomsmeding> | if you're not tied to exactly the 'repa' library, tough, repa code and accelerate code should be fairly easily inter-convertible |
| 2025-07-14 17:22:36 +0200 | CiaoSen | (~Jura@2a02:8071:64e1:da0:5a47:caff:fe78:33db) (Ping timeout: 276 seconds) |
| 2025-07-14 17:22:35 +0200 | <juri_> | tl;dr: future problems. right now, i'm still implementing the transformer block. |
| 2025-07-14 17:22:11 +0200 | <juri_> | oh. yeah. :) |
| 2025-07-14 17:22:11 +0200 | <tomsmeding> | you can certainly interpret accelerate into repa but that's... a bit pointless? |
| 2025-07-14 17:21:53 +0200 | <tomsmeding> | that's the other way round |
| 2025-07-14 17:21:48 +0200 | <tomsmeding> | yeah repa's typing is not suitable for a deeply embedded language |
| 2025-07-14 17:21:47 +0200 | <juri_> | https://github.com/blambo/accelerate-repa looks like someone's attempt.. 14 years back. :) |
| 2025-07-14 17:20:43 +0200 | <tomsmeding> | I expect that the typing of repa is too haskelly to be able to make a proper "accelerate backend" for repa |
| 2025-07-14 17:19:54 +0200 | <juri_> | accelerate-io-repa is part of it. haven't done the "does it work" legwork. |
| 2025-07-14 17:18:35 +0200 | <tomsmeding> | juri_: where's this repa-accelerate? |
| 2025-07-14 17:17:43 +0200 | ttybitnik | (~ttybitnik@user/wolper) (Remote host closed the connection) |
| 2025-07-14 17:10:29 +0200 | trickard_ | trickard |
| 2025-07-14 17:09:00 +0200 | <Clint> | cool |
| 2025-07-14 17:06:35 +0200 | <juri_> | I'm not, right now. I've got a crate of FPGA accelerators, for when i rewrite everything in clash. it looks like my code should be trivial to run with accelerate, thanks to repa-accelerate. |
| 2025-07-14 17:06:09 +0200 | trickard_ | (~trickard@cpe-92-98-47-163.wireline.com.au) |
| 2025-07-14 17:05:12 +0200 | poscat | (~poscat@user/poscat) poscat |
| 2025-07-14 17:05:05 +0200 | <Clint> | juri_: what are you doing for hardware acceleration? |
| 2025-07-14 17:04:54 +0200 | poscat | (~poscat@user/poscat) (Read error: Connection reset by peer) |
| 2025-07-14 17:03:35 +0200 | poscat | (~poscat@user/poscat) poscat |
| 2025-07-14 17:02:44 +0200 | <juri_> | I got multi-head attention working yesterday. now drawing the rest of the owl. |
| 2025-07-14 17:02:03 +0200 | <juri_> | I don't suppose anyone *else* is doing AI with haskell? :) |
| 2025-07-14 17:00:22 +0200 | poscat0x04 | (~poscat@user/poscat) (Remote host closed the connection) |
| 2025-07-14 17:00:21 +0200 | trickard_ | (~trickard@cpe-92-98-47-163.wireline.com.au) (Ping timeout: 252 seconds) |
| 2025-07-14 16:58:33 +0200 | AlexZenon | (~alzenon@178.34.150.30) |
| 2025-07-14 16:55:29 +0200 | dtman34 | (~dtman34@2601:447:d182:6512:c2f9:c3a:b83d:6490) dtman34 |
| 2025-07-14 16:55:09 +0200 | trickard_ | (~trickard@cpe-92-98-47-163.wireline.com.au) |
| 2025-07-14 16:54:34 +0200 | mceresa | (~mceresa@user/mceresa) mceresa |
| 2025-07-14 16:54:33 +0200 | trickard_ | (~trickard@cpe-92-98-47-163.wireline.com.au) (Ping timeout: 245 seconds) |
| 2025-07-14 16:54:30 +0200 | dtman34 | (~dtman34@2601:447:d182:6512:c2f9:c3a:b83d:6490) (Quit: ZNC 1.8.2+deb3.1+deb12u1 - https://znc.in) |
| 2025-07-14 16:54:04 +0200 | AlexZenon | (~alzenon@178.34.150.30) (Ping timeout: 260 seconds) |
| 2025-07-14 16:53:46 +0200 | mceresa | (~mceresa@user/mceresa) (Remote host closed the connection) |
| 2025-07-14 16:51:19 +0200 | user363627 | (~user@user/user363627) (Ping timeout: 268 seconds) |
| 2025-07-14 16:47:32 +0200 | weary-traveler | (~user@user/user363627) user363627 |
| 2025-07-14 16:47:14 +0200 | dtman34 | (~dtman34@2601:447:d182:6512:c2f9:c3a:b83d:6490) dtman34 |
| 2025-07-14 16:46:22 +0200 | dtman34 | (~dtman34@c-73-242-68-179.hsd1.mn.comcast.net) (Quit: ZNC 1.8.2+deb3.1+deb12u1 - https://znc.in) |