403 Commits

Author SHA1 Message Date
Jake Poznanski
370dbba2bc new build 2024-11-15 15:17:10 -08:00
Jake Poznanski
9ce243ee6c no weka on augusta 2024-11-15 14:19:18 -08:00
Jake Poznanski
eefb045859 Single cluster fix 2024-11-15 13:30:27 -08:00
Jake Poznanski
2e1d0b6a90 Fix 2024-11-15 13:21:55 -08:00
Jake Poznanski
748b095e0f Fix 2024-11-15 13:19:23 -08:00
Jake Poznanski
80ba562bb2 Fixing timeout situation 2024-11-15 13:18:13 -08:00
Jake Poznanski
65763de178 Don't retry accessdenied errors 2024-11-15 13:02:38 -08:00
Jake Poznanski
2c52664301 Cleaner exit 2024-11-15 12:54:45 -08:00
Jake Poznanski
77c82fdb3a New version with aiohttp fixes 2024-11-15 12:48:36 -08:00
Jake Poznanski
ae1e4bc07e More realistic results 2024-11-15 11:35:10 -08:00
Jake Poznanski
770da2b7ae Docker 2024-11-15 11:25:24 -08:00
Jake Poznanski
bfe4211dcc Debugging timeout errors and other things 2024-11-15 11:23:38 -08:00
Jake Poznanski
fd17652d55 Trying to make it faster 2024-11-15 11:06:50 -08:00
Jake Poznanski
278422b8ff Fixing one max context issue 2024-11-15 10:03:26 -08:00
Jake Poznanski
62de9fe9b5 weka fix 2024-11-14 14:52:19 -08:00
Jake Poznanski
9a1e82f9d9 Logging 2024-11-14 14:47:19 -08:00
Jake Poznanski
fe0574c725 Cleanup code, s3 retries 2024-11-14 14:13:04 -08:00
Jake Poznanski
2c7686f8ff I think I have error handling better now 2024-11-14 13:44:54 -08:00
Jake Poznanski
8217e49153 Page calc 2024-11-14 13:38:58 -08:00
Jake Poznanski
4eab90f69b Fixing bugs 2024-11-14 13:13:27 -08:00
Jake Poznanski
b67d8e7555 Fixing work queue population 2024-11-14 12:48:46 -08:00
Jake Poznanski
827b77e8df Working on task groups 2024-11-14 12:06:13 -08:00
Jake Poznanski
a58efea133 better logging 2024-11-14 09:55:37 -08:00
Jake Poznanski
a9cf2e0272 Allow setting beaker priority 2024-11-14 09:10:28 -08:00
Jake Poznanski
41c8d552a0 exponential backoff 2024-11-14 09:02:49 -08:00
Jake Poznanski
4dcf9ed5d4 more fixes 2024-11-14 08:55:20 -08:00
Jake Poznanski
06331d740a Fix timeout 2024-11-14 08:50:33 -08:00
Jake Poznanski
8e16780b82 Beaker stuff 2024-11-14 08:49:12 -08:00
Jake Poznanski
4c3bf7045d Beaker fixes 2024-11-13 14:24:23 -08:00
Jake Poznanski
3172a1c16a Shuffling 2024-11-13 13:23:29 -08:00
Jake Poznanski
fe3c9a2709 Creds and other things 2024-11-13 13:14:33 -08:00
Jake Poznanski
a3b6962d21 fix 2024-11-13 13:05:57 -08:00
Jake Poznanski
83bb1dcd3b Dockerfile fixes 2024-11-13 12:59:52 -08:00
Jake Poznanski
6c9c785130 Using version strings 2024-11-13 12:35:40 -08:00
Jake Poznanski
9610eac4f0 Secrets management 2024-11-13 11:26:46 -08:00
Jake Poznanski
39256c19bb Beaker running 2024-11-13 10:25:35 -08:00
Jake Poznanski
867e2c9a36 Docker builds 2024-11-13 09:46:08 -08:00
Jake Poznanski
a091412079 Starting to play with docker too 2024-11-13 09:35:34 -08:00
Jake Poznanski
bce85e669a pipeline 2024-11-13 08:00:14 -08:00
Jake Poznanski
a085e8c7b3 Beaker test 2024-11-12 15:56:51 -08:00
Jake Poznanski
910c2ebcfc Downloads from s3 based on hash 2024-11-12 15:18:04 -08:00
Jake Poznanski
6598e2dc45 Control http session at the worker level 2024-11-12 13:54:45 -08:00
Jake Poznanski
fbacdd06e0 Stuff 2024-11-12 13:44:20 -08:00
Jake Poznanski
ae9b1c405d Better stats 2024-11-12 13:28:39 -08:00
Jake Poznanski
9ce28c0504 Measuring metrics better now 2024-11-12 12:56:35 -08:00
Jake Poznanski
193e5214d1 Semaphore timeout 2024-11-12 11:53:29 -08:00
Jake Poznanski
102c0e4cfc new version of sglang, server restarts, semaphore timeouts 2024-11-12 10:49:13 -08:00
Jake Poznanski
918e2f3542 Pipeline stuff 2024-11-12 09:33:53 -08:00
Jake Poznanski
691cc5a13c A few items 2024-11-12 08:34:25 -08:00
Jake Poznanski
4f2f4fda7d Quicker results by limited workers via semaphore while still utilizing gpu 2024-11-12 08:18:22 -08:00