Commit Graph

27 Commits

Author SHA1 Message Date
James Ravenscroft
215a69b5af update clblast code in gpt-j model 2023-08-26 16:16:01 +01:00
James Ravenscroft
0b408510f4 add gpu offload for gpt-j models (codegen) 2023-08-26 16:11:16 +01:00
James Ravenscroft
604183380d tidy up prints in stablecoder and starcoder 2023-08-26 16:04:41 +01:00
James Ravenscroft
4a47251822 update for gpu build 2023-08-26 15:13:08 +01:00
James Ravenscroft
b2b4a1480f increase scratch on starcoder 2023-08-26 15:12:41 +01:00
James Ravenscroft
5f7155a314 add gpu offload for gptneox 2023-08-26 15:12:41 +01:00
James Ravenscroft
dc81abbc52 Merge branch 'main' into feature/model-lock 2023-08-24 14:58:44 +01:00
James Ravenscroft
f69a8f65d4 fix build? 2023-08-24 13:49:14 +01:00
James Ravenscroft
e8beac34e7 more attempts to build with boost threads 2023-08-24 13:17:51 +01:00
James Ravenscroft
cceee41f79 add boost threads 2023-08-24 13:03:12 +01:00
James Ravenscroft
2d617b458e expose batch size flag to cli 2023-08-24 11:40:19 +01:00
James Ravenscroft
0c1fc1a04e implement locking of model per request to prevent crashing when multiple requests reeived 2023-08-24 11:30:44 +01:00
James Ravenscroft
6a72e32dab add debug output to starcoder 2023-08-23 15:53:38 +00:00
James Ravenscroft
bad53ad190 add debug logs to codegen 2023-08-23 14:57:31 +00:00
James Ravenscroft
cd9baacdd7 add debug logs to stablecode 2023-08-23 14:56:36 +00:00
James Ravenscroft
88c5ede768 fix cast problem due to incorrect float notation 2023-08-23 08:12:53 +00:00
James Ravenscroft
757cc64a59 add temp and top p as cli args 2023-08-23 06:27:37 +01:00
James Ravenscroft
f7f1991e2c add huggingface request handler and refactor old req handler 2023-08-10 09:26:54 +01:00
James Ravenscroft
84869ff0f3 added support for gptneox models 2023-08-10 08:39:14 +01:00
James Ravenscroft
430733c7b8 sort out args 2023-08-05 08:29:18 +01:00
James Ravenscroft
c48cd786ed replace crow submodule with header-only impl 2023-08-04 22:48:11 +01:00
James Ravenscroft
7dee51782d tidy up arm args 2023-08-04 22:18:30 +01:00
James Ravenscroft
00ec442860 try to move arm options to top level 2023-08-04 07:14:38 +01:00
James Ravenscroft
3f3bff73b6 add processor specific flags for arm 2023-08-04 07:05:20 +01:00
James Ravenscroft
dfa4b5e74f add starcoder/wizardcoder/santacoder support 2023-07-29 15:28:07 +01:00
James Ravenscroft
fd3a127aaa add server component 2023-07-23 17:50:53 +01:00
James Ravenscroft
887d348188 add new code 2023-07-23 17:06:44 +01:00