Commit Graph

20 Commits

Author SHA1 Message Date
Yannic Kilcher 48212079f4 unified queueing 2023-02-11 01:31:25 +01:00
Yannic Kilcher 097c918cb9 bugfix 2023-02-10 23:30:42 +01:00
Yannic Kilcher 91e5b11315 touch 2023-02-10 23:01:34 +01:00
Yannic Kilcher d1aea98ad5 added dockerfile for worker-full 2023-02-10 22:53:40 +01:00
Yannic Kilcher 90c3d5640e Added database to inference server (#1446)
* added db for inference

* fixed dockerfiles for inference
2023-02-10 22:51:35 +01:00
Andrew Maguire b60eb1e1ae minimal fastapi prom metrics (#1426)
* minimal fastapi prom metrics
2023-02-10 14:37:43 +01:00
Yannic Kilcher 4076afd0d8 added token buffer for catchiing stop sequences 2023-02-09 23:46:44 +01:00
Yannic Kilcher ed7d920e5d robustifying inference 2023-02-09 15:31:46 +01:00
Yannic Kilcher c53d8e9bce changed return type of GET chat 2023-02-09 08:52:45 +01:00
Yannic Kilcher a85cc0a47d endpoint to list chats 2023-02-09 08:49:01 +01:00
Yannic Kilcher 27671e3220 HF inference fixes 2023-02-09 00:27:51 +01:00
Yannic Kilcher bab056a73b switched to HF text-generation-inference 2023-02-08 23:52:39 +01:00
Alan Jean 7961fe3b71 chore(inference): add docs about the containerized inference stack 📝
Fixes #913
2023-01-29 05:30:00 +04:00
Yannic Kilcher f885e27961 added horizontal layout to inference dev setup 2023-01-29 00:37:28 +01:00
Yannic Kilcher 45a4b09eae added setup instructions to readme 2023-01-27 15:42:01 +01:00
Yannic Kilcher ae5d16f394 added a tmux inference dev setup script 2023-01-27 15:40:21 +01:00
Yannic Kilcher a0f4449e9f added seed to parameters 2023-01-27 15:23:19 +01:00
Yannic Kilcher 040344a41f made inference server a bit more robust 2023-01-26 21:01:52 +01:00
Yannic Kilcher f1edcc8a28 added streaming worker 2023-01-26 16:41:57 +01:00
Yannic Kilcher 1709dc0324 Initial implementation of the inference system (#869)
* very primitive implementation of inference

* re-worked with security in mind

* removed polling from clients

* switched workers to websockets

* implemented back and forth chats
2023-01-21 22:38:18 +01:00