Commit Graph

233 Commits

Author SHA1 Message Date
f1f1a42fb3 Fixes after local testing 2025-05-02 18:28:07 +02:00
e78fca8233 Gemma notebook working on runpod 2025-05-02 17:05:28 +02:00
f63e5db996 Move models directory 2025-05-01 17:28:59 +02:00
d022d41a5d Update docker image 2025-05-01 17:28:33 +02:00
Niels Geens
ff6a6ab135 Add bitsandbytes to gemma notebook 2025-05-01 10:46:33 +02:00
72b902117c Simplified bootstrap even more 2025-05-01 09:26:29 +02:00
Niels Geens
95546208e8 Gemma notebook 2025-04-30 13:36:24 +02:00
8544f12d41 Added example notebook 2025-04-29 14:58:27 +00:00
e07a62d488 Simplified bootstrap 2025-04-29 14:57:15 +00:00
Niels Geens
3557270c89 Merge branch 'master' of https://git.nielsgeens.be/llm/SIA 2025-04-29 16:50:13 +02:00
Niels Geens
2179bbe00d Create separate venv for running notebooks 2025-04-29 16:49:24 +02:00
Niels Geens
07aa80fa75 Create separate venv for running notebooks 2025-04-29 16:24:19 +02:00
36d718d326 Test train qwen 3 2025-04-29 10:17:03 +00:00
725bd4cacd wip vllm 2025-04-25 19:39:18 +00:00
9b2c2ece70 Flash attention in notebook 2025-04-23 19:38:56 +00:00
Niels Geens
4b124fa3ed Prefer previous generation gpu 2025-04-23 18:47:01 +02:00
e2e346d134 Implement 4-bit quant for train script and inference 2025-04-23 16:43:55 +00:00
2c4b254bb0 Merge branch 'master' of https://git.nielsgeens.be/llm/SIA 2025-04-23 16:38:05 +00:00
41c336165f Updated notebooks to support 4-bit quantization 2025-04-23 16:37:10 +00:00
Niels Geens
47491fc20e Redo jupyter install 2025-04-23 17:27:38 +02:00
Niels Geens
1760532e7f Merge branch 'master' of https://git.nielsgeens.be/llm/SIA 2025-04-23 15:18:39 +02:00
Niels Geens
c834141ead Retry deploy if not available 2025-04-23 15:16:41 +02:00
d7c340a937 Updated notebooks wip 2025-04-23 10:32:33 +00:00
Niels Geens
bf94a0838a Enable 2x faster inference with unsloth 2025-04-22 11:47:27 +02:00
3e471eced3 Use Unsloth for QwQ inference 2025-04-21 10:42:35 +00:00
568ec4b66d Improved docker build speed 2025-04-20 16:06:47 +02:00
def822d61f Load response when loading iteration 2025-04-20 12:31:55 +02:00
7a4028085f Simplified dockerfile 2025-04-20 12:08:57 +02:00
243e4177cb Reject whitespace in iteration parser 2025-04-20 12:08:43 +02:00
a165772411 Updated user_communication flowchart 2025-04-19 20:55:04 +02:00
c1cef08941 Speed up inference 2025-04-19 14:12:00 +00:00
9315f87f53 Fix stop bug 2025-04-19 12:15:23 +00:00
5ac701316e Fix char_trie to store all token ids 2025-04-19 11:49:09 +00:00
33aaaf4455 Fix start of assistant message detection 2025-04-18 17:46:49 +00:00
23b8c3c54d Fix CharTie multi-threading 2025-04-18 18:10:05 +02:00
64fc621d44 Fix error print in dataset 2025-04-18 12:42:24 +00:00
5e47200ece Fixes for QwQ llm engine 2025-04-18 12:30:47 +00:00
Niels Geens
c09f0766c1 Fixed auto approver and inference continuation 2025-04-18 11:36:17 +02:00
Niels Geens
023f6bba9a Wip web interface 2025-04-17 22:09:15 +02:00
Niels Geens
813c6a3f8f Update unit tests 2025-04-17 18:32:44 +02:00
Niels Geens
34fb5d814f Shared response buffer backend 2025-04-17 18:32:23 +02:00
Niels Geens
9477575421 Multi-thread chartie 2025-04-17 09:48:04 +02:00
51fed5dfd3 Implemented commend and cdata 2025-04-14 21:32:36 +02:00
bfe03dbdc2 WIP cdata and comment tokens 2025-04-13 22:02:29 +02:00
773134e65b Improved tests for tokens 2025-04-13 19:05:10 +02:00
ac2c1ae08b Added lib to clean-python in Makefile 2025-04-13 18:19:32 +02:00
5bcfe89cb9 Disabled community cloud 2025-04-13 18:15:19 +02:00
be48130a8f Use CharTrie and fix multi-byte char issue 2025-04-13 16:05:27 +02:00
f4de9e7571 Fixed token cache for whitespace bug 2025-04-13 10:03:14 +02:00
ff82173f2a Removed auto enable for QwQ 2025-04-13 09:56:01 +02:00