34 Commits

Author SHA1 Message Date
CoolBoy
2c748dcec2 perf: complete history bootstrap in one streaming phase 2026-09-04 11:35:43 +08:00
CoolBoy
54ed4351a0 fix: stream cold-start history into decision state 2026-09-04 11:20:51 +08:00
CoolBoy
d60551e130 refactor: retain decision state instead of full logs 2026-09-04 10:49:04 +08:00
CoolBoy
ff73768537 feat: compact durable logs and refine framework routing 2026-09-04 10:30:21 +08:00
CoolBoy
5b1ec4d3eb feat: add tiered durable state and memory bounds 2026-08-22 14:15:25 +08:00
CoolBoy
6eb7ded984 fix: keep agent alive across worker crashes 2026-08-21 03:38:06 +08:00
CoolBoy
85f6cb5157 fix: remove slow system git build dependency 2026-08-15 19:47:24 +08:00
CoolBoy
467eaccd59 fix: install git for durable state recovery 2026-08-15 19:33:07 +08:00
CoolBoy
4260793c03 feat: add durable success-first modelhub agent 2026-08-15 19:24:26 +08:00
CoolBoy
91d1d3d87d feat: make model age admission-only 2026-08-12 10:18:09 +08:00
CoolBoy
86088ce577 fix: enrich cleanup from active queue context 2026-08-12 09:00:51 +08:00
CoolBoy
d8ed7fa1c2 Bootstrap architecture rules from task history 2026-08-12 08:41:07 +08:00
CoolBoy
7ec875563e feat: dynamically clean incompatible architectures 2026-08-12 08:19:53 +08:00
CoolBoy
615bcad124 feat: learn framework architecture incompatibilities 2026-08-12 08:02:51 +08:00
CoolBoy
d908706f9a fix: protect running validation tasks from queue cleanup 2026-08-12 01:04:34 +08:00
CoolBoy
2065ad6abc feat: derive recent queue reserve from account capacity 2026-08-11 01:43:40 +08:00
CoolBoy
f12d96b138 feat: reserve queue capacity for recent models 2026-08-11 01:29:54 +08:00
CoolBoy
38ab25fc3c refactor: remove LLM from online decisions 2026-08-11 00:59:40 +08:00
CoolBoy
9645973468 feat: clean deterministic OOM tasks on startup 2026-08-11 00:52:16 +08:00
CoolBoy
3d15f60284 feat: add failure-aware preflight and Qwen review 2026-08-10 21:44:42 +08:00
CoolBoy
5e47d9e695 feat: prioritize proven GPU framework combinations 2026-08-05 18:21:59 +08:00
CoolBoy
882479e43e feat: add queue-aware adaptive scheduling 2026-08-04 20:22:08 +08:00
CoolBoy
ccae7ff8f3 fix: prevent repeated model GPU submissions 2026-08-02 18:48:47 +08:00
CoolBoy
8e68c6f611 fix: refill from expanded candidate history 2026-08-02 17:45:21 +08:00
CoolBoy
eab5ab6dce feat: add adaptive GPU scheduling 2026-08-02 16:59:44 +08:00
CoolBoy
80a1b8518d fix: respect ModelScope model page limit 2026-08-02 16:03:44 +08:00
CoolBoy
0e42ef73dc fix: replace duplicate submissions while refilling queues 2026-08-02 15:44:28 +08:00
CoolBoy
670c76fe4e fix: reject invalid GPU submission templates 2026-08-01 17:16:03 +08:00
CoolBoy
0cabc43abf fix: coordinate concurrent account capacity filling 2026-08-01 16:50:50 +08:00
CoolBoy
c06169d906 Rebuild agent on original pooled runner 2026-07-10 02:14:08 +08:00
CoolBoy
3a2fa86e1e Switch agent discovery to ModelScope 2026-07-10 01:44:16 +08:00
CoolBoy
5460fbd108 Add embedded Hugging Face token fallback 2026-07-10 01:02:25 +08:00
CoolBoy
597d3b1d86 Add embedded ModelHub auth fallback 2026-07-10 00:54:26 +08:00
CoolBoy
8c4bbff413 adapt for ModelHub agent platform 2026-07-10 00:37:43 +08:00