Commit History

Автор SHA1 Съобщение Дата
  Daniel Hiltgen 92c81e8117 Give the final model loading more time преди 11 месеца
  Lei Jitang 7487229c34 llm/server.go: Fix 2 minor typos (#4661) преди 11 месеца
  Daniel Hiltgen 0165ba1651 Merge pull request #4638 from dhiltgen/better_error преди 11 месеца
  Daniel Hiltgen c4209d6d21 Report better warning on client closed abort of load преди 11 месеца
  Patrick Devine 4cc3be3035 Move envconfig and consolidate env vars (#4608) преди 11 месеца
  Daniel Hiltgen b37b496a12 Wire up load progress преди 11 месеца
  Jeffrey Morgan 38255d2af1 Use flash attention flag for now (#4580) преди 11 месеца
  Sam e15307fdf4 feat: add support for flash_attn (#4120) преди 11 месеца
  Patrick Devine d1692fd3e0 fix the cpu estimatedTotal memory + get the expiry time for loading models (#4461) преди 11 месеца
  Daniel Hiltgen 853ae490e1 Sanitize the env var debug log преди 11 месеца
  Patrick Devine 6845988807 Ollama `ps` command for showing currently loaded models (#4327) преди 11 месеца
  jmorganca 92ca2cca95 Revert "only forward some env vars" преди 11 месеца
  Daniel Hiltgen c4014e73a2 Fall back to CPU runner with zero layers преди 11 месеца
  Jeffrey Morgan bb6fd02298 Don't clamp ctx size in `PredictServerFit` (#4317) преди 11 месеца
  Michael Yang cf442cd57e fix typo преди 11 месеца
  Michael Yang ce3b212d12 only forward some env vars преди 11 месеца
  Michael Yang 58876091f7 log clean up преди 11 месеца
  Daniel Hiltgen d0425f26cf Merge pull request #4294 from dhiltgen/harden_subprocess_reaping преди 11 месеца
  Bruce MacDonald cfa84b8470 add done_reason to the api (#4235) преди 11 месеца
  Daniel Hiltgen 84ac7ce139 Refine subprocess reaping преди 11 месеца
  Daniel Hiltgen 920a4b0794 Merge remote-tracking branch 'upstream/main' into pr3702 преди 1 година
  Daniel Hiltgen ee49844d09 Merge pull request #4153 from dhiltgen/gpu_verbose_response преди 1 година
  Daniel Hiltgen bee2f4a3b0 Record GPU usage information преди 1 година
  Daniel Hiltgen 72700279e2 Detect noexec and report a better error преди 1 година
  Daniel Hiltgen 380378cc80 Use our libraries first преди 1 година
  Jeffrey Morgan ed740a2504 Fix `no slots available` error with concurrent requests (#4160) преди 1 година
  Jeffrey Morgan 1b0e6c9c0e Fix llava models not working after first request (#4164) преди 1 година
  Daniel Hiltgen f56aa20014 Centralize server config handling преди 1 година
  Mark Ward 321d57e1a0 Removing go routine calling .wait from load. преди 1 година
  Mark Ward ba26c7aa00 it will always return an error due to Kill() discarding Wait() errors преди 1 година