Commit Graph

  • bb6d3bc7fe docs: refresh main AI Horde documentation main tazlin 2026-11-04 13:33:05 -05:00
  • f4145e91b1 docs: add how-to for native text generation requests tazlin 2026-11-04 13:32:27 -05:00
  • 8928d4d49f docs: rewrite readmes and point FAQ to website tazlin 2026-11-04 12:57:31 -05:00
  • 99f09720ad docs: fix index for ADR 42 status tazlin 2026-11-04 12:43:58 -05:00
  • cba3df6786 docs: mark ADR 42 accepted tazlin 2026-11-04 12:12:06 -05:00
  • 1f17469c70 fix(kudos): the migration's index builds and CHECK validation run without lock or statement timeouts tazlin 2026-11-04 00:47:14 -05:00
  • 7337c844f1 feat(kudos): legacy sub-ranges become eligible on their own age tazlin 2026-11-04 00:24:25 -05:00
  • 1215a14bee feat(kudos): retention health and a gauge report pending units per dataset tazlin 2026-11-03 23:44:55 -05:00
  • 4580ec70a2 fix(kudos): the ledger swap holds the applier lock and bounds its lock waits, and retention guards lost statistics dumps and restored partitions tazlin 2026-11-03 23:12:54 -05:00
  • c98476c921 feat(kudos): the migration creates eight weekly partitions and the drop grace defaults to seven days tazlin 2026-11-03 19:23:31 -05:00
  • 65d32bbad5 feat(kudos): archive audit reports noncurrent bytes and delete markers per tier tazlin 2026-11-03 18:58:34 -05:00
  • f4608ac7f1 docs(kudos): eviction on a versioned bucket keeps the deleted object as a noncurrent version tazlin 2026-11-03 18:40:00 -05:00
  • 214da72d09 feat(kudos): archive manifests record column types and restore builds the scratch table from the manifest alone tazlin 2026-11-03 18:32:44 -05:00
  • f89f00b8fd fix(kudos): archive checks before the drop, snapshot invalidation across a recheckpoint and a measured upload budget tazlin 2026-11-03 18:05:39 -05:00
  • 0878861113 fix(kudos): keep checkpoints and archive objects consistent with the catalog; per-tier archive budgets tazlin 2026-11-03 13:23:30 -05:00
  • 59e6defc97 fix(kudos): attach new partitions without an exclusive lock; size the key backfill to the key lifetime tazlin 2026-11-03 13:23:30 -05:00
  • 4063734bd1 docs(kudos): event retention explanation, reference and operations guide tazlin 2026-11-03 12:30:19 -05:00
  • 77b6a9d752 feat(kudos): retention tick, metrics, operator commands and the privacy disclosure tazlin 2026-11-03 12:30:06 -05:00
  • e494ee788e feat(kudos): bounded idempotency keys and checkpoint-aware reconciliation tazlin 2026-11-03 12:29:50 -05:00
  • 0d22544190 feat(kudos): retention engine with rollups, checkpoints, archive export and guarded drops tazlin 2026-11-03 12:29:33 -05:00
  • 5564c077bd feat(kudos): S3 archive client with a byte budget and verified restore tazlin 2026-11-03 12:29:16 -05:00
  • aef5f44d15 feat(kudos): partition the event tables by created and convert in place tazlin 2026-11-03 12:29:05 -05:00
  • 871f6f6b0b docs(kudos): decide day-grain statistic rollups and a currency archive after a 30-day online window tazlin 2026-11-03 12:20:41 -05:00
  • e64f009f35 chore: bump GitHub Actions versions in CI workflows tazlin 2026-11-02 20:44:23 -05:00
  • 0822891e84 docs: wording, structure, and chat accuracy tazlin 2026-11-02 20:38:38 -05:00
  • 24be7096c5 docs: update zensical to latest tazlin 2026-11-02 17:41:06 -05:00
  • aed2df12ed chore: update haidra-assets pointer tazlin 2026-11-02 16:35:27 -05:00
  • 58a4c141ea fix(pop): trusted-only requests in an untrusted owner's pop page tazlin 2026-11-02 16:29:23 -05:00
  • d33285d12e feat(front): add OpenAPI service description tazlin 2026-11-02 16:06:55 -05:00
  • 780359f0db feat(front): label every request's outcome instrument tazlin 2026-11-02 15:01:29 -05:00
  • 1e80b5ddce ci: chat contract mix in the Locust smoke; image acceptance jobs tazlin 2026-11-02 14:11:59 -05:00
  • c3f38e34c2 test(stress): expect the lowered max_tokens and the parallel_tool_calls card entry tazlin 2026-11-02 14:11:59 -05:00
  • 3d28dcf895 docs(fix): add missing adr flowchart pointer tazlin 2026-11-02 13:34:46 -05:00
  • b3eb8a8877 docs: adjust copy for legibility, tone and readability tazlin 2026-11-02 10:16:25 -05:00
  • 7b5420388b test: share the wire-example builders and fixtures through _fake_chat_worker and conftest tazlin 2026-11-02 10:06:02 -05:00
  • 9d26500362 docs: add checked chat wire examples and link the reference pages to them tazlin 2026-10-31 14:22:24 -04:00
  • f64dc65c48 feat(tools): add a minimal chat worker for OpenAI-compatible backends tazlin 2026-10-31 14:22:14 -04:00
  • 86dcf53b90 test: share the chat request and worker helpers through _fake_chat_worker tazlin 2026-10-31 14:22:02 -04:00
  • f26a7ba5b0 docs: describe koboldcpp 1.124's embedded chat worker and text worker details tazlin 2026-10-31 11:28:40 -04:00
  • 0fa5c4a97d fix(chat): require an image transport of a capable vision worker tazlin 2026-11-02 12:27:51 -05:00
  • b0e6f94c3b fix(chat): offer chat requests over every worker's capacity to slower workers tazlin 2026-11-02 10:53:54 -05:00
  • 844c7a6a01 feat(chat): instrument the reasoning default, budget markers and completion content tazlin 2026-11-02 09:52:18 -05:00
  • 7d13ce4508 feat(chat): send reasoning_effort medium to declaring workers when the client omits it tazlin 2026-11-01 15:34:25 -05:00
  • 993e45c6cf fix(chat): count reasoning in the charge, the reward and the speed sample tazlin 2026-11-01 15:17:57 -05:00
  • 1d8ee46c4a feat(front): return reasoning by default and pace a post's reasoning before its text tazlin 2026-11-01 15:09:56 -05:00
  • 451ad32dc4 feat(front): pace relayed chat deltas over the worker's post interval tazlin 2026-11-01 11:20:13 -05:00
  • d829709d9a fix(chat): refuse unserved models, lower max_tokens to declared limit tazlin 2026-11-01 11:20:12 -05:00
  • 6bdd5b122d fix(worker): apply each pop's kudos policy before the check-in debounce tazlin 2026-10-31 12:27:44 -04:00
  • 3838a8775a fix(worker): apply each pop's kudos policy before the check-in debounce fix/image-pop-policy-inputs tazlin 2026-10-31 12:27:44 -04:00
  • a32df4a734 test: fail CI when a text worker column has no place in worker details tazlin 2026-10-31 11:31:17 -04:00
  • 78b3f6c250 feat: expose the chat declaration in text worker details tazlin 2026-10-31 11:20:59 -04:00
  • f60aff9e99 feat: expose chat/prompt worker capability db0 2026-10-31 15:26:12 +01:00
  • 9dea799a1f test: fail CI when a text worker column has no place in worker details tworkerinfo tazlin 2026-10-31 11:31:17 -04:00
  • 37b89bfea9 feat: expose the chat declaration in text worker details tazlin 2026-10-31 11:20:59 -04:00
  • 012a14c237 feat: expose chat/prompt worker capability db0 2026-10-31 15:26:12 +01:00
  • 017a0f8a2d ci: export the evidence store credentials and region from the test stack tazlin 2026-10-31 10:23:39 -04:00
  • 64745bcaa4 ci: keep the limiter on for the in-process suite and export the evidence bucket to stack jobs tazlin 2026-10-31 01:26:39 -04:00
  • 17ede0f868 fix(chat): fit chat requests as truncate_oldest by default tazlin 2026-10-31 01:03:39 -04:00
  • ab42a730d7 feat(front): label UnsupportedChatParameter refusals by the refused parameter tazlin 2026-10-31 01:04:20 -04:00
  • 2a7361dabb fix(text): refuse NoEligibleWorker in place of a kudos refusal when no worker can serve the request tazlin 2026-10-31 00:58:24 -04:00
  • ec54cb772d ci: run the in-process pytest suite against the test stack with the limiter off tazlin 2026-10-31 00:04:29 -04:00
  • 05430d4a14 chore: add the licence header to the Lua script and drop a duplicate nav entry tazlin 2026-10-30 23:51:32 -04:00
  • 8f911bfef0 feat(text): count chat worker faults and split stale aborts and settlements by chat tazlin 2026-10-30 23:36:06 -04:00
  • 5b243c130f feat(text): export the free output allowance decisions and queue waits tazlin 2026-10-30 22:49:56 -04:00
  • d120b922e9 feat(text): enforce the speed gate for chat requests and shadow it for legacy text tazlin 2026-10-30 22:44:42 -04:00
  • 7158021d5b fix(chat): refuse a reply for an aborted generation before the conformance check tazlin 2026-10-30 22:46:46 -04:00
  • cbf1d02beb fix(chat): return ChatJobAborted on a progress post for an aborted generation tazlin 2026-10-30 22:41:46 -04:00
  • cada266d67 docs(text): state the lease as a range of one cleanup pass and the upfront refusal with both figures tazlin 2026-10-30 22:36:48 -04:00
  • 394a805f69 fix(text): skip the quote cache for a dry run that may need upfront kudos tazlin 2026-10-30 22:26:33 -04:00
  • 08ce300e63 refactor(db): compare booleans and nulls with the SQLAlchemy helpers tazlin 2026-10-28 18:43:47 -04:00
  • a1980952ae docs(decisions): record the chat and text-API decisions in their final form tazlin 2026-10-28 18:41:56 -04:00
  • 7db39eb0fd feat(text): size the free output allowance by the estimated wait and count competing work per worker tazlin 2026-10-28 18:41:38 -04:00
  • 553addf0ca fix(text): lease text jobs for a fixed term renewed by progress tazlin 2026-10-28 18:41:09 -04:00
  • 40ec86e376 fix(text): lease a text job by its requested output length tazlin 2026-10-26 23:06:33 -04:00
  • 12caf086d1 chore: use required horde_sdk v0.33.3 tazlin 2026-10-26 19:14:03 -04:00
  • 4189a4db07 feat(text): size the free output allowance by the request's eligible capacity tazlin 2026-10-24 13:45:26 -04:00
  • dd546f0383 feat(chat): cap chat images by total bytes per trust level tazlin 2026-10-24 13:45:07 -04:00
  • d0868b30c0 feat(chat): accept reasoning_effort and route it by a declared capability tazlin 2026-10-24 13:43:40 -04:00
  • 535e4a40c4 tools: track the chat stack drivers, probes and recipes with a usage page tazlin 2026-10-22 11:56:54 -04:00
  • 8eb390528a test(stress): chat front scale demo and chat contract mix tazlin 2026-10-17 15:43:24 -04:00
  • a9477aa15a feat(text): estimate a request's wait from the capacity eligible to serve it tazlin 2026-10-22 11:41:13 -04:00
  • 919dd78c5f perf(eligibility): select a request's eligible workers and their collections in one statement tazlin 2026-10-22 11:41:12 -04:00
  • 9b3e84076f feat(observability): count worker availability cache lookups by outcome tazlin 2026-10-22 10:37:22 -04:00
  • 84bfd2aaa0 feat(oai): record queue refusal facts, profile the front and trace relay reads tazlin 2026-10-22 10:37:22 -04:00
  • 82461a66d2 feat(chat): accept a list of acceptable models and serve whichever an online worker runs tazlin 2026-10-21 18:50:31 -04:00
  • 0bd2cb4e4b feat(chat): fit a long history by dropping its oldest turns under context_fit truncate_oldest tazlin 2026-10-21 18:04:37 -04:00
  • ffb26ba6b5 feat(chat): settle a cancelled streamed generation on the text it posted tazlin 2026-10-21 17:10:06 -04:00
  • 34ffc4d7bf feat(chat): give chat submits separate per-address request limits tazlin 2026-10-22 11:41:14 -04:00
  • 1f909a91d3 feat(chat): availability_mode fail_fast and the max_kudos cap tazlin 2026-10-21 16:12:14 -04:00
  • 28e78103ae feat(chat): carry a reply's reasoning in a separate field, hidden unless asked for tazlin 2026-10-21 14:24:08 -04:00
  • aaa8ef2fb8 feat(chat): refuse a request over every online worker's chat limits at once tazlin 2026-10-21 14:24:07 -04:00
  • d8068eea4b feat(chat): stream worker progress to clients through a retained Redis stream tazlin 2026-10-21 11:11:19 -04:00
  • 5de56325e8 feat(chat): offload a large request's messages to one object in a configurable bucket tazlin 2026-10-21 11:11:19 -04:00
  • 444939fb31 perf(status): read generation counts from the completion snapshot in one grouped pass tazlin 2026-10-20 12:12:59 -04:00
  • 295dc624ad perf(scheduling): skip stall evidence inside the minimum window and seed availability at creation tazlin 2026-10-20 12:12:58 -04:00
  • 7c90b3236a feat(oai): trace the front and size it for stalled queues tazlin 2026-10-19 23:15:04 -04:00
  • b47dae8d9c docs: describe the chat endpoints in the API docs tazlin 2026-10-19 20:44:38 -04:00
  • 1f89539b24 refactor(chat): read the chat vocabulary, limits and defaults from horde_sdk 0.33.1 tazlin 2026-10-19 17:50:05 -04:00
  • 0dd3aef41c feat(chat): refusal reasons, worker fault reasons and nonconformant reply accounting tazlin 2026-10-19 16:25:11 -04:00
  • ae799932b1 feat(chat): align request validation with OpenAI tazlin 2026-10-19 14:40:32 -04:00