Commit History
-
Move default template so it doesn't override tokenizer (#987)
Unverified signature9437108 -
Set ChatML as default chat prompt template (#985)
Unverified signature619722b
-
quick fix error message (#977)
Unverified signatured323cb0 -
Add support for cohere endpoints (#976)
Unverified signature* Add support for cohere endpoints * readme updatea99cca3 -
Use jinja template for chat formatting (#730) (#744)
Unverified signature* Use jinja template for chat formatting * Add support for transformers js chat template * update to latest transformers version * Make sure to `add_generation_prompt` * unindent0819256
-
Implement Cloudflare Workers AI endpoint (#907) (#972)
Unverified signature* Implement Cloudflare Workers AI endpoint (#907) * Renamed to Cloudflare Workers AI in docs * Add note about sampling parameters * clean up env examplecb000d3
-
Google Vertex API support (#950)
Unverified signature* Added Google Vertex API support Google Application credentials can be set via GOOGLE_APPLICATION_CREDENTIALS = clientid.json * Updated the readme file * Updated the package-lock file * Updated the package-lock file Fixed minor issues * Moved "@google-cloud/vertexai" to optional dependencies * Updated for lint checker3ebd3a8
-
[Assistants] trending feature (#938)
Mishig +1: coyotte508Unverified signature* [Assistants] trending feature * cleaner sort query * create & use `AssistantStats` collection * fix grouping * rm usage of migration * ♻️ Refacto lock usage / AssistantStat * 🩹 Fix last24HoursCount.count => last24HoursCount * 🐛 Fix collection name * 🐛 Fix DB query * 🐛 Fix rate limit on assitant creation * 🐛 Fix aggregation query * ✨ Only run refreshAssistants if assistants are enabled * hide UI until we have enough stats collection --------- Co-authored-by: coyotte508 <coyotte508@gmail.com>a4c3fca
-
Expose sampling controls in assistants (#955) (#959)
Unverified signature* Expose sampling controls in assistants (#955) * Make sure all labels have the same font size * styling * Add better tooltips * better padding & wrapping * Revert "better padding & wrapping" This reverts commit 1b44086465040f2cb6bc906983cfc8d95820d6fe. * ui update * tooltip on mobile * lint * Update src/lib/components/AssistantSettings.svelte Co-authored-by: Mishig <mishig.davaadorj@coloradocollege.edu> --------- Co-authored-by: Victor Mustar <victor.mustar@gmail.com> Co-authored-by: Mishig <mishig.davaadorj@coloradocollege.edu>d4016bc
-
[Websearch] change context schema (#944)
MishigUnverified signature2e2f16c
-
Count system prompt tokens (#850)
Mishig +1: nsarrazin Unverified signature* Count sysmte prompt tokens * show error only once * fix * fix css * simplify * fix * simplify * Revert "simplify" This reverts commit 61c0a2281f9795c4a4351d9ad473993a3c01c59f. * `model.tokenizer` config & fix reactivity issues * rm gated tokenizer * use `truncate` --------- Co-authored-by: Nathan Sarrazin <sarrazin.nathan@gmail.com>50d8483 -
Enhance Dynamic User Attribute Handling in OIDC Integration (#885)
Chang Chi, Meng +2: nsarrazin, ‘XGungo’ Unverified signature* Refactor `updateUser` to Support Dynamic `nameClaim` Configuration * style: run Prettier to fix lint errors * Validate NAME_CLAIM in OIDConfig schema * Refactor userData parsing --------- Co-authored-by: Nathan Sarrazin <sarrazin.nathan@gmail.com> Co-authored-by: ‘XGungo’ <‘daniel2000890311g@gmail.com’>c6129c3
-
Prepend domain filters in search query generation (#932)
Unverified signature233d2a9 -
Show error when webpage cannot be reached or parsed (#930)
Unverified signaturee04d4e8 -
Fix multi domains for assistants (#929)
Unverified signaturecbf4205 -
Anthropic Endpoint Support (#923)
Liam DyerUnverified signaturefeat: anthropic endpoint5a31018
-
Add websearch controls for assistants (#812)
Unverified signature* remove query modifiers from generateQuery * Add backend for assistant RAG * Add front-end for updating RAG assistant * enable web parser to return plaintext directly for matching headers * Update websearch flow for handling assistant rag preferences * Add our old blocklist to .env.template * Enable websearch to run on messages depending on assistant requirements * reorganized imports * Rename vars * use projection * Add environment variable for assistant rag * fix assistant rag on runwebsearch * fix styling if rag is disabled * make sure we always omit credentials when fetching web pages * Add new checks for SSRF, with a new env var `ENABLE_LOCAL_FETCH` * Use DNS to check if the links are local or not * Add a websearch indicator * Add more tags to parser * Add indicators * Display RAG options in settings view * ui * fix rag detection * bit more spacing * fix button position in assistant form * wording (mainly) …3f5871c
-
Fix prompt caching on llama.cpp endpoints (#920)
Unverified signatureExplicitly enable prompt caching on llama.cpp endpoints Co-authored-by: Nathan Sarrazin <sarrazin.nathan@gmail.com>eb071be
-
Add openai embeddings (#915)
Martok88 +1: nsarrazin Unverified signature* Add OpenAI embedding compatibility * Use OPENAI_API_KEY by default * lint * Add default OpenAI URL replace `authorization` by `apiKey` * Add a note in readme --------- Co-authored-by: Nathan Sarrazin <sarrazin.nathan@gmail.com>f7db219
-
Include last chunk in websearch context (#912)
Unverified signatureFix bug where last chunk was dropped in websearch8e839b8 -
Make sure preprompt is set on open ai endpoint type (#913)
Unverified signature* Make sure preprompt is set on open ai endpoint type * fix preprompt passing to conversationbcfa394
-
Add limits on API endpoints (#886)
Unverified signature* Add limits on messages, conversations, assistants and messages/minute * Add max message length limit * remove rate limits from public config * add `RATE_LIMITS` to secrets * Add `MESSAGES_BEFORE_LOGIN` to secrets * replace `RATE_LIMITS` by `USAGE_LIMITS` * replace `RateLimits` by `usageLimits` and only get nEvents if needed * rename schema too * replace \r\n by \n537b6f5
-
Automated migration setup (#897)
Unverified signature* Initial migration setup for chat-ui * refresh the lock regularly while the migrations are running * add index for migrationResults * clean up code a bit * Don't try to run migrations when sveltekit is building * simplified lock code * Add early check for migrations being done Reduce timer for lock * migration use `generateSearchTokens` * Update src/lib/migrations/migrations.spec.ts Co-authored-by: Mishig <mishig.davaadorj@coloradocollege.edu> --------- Co-authored-by: Mishig Davaadorj <dmishig@gmail.com> Co-authored-by: Mishig <mishig.davaadorj@coloradocollege.edu>4dbcbb6 -
[Assistants] Filter on names (#841)
Mishig +3: Eliott C., nsarrazin, victor Unverified signature* [Assistants] Filter on names * Add `$text` index on `assistant.name` (#844) * add maxlength * better experience * use `$meta: "textScore"` * Update src/routes/assistants/+page.server.ts Co-authored-by: Eliott C. <coyotte508@gmail.com> * null, not undefined * [Assistants] Filter on names (using searchTokens) (#873) Filter with `searchTokens` * input * rm extra whitespace * hide UI before migration * rm ad-hoc migration --------- Co-authored-by: Eliott C. <coyotte508@gmail.com> Co-authored-by: Nathan Sarrazin <sarrazin.nathan@gmail.com> Co-authored-by: Victor Mustar <victor.mustar@gmail.com>10dbbd6 -
Fix issue with "continue" feature on llama.cpp endpoints (#898)
Unverified signature714ff2c -
Bug Fix: Json Decoder aggessively pulls json (#867)
Matthew Current +1: nsarrazin Unverified signature* wait for newline char b4 decoding json * remove noisy console log * linters --------- Co-authored-by: Nathan Sarrazin <sarrazin.nathan@gmail.com>9c5a826
-
Pass websearch to `preprocessMessages` (#876)
Unverified signatureThere was a bug where the websearch was fetching relevant chunks for context correctly but it wasn't getting injected into the prompt, this PR fixes that.00443e1
-
Add gemma model to prod config (#854)
Unverified signature* Add `google/gemma-7b-it` to production config * Add gemma prompt format to PROMPTS.md * Make sampling parameters optional5f78b90 -
Models page (#849)
Unverified signature* add models page * add logo * models discussion * add default hint * Update +page.svelte * mishig review * fix typecheck --------- Co-authored-by: Nathan Sarrazin <sarrazin.nathan@gmail.com>25c844d
-
✨ Add stats on conversations (#828)
Eliott C +1: nsarrazin Unverified signature* ✨ Add stats on conversations * 🩹 Fix conversation stats compute * ⚡️ Slightly better index * 🔊 * 🩹 Some fixes * 🐛 Fix aggregation query * 🐛 Fix $merge stage of query * 🐛 Fix query on sessionId * ✨ Compute weekly/monthly unique users * 🩹 Final tweaks * ⚡️ Split aggregations for better perf No need to recompute DAUs / WAUs and so on * ♻️ Deprecate PARQUET_EXPORT_SECRET & fix sec vuln * Add ADMIN_API_SECRET to CD action --------- Co-authored-by: Nathan Sarrazin <sarrazin.nathan@gmail.com>634bd69
-
[Mongo] Optimize `reports` collection query (v2) (#834)
MishigUnverified signature* [Mongo] Optimize `reports` collection query (#832) * [Mongo] Optimize `reports` collection query * perform reports query only once * rm unused async * keep both indices * correct Objectid comparison53e0136 -
Revert "[Mongo] Optimize `reports` collection query (#832)" (#833)
MishigUnverified signatureThis reverts commit 19329abf32a4f49ff048585d0f6df5887cea8465.c755352 -
[Mongo] Optimize `reports` collection query (#832)
MishigUnverified signature* [Mongo] Optimize `reports` collection query * perform reports query only once * rm unused async * keep both indices19329ab -
Allow customising defaultQuery for native Azure OpenAI support (#822)
Unverified signatureThe api-version is a required query parameter for Azure OpenAI, by supporting this we can support it without a gateway like portkey. Co-authored-by: Nathan Sarrazin <sarrazin.nathan@gmail.com>7c4d92a
-
Conversation trees (#223) (#807)
Unverified signature* work on branching * branching * work on input field * wip * wip * helper stuff * pass tests * add type guards and clean up type requirements * fixed shared conv type * more helpers stuff * clean up code * addChildren helper * addSibling * add test for addSibling & refacto addChildren tests * backend work pt. 1 * backend done * add children property to messages for easier frontend rendering * fix message id type rail * front-end work on simple linear conversation * fix title generation * convert conversations on post * server side retry works with branching * clean up buttons * fix retry feature backend * Send new messages in any subtree * make edit previous prompts feature work * fix padding * revert unneeded changes * bring back pending message * fix front-end streaming * fix initial message * Revert "fix initial message" This reverts commit 6257fe8789c52a20e7506aa453efda55c16e95bb. * …e6addfc -
[Assistants] Add new indices for efficient querying (#810)
MishigUnverified signature* [Assistants] Add new indeices for efficient querying * update indices & use `createdById` * stronger typingdb943f9
-
Make lorax openai endpoint working (#755)
Unverified signaturemake base model for lorax inference working Co-authored-by: Nathan Sarrazin <sarrazin.nathan@gmail.com>10d7a5a -
Added support for SearXNG as a websearch source (#805)
Karlostavitch1 +1: nsarrazin Unverified signature* Created searchSearxng.ts for querying and parsing searxng instance. Added searxng to the searchWeb.ts to ensure inclusion in the search initialisation script. Added searxng to +layout.server.ts to ensure the search button is enabled if url provided in .env.local assignment Updated .env to include searxng configuration Updated README.md under Web Search Config to mention addition of searxng URL use in .env * Changes requested consolidated env variable to SEARXNG_QUERY_URL (felt considering the content the word QUERY was more relevant) refactored the searchSearxng.ts to generate the query with the new variable updated related references to variable (.env, searchWeb.ts and +layout.server.ts) Updated readme and variable comment on .env to reflect the same * urefectored to display SearXNG in search dialoge when used. * lint * types --------- Co-authored-by: Nathan Sarrazin <sarrazin.nathan@gmail.com>7c22da3 -
🗃️ Create index on users.username (#819)
Eliott CUnverified signature* 🗃️ Create index on `users.username` * 🗃️ Do not do db.assistants.find({modelId: {$exists: true}) since it's always true881070b
-
Only pull from settings where the assistant is present (#806)
Unverified signature3c372b8 -
fix continue button on openai endpoint (#803)
Unverified signature12e7e30
-
Several QoL contributions (#760)
Luke +1: nsarrazin Unverified signature* allow customizing disclaimer as `PUBLIC_APP_DISCLAIMER_MESSAGE` * support passing `defaultHeaders` to `openai` endpoint * add azure openai, claude, mistral examples using `defaultHeaders` & `openai` endpoint * fix streaming being buffered behind cloudflare tunnel might help to relieve issue #598 * support new lines in model description * don't automatically generate modelUrl to huggingface fixes broken links for self-hosted or custom-named model * add `PUBLIC_APP_DISCLAIMER_MESSAGE` in `.env` * `npm run format` --------- Co-authored-by: Nathan Sarrazin <sarrazin.nathan@gmail.com>73a5c0d
-
Bring back featured check on top assistants page (#765)
Unverified signature0b2a549
-
Top assistants page (#740)
Unverified signature* Basic top assistants page * remove featured check * fix edit * ui update * ui * mishig review * fix * move feedback link to settings * misc * Hide unlisted models in assistants flow * settings + public label * tweak * font-size * Hide top assistant page for soft release * always show author --------- Co-authored-by: Victor Mustar <victor.mustar@gmail.com>786115c
-
Web search: UI update (#749)
Unverified signature* wip * ui * font-size * leading * smaller * move icon and remove loading props * rm inline svg * rm animationbccd811
-
Assistants feature (#639)
Unverified signature* First push on assistants * push fixes * fix add assistant * Sign up works * lint * mobile layout fixes * design fixes * Merge branch 'main' into feature/assistants * fix copy button * add error feedback * hide duplicate feature * remove wrong comments * add autoredirect if assistant is missing * latest changes: - add edit feature - hash assistant avatar - get rid of ugly line - check for non existent avatar - make a better looking upload icon * Update src/routes/conversation/+server.ts Co-authored-by: Mishig <mishig.davaadorj@coloradocollege.edu> * reused type more cleanly * fix type in shared conversation * fixed feature * fix: share conv with an assistant * delete assistant avatars in db when deleting avatar * affordance on avatar upload * improve assistant conv start on mobile * settings modal fly in * better mobile intro * mobile padding * link affordance * Make assistants disabled by default, b…992a8de -
[Fix] small fixes in websearch generateQuery (#727)
MishigUnverified signature13489e8 -
[SmallModel] Nit update the comment (#726)
MishigUnverified signature* [SmallModel] Nit update the comment * even more precise77adc4d
-
Continue generation feature (#707)
Unverified signature* Initial work on continue feature * Move continue button * Fix websearch with continue * Make it work with every model * Update src/routes/conversation/[id]/+server.ts Co-authored-by: Mishig <mishig.davaadorj@coloradocollege.edu> * fixes * async all the things * add reduce comment * remove log * Only show loading indicator if not continuing --------- Co-authored-by: Mishig <mishig.davaadorj@coloradocollege.edu>77399ca -
Use JSON5 for embeddings (#719)
Unverified signature6e0b0ea