Releases: defenseunicorns/leapfrogai
Releases · defenseunicorns/leapfrogai
v0.14.0
0.14.0 (2024-10-10)
⚠ BREAKING CHANGES
- vllm: upgrade vllm backend and refactor deployment (#854)
Features
- api: add prometheus monitoring (#1166) (81ff6a3)
- api: openai compliant annotations and vector_content retrieval (#1164) (96d89f0)
- api: reranking backend integrated in with rag (#1090) (2f80d87)
- use chunk data in NIAH and QA evals (#1176) (ad697cd)
- vllm: upgrade vllm backend and refactor deployment (#854) (fd3cbc4)
Bug Fixes
- api: make model field optional in ModifyAssistantRequest type (#1159) (75cf696)
- docs: cron-job badges should look at main branch workflows only (31178d1)
- nightly snapshots repository typo (#1177) (432b380)
- remove nightly snapshots release workflow from release-please (#1163) (48507da)
- replace file ids with replacement text (#1201) (bf14dde)
- ui: allow large transcribe and translate responses (185dcbb)
- ui: assistants chat (#1151) (04ae4b0)
- ui: bullets item styling (b6fff15)
- whisper: failing gpu pod due to mangled env variable (bb5e58c)
Miscellaneous
- add Made for UDS badge verification task and workflow (#1149) (228c55c)
- api, tests: refactor api tests to centralize file pathing (#1165) (e2ce0f4)
- deps: update actions/checkout action to v4.2.0 (#1059) (fcd8087)
- deps: update actions/upload-artifact action to v4.4.0 (#1133) (e2bd44f)
- deps: update dependency defenseunicorns/uds-cli to v0.16.0 (#1139) (c3824c2)
- deps: update dependency grpcio to v1.66.1 (#1145) (cd5ca6f)
- deps: update dependency huggingface-hub to v0.25.1 (#1189) (8129c34)
- disables renovate and cron-job workflows (#1202) (d1e26df)
- docs: add workflow badges to readme (#1152) (6880569)
- release 0.14.0 (#1203) (b95dd44)
- test, release: nightly snapshot release and upgrade testing (#1141) (b43c30c)
- ui: better assistant error handling (#1158) (0596115)
- ui: bump @sveltejs/kit and fix e2es (b9f6413)
- ui: style updates part1 (#1168) (bab2dcd)
- update registry1 weekly to 0.13.1, API, and testing workflows (#1101) (022e98d)
v0.13.1
0.13.1 (2024-09-26)
Features
Bug Fixes
- fix indientation in text-embeddings and whisper deployments (#1155) (c47add1)
- whisper: fix whisper transcription of non-english audio (#1066) (8dd467a)
Miscellaneous
- deps: update actions/setup-node action to v4.0.4 (#1060) (69e7358)
- deps: update actions/setup-python action to v5.2.0 (#1132) (5f58be0)
- deps: update davidanson/markdownlint-cli2-action action to v17 (#1112) (438553e)
- deps: update defenseunicorns/uds-common action to v0.13.1 (#1134) (3150873)
- deps: update defenseunicorns/uds-common digest to 24c8a2a (#1058) (605b942)
- deps: update dependency @ai-sdk/openai to ^0.0.61 (#1061) (78222dd)
- deps: update dependency aiostream to v0.6.2 (#1136) (e0845ab)
- deps: update dependency anthropic to v0.34.2 (#1069) (0e08480)
- deps: update dependency ctranslate2 to v4.4.0 (#1137) (3fa92ea)
- deps: update dependency eslint to ^8.57.0 (#1070) (a51ed24)
- deps: update dependency instructor to v1.4.3 (#1071) (7c9d855)
- deps: update dependency markdownlint-cli to v0.42.0 (#1111) (ae5af5d)
- deps: update dependency numpy to v1.26.4 (#1092) (2ca0b70)
- deps: update dependency playwright to ^1.47.0 (#1093) (28898ff)
- deps: update dependency postgrest to v0.16.11 (#1096) (41fea26)
- deps: update dependency python-multipart to v0.0.10 (#1097) (7e6027e)
- deps: update dependency storage3 to v0.7.7 (#1099) (fb2b437)
- deps: update dependency tqdm to v4.66.5 (#1100) (be5bf2c)
- deps: update dependency unstructured to v0.15.13 (#1102) (590e946)
- release 0.13.1 (#1157) (a0d9bc0)
- ui: improve e2e playwright test stability (#1116) (c4c7e9d)
- ui: streaming text response deprecated (#1104) (b360d15)
v0.13.0
0.13.0 (2024-09-20)
⚠ BREAKING CHANGES
- add containerization and packaging manifest lints (#937)
- uds bundle refs for upstream flavor and namespace (#977)
Features
- add LLM as judge evaluations (#960) (3e5f1e0)
- api,vllm,llama-cpp-python: add token counting feature with gRPC support (#964) (e45caef)
- ui: summarization (#1087) (ee0d8d8)
- ui: transcription (#1062) (8e1641d)
- ui: translation (#1046) (d954c33)
Bug Fixes
- api, tests: configmap reloading core dump (#1085) (4a19c8e)
- make: make clean changed to make clean-artifacts (#1073) (480a35b)
- sdk: best_of in completions was an int and should be a string (#1031) (4e5c1b6)
- ui: broken playwright authentication for cleanup (#1016) (6d17187)
- ui: delete avatar (#978) (3ddb98c)
- ui: header flicker and hamburger on thread page (#1024) (84a6d70)
- ui: small screen chat page style (#1025) (8fee760)
- ui: temperature slider (#1086) (f7a37c4)
Miscellaneous
- add containerization and packaging manifest lints (#937) (c4d0835)
- api: update langchain to 0.2.12 (#1048) (3994547)
- deps: ensure a dependency workflow checks all upstream resources and dependencies (#1042) (c37c698)
- new 'types' module (#1009) (a6ad11c)
- release 0.13.0 (#1088) (a343a8d)
- sdk, api: fix types, enhance chat, audio and completion tests (#1038) (014329c)
- split up make clean target (#1029) (2d9cff2)
- test: fix and re-activate integration and unit tests (#1004) (d32bd72)
- test: prevent E2E tests from running in draft PRs (#1015) (11526e8)
- uds bundle refs for upstream flavor and namespace (#977) (caf4f9c)
- ui: html/markdown security (#1006) (6a0038a)
- ui: package updates (8fbe561)
- variablize uds distro for uds make target (#1065) (eb35830)
v0.12.2
v0.12.1
v0.12.0
0.12.0 (2024-09-06)
⚠ BREAKING CHANGES
- standardize flavors, bundle dir and ghcr namespace (#933)
- api: initial integration of registry1 api flavor (#920)
Features
- add RAG Evals Runner and Needle in a Haystack Evaluation (#945) (c191b54)
- api: adds migration for tracking vector store indexing status (#830) (eee3ed7)
- api: initial integration of registry1 api flavor (#920) (a8e93fb)
- ui: chat completion with files (#951) (020dc07)
- ui: persist file upload status (#967) (cb0650d)
- ui: refresh token early (#922) (4f0449d)
Bug Fixes
- api: search vectorstore using only last message (#939) (8a1d61e)
- correct fsGroup configuration in various deployments (#974) (86766db)
- run on thread containing an assistant without files (#995) (1072654)
- ui: csp prevents iframe content from showing (ff71c56)
- ui: file upload data shows 1970 (#941) (1a80646)
- ui: playwright logout causes others to fail (#923) (464ddbd)
- vllm: initializes dictionary entry prior to usage (#959) (d0e09cf)
Miscellaneous
- add logger configuration to our various main entrypoints (#918) (03bf0fd)
- api: add background task for processing vectors (#942) (5f259f1)
- commitlint and labelint workflows (#931) (186a92c)
- deps: pin all python deps and standardize pyprojects (#929) (af74924)
- docs: markdown linting and improved documentation (#902) (046a466)
- pin python dependencies in root pyproject (#958) (cef1535)
- release 0.12.0 (#993) (547988f)
- rename indexing_status migration version (#996) (c39ff94)
- standardize flavors, bundle dir and ghcr namespace (#933) (985642a)
- supabase: enable realtime for the file_objects table (#985) (38e9705)
- supabase: fix supabase-realtime (#855) (b1ee076)
- tests: Conformance testing templates (#910) (b9f8d6a)
- ui: add test for assistant with no files (#997) (d9b9361)
- ui: use keycloak for e2es in workflow (#909) (cdd3e61)
- Update CODEOWNERS (#971) (ae203b6)
- update release workflow trigger type (#988) (ce4952b)
v0.11.0
0.11.0 (2024-08-16)
Features
- ui: 663- skeleton text for assistant responses (#898) (a4b0e1b)
- ui: 669-annotations (#908) (9d14dbf)
- ui: download files (0ff7aa7)
- ui: settings/profile header btn re-design (941da3a)
Bug Fixes
- api: Adds migration to prevent null metadata messages (#895) (44aafca)
- api: Fix invalid nltk reference and bug (#914) (6e3e5b8)
- api: Updates unstructured to fix punkt_tab issue (#905) (67a05bb)
- batch requests to the CreateEmbedding stub (#887) (8ed4328)
- Remove explicit requests install (#906) (863e9d6)
- ui: CSP (#913) (b83ef04)
Miscellaneous
- add default GH PR workflow triggers to test workflows (#892) (c418a4c)
- Bumps all the uds core versions to 0.25.0 and updates deployment docs (#874) (6ca39f1)
- Remove "review_requested" as trigger to prevent unintentional test re-runs (4c20d9e)
- ui: content security policy (#903) (c2d2262)
- ui: improve reliability of playwright test in workflow (29703cb)
- ui: improve sveltekit store mocking (90bef05)
- ui: typescript updates and fix rag e2e tests (#883) (8cd3011)
v0.10.0
0.10.0 (2024-08-02)
⚠ BREAKING CHANGES
- api: updating api endpoints (#817)
Features
- backend: add k3d gpu image builder (#797) (4504085)
- backend: nvidia runtimeclass (#787) (106997d)
- silent parallel make targets for build and deployment (#824) (681aafd)
- ui: remove carbon and replace with Flowbite (#862) (921a864)
Bug Fixes
- api: fix indexing files with api key auth (#852) (c4d9c3f)
- helm template evluation for whisper GPU_REQUEST envvar (#859) (5320890)
- make errors when supabase is unavailable (#814) (976635c)
- ui: logout (#849) (f71af5d)
- ui: playwright login without keycloak (#833) (fd1e3dd)
- whisper: Including missing cuda dependencies required for GPU runtimes (2aba4af)
Miscellaneous
- api: Adds API load testing (#801) (67b9755)
- api: Indexing performance improvements (#799) (e679ad2)
- api: updating api endpoints (#817) (6ff292f)
- deps: bump torch from 2.1.2 to 2.2.0 in /packages/text-embeddings (#831) (22c75cb)
- make python test dependencies optional (#815) (89ff0a6)
- optimize vLLM Dockerfile to reduce layer sizes (#805) (0fec864)
- release 0.10.0 (#864) (7e6f574)
- Remove model weights from container images (#786) (33e4efb)
- restrict daemonset to cuda compute, utility (#836) (4bf9124)
- set kong service to default to cluster IP (#857) (08f1d10)
- split e2e tests into multiple workflows (#808) (c993ad5)
- Update defenseunicorns/zarf to zarf-dev/zarf (#832) (cc18cea)
- update release workflow to install necessary python deps (#867) (1e667a4)
- whisper: Pass through variables down to whisper (#840) (4e8092a)