0.0.54 (2024-12-18)
- adapters: add embedding support for Groq (4673b5e), closes #176
- adapters: add embedding support for IBM vLLM (#251) (2925dfc)
- adapters: add presets for Granite 3.1 in WatsonX and IBM vLLM (#250) (972681f)
- adapters: extends Ollama embedding options (d3c9364), closes #176
- adapters: extends OpenAI embedding options (ff96251), closes #176
- agents: extend bee start event (#224) (368aa2a)
- agents: granite 3.1 support (#257) (56045d6)
- agents: improve streamlit agent prompt (#256) (747a052)
0.0.53 (2024-12-13)
- agents: improve parsing in Streamlit agent (686b59b)
- agents: update Bee system prompt (#249) (4fff94a)
- tools: prevent python tool use for summarization (#247) (0e69d11)
- tools: propagate agent's runner memory to tools (#242) (0407c66)
- agents: granite runner fixes, docs and examples updates (#243) (6d0c7c5)
- agents: use the userEmpty template for an empty input in Bee (637368d)
- instrumentation: assert raw prompt (#248) (24546c9)
- llms: add missing events for stream method (9a82d29)
- tools: verify that the end date is later than the start date in OpenMeteo tool (#244) (eee4cb5)
0.0.52 (2024-12-10)
- adapters: implement embed (#210) (84a3bf4)
- agents: update streamlit agent prompt (#240) (b73a2e5)
- agents: update streamlit agent prompt to write async code (#239) (2bd9c39)
- cache: propagate self to custom cache keys (85a27d2)
- tools: update cache policies (9613472)
- agents: connect RePlan emitter to the root one (7f79d5f)
- cache: retrieve descriptors for inherited members (df52337)
- feat(adapters): add llama3.3 template (#235) (2778a9a), closes #235
- feat(adpters): add ollama structured output and version retrieval (#237) (821364e), closes #237
- feat(agents): add experimental RePlan (#236) (3b5c20b), closes #236
- feat(agents): base updates (8b740d4)
- feat(internals): handle async middlewares/observes (b802216)
- chore(agents): move RePlan example (29fbcf0)
- chore(llms): return driver response type (28293a6)
- test: log tool errors (99e5f49)
- test: remove dummy value (55847e5)
- fix: auto install pre-commit hooks (b5c4e64)
- feat(llms)!: extend driver response type (16d4bfd)
- chore: update .env.template (36a4a83)
- chore(agents): remove optional type for streamlit (d552506)
- chore(deps): upgrade dependencies (a3352d9)
- chore(examples): update bee advanced (10314af)
- feat: (instrumentation): increase timestamp precision (#234) (72c8717), closes #234
- feat(agents): update Thought in Bee agent system prompt (#221) (cdb3731), closes #221
- docs(agents): add granite bee documentation (#229) (9199ff5), closes #229
0.0.49 (2024-12-05)
- adapters: propagate WatsonX API errors (46dc651)
0.0.48 (2024-12-04)
- adapters: add watsonx debug example (a737327)
- agents: update system prompt for Bee (#199) (a38804d)
- agents: use prompt templates from a concrete runner (#223) (b868974), closes #219
- internals: extends RestfulClient by emitting events (e7a95c0)
- tools: rename CustomToolEmitter to ToolEmitter (a22a4d9)
0.0.47 (2024-12-03)
- agents: set auto retry for Bee as a default behaviour
- agents: extends the error-related context of LinePrefixParser (740f1ef)
- agents: set auto retry for Bee as a default behaviour (04b5c45)
- example agent using IBM Granite LLM (#213) (9745920)
- internals: update type for context in FrameworkError (6d457a1)
0.0.46 (2024-12-03)
- llms: make "options" parameter for generate/stream always partial
- tools: make run options always partial with fallback to an empty object
- tools: improve type support for events (#209)
- adapters: add Azure OpenAI LLM adapter (#201) (a6a0653)
- llms: make "options" parameter for generate/stream always partial (20fbe71)
- tools: improve type support for events (#209) (456ff11)
- tools: make run options always partial with fallback to an empty object (ff65e0c)
- adapters: handle undefined values for CD in BAM/vLLM (0f45b64)
0.0.45 (2024-12-03)
- agents: add retry for parser errors (#204) (226aaf5)
- llms: improve type support for events (74713a6)
- adapters: vLLM JSON Schema guided decoding (#207) (7d83d21)
- agents: propagate groupId to events (e63bc5b)
0.0.44 (2024-11-29)
- agents: secure thought single-line format via system prompt (#198) (e67dabe)
- agents: update constraint decoding for Bee (41ea34f)
- tools: add tool for Milvus (Vector DB) (#188) (bf07a46)
- adapters: gcp vertexai llm adapter parameters (#194) (54819bf)
- agents: handle nodes without newlines in PrefixParser (bb5e3df)
- tools: update ES Tool typings to work with newer versions (cf19ba0)
0.0.43 (2024-11-26)
- adapters: add GCP VertexAI LLM adapter (#164) (d2eb5b3)
- agents: generalize BeeAgent (#193) (f6fb4d8), closes #178
- adapters: correctly format messages in OpenAI adapter (6728ef6)
- tools: run's method return type (623dbe1)
0.0.42 (2024-11-21)
0.0.41 (2024-11-19)
- adapters: add AWS Bedrock LLM provider (#168) (23656bd), closes #122
- adapters: add llama-3-1-8b to vllm, remove qwen2 templates (#172) (b333594)
- agents: serialize isRunning state (f5b7abe)
- internals: add recursiveSplitString and mergeStrings utility functions (d82be6e)
- internals: make findFirstPair handles longer sequences (c00efbe)
- tools: make tools composable (#169) (33f1db5), closes #166
- tools: reduce number of iterations needed for SQLTool (#174) (cc221a3)
0.0.40 (2024-11-14)
- agents: improve constraint decoding regex for granite agent (cc6a32c)
- tools: correctly infer input type for DynamicTool (90db143)
0.0.39 (2024-11-13)
- adapters: load watsonx region from env (7ab61f0)
- agents: granite agent (#159) (102b9be), closes #145
- agent: use constraint decoding for tool names in Bee (#161) (ad55531)
- observe: implement OpenTelemetry instrumentation (2cc1ef4)
- tool: add elasticsearch (#138) (31c3559)
- tools: openMeteo optimization (#155) (c228ee2), closes #110
- llms: correctly handles a signal abortion (1b0d514)
0.0.38 (2024-11-08)
- agent: improve Bee extensibility, update events (705f142)
- agent: improve extensibility of Bee parser (b5f3b7c)
- memory: add splice and remove method (3dd1535)
- tool: handle duckduckgo anomaly detection (41a1048)
0.0.37 (2024-11-07)
- agent: remove tool caption from Bee agent (#154)
- tool: remove pagination functionality from GoogleSearch and DuckDuckGo (#152)
- tool: remove pagination from search tools
- agent: remove tool caption from Bee agent (#154) (d84beeb), closes #153
- llm: add Granite 3.0 models to WatsonXChatLLM Preset (1e6aa37)
- llm: set temperature to 0 for OpenAI LLM adapter (ea84808)
- template: make validateInput method public (34156b7)
- tool: add public parse method (f97f73a)
- tool: improve wikipedia results filtering (#143) (2529a0c), closes #142
- tool: remove pagination from search tools (f93d181)
- tool: remove pagination functionality from GoogleSearch and DuckDuckGo (#152) (59424de), closes #151
0.0.36 (2024-11-04)
- agent: remove unused agent manager (b22b1d0)
- agent: update custom agent example (e942b18)
- llm: update the granite chat template for tools (#125) (b554031)
- tool: add optional input preprocessor (#128) (cec2cb9)
0.0.34 (2024-10-28)
- add additional validation for DynamicTool (#100) (b69d64c)
- agent: make bee runner easier to override (e7b99a0)
- agent: make bee runner serializable (c8960ce)
- llm: export templateSchemaFactory (8fce700)
- llm: set ollama host from env (4712e87)
- tools: improve python PDF capabilities (#108) (9e107d7)
- examples: update open library tool (9054deb), closes #104
- GH: update CODEOWNERS (#109) (3ea6fe5)
- serialization: handle constructor references (8a9a91b)
0.0.33 (2024-10-19)
0.0.32 (2024-10-16)
0.0.30 (2024-10-15)
- add info on running with remote ollama (#83) (ed6405d)
- agent: ignore newlines for a tool name (a2c6202)
- serialization: handle bounded functions (58d072f), closes #92
0.0.29 (2024-10-15)
- agent: add endOnRepeat option to the LinePrefixParser (de24beb)
- agent: add waitForStartNode option to the LinePrefixParser (a27a60a)
- agent: update parser settings (42027e2)
- agent: update parser settings (fdfca98)
- agent: update SQL tool example (325e1ed)
- code-interpreter: improve file input prompt (#86) (e0c5985)
- code-interpreter: remove functionality (739036c)
- code-interpreter: remove functionality (18302f9)
- code-interpreter: rename "hash" to "pythonId" (#90) (c25cbc5)
- examples: replace process exit with reader close (65b7da4)
- llm: update credentials loading for watsonx (44a79d3)
- llm: update default model for groq adapter (67c3270)
- serialization: allows to pass extra classes to bootstrap (60f8953)
- internals: fix race condition in emitterToGenerator (dfe63f8)
- llm: handles OpenAI structured output (23e4e0e)
- memory: make shallow copy of handlers (be82ed5)
- serialization: properly deserialize arrow functions (25234df)
0.0.28 (2024-10-10)
- agent: update system prompt (#76) (46e45b7)
- improve Python tool (#79) (feb2f4a)
- internals: createURLParams handles nested objects (2f0ec77)
- tools: propagate run context to the CustomTool (b3cc10a)
- agent: LinePrefixParser - handling termination nodes (2a5ce16)
- llm: templates serialization (d7584ef), closes #78
0.0.27 (2024-10-08)
- agent: improve time awareness (89e8f0a)
- agent: update system prompt and open meteo schema (#72) (c235522)
- memory: add iterator (23bba65)
- memory: rename SlidingWindowMemory, add removalSelector functionality (69702b6)
- memory: switch from LLM to ChatLLM in SummarizeMemory (022c05a)
- serializer: add aliases support (72a6640)
- tools: auto coerce input for google search (7236ba7)
0.0.26 (2024-10-07)
- cache: add getter for fullPath in FileCache (d2beaea)
- cache: improve fileCache typings (4c22dae)
- llm: add caching support (e330fb2), closes #69
- tools: add type coercion to the schema (f0995ea)
- tools: update duckduckgo defaults propagation (9ef508c)
0.0.25 (2024-10-04)
- agent: simplify tools' schema that is passing to prompt (c927201)
- agent: update CD regex, update tools description (#60) (50f86e6)
- code-interpreter: update to a new version (5357215)
- code-interpreter: update to a new version (699e45f)
0.0.24 (2024-10-03)
- agent: update line prefix parser fallback and add tests (6932379)
- agent: use fallback only if LLM output is not empty (a6e582d)
- llm: update ollama default parameters (a9dc1a9)
0.0.23 (2024-10-03)
- agent: add fallback for invalid LLM output (5ac89c1), closes #55
- agent: improve LinePrefixParser error messages (710f75b)
- agent: update bee system prompt (#56) (08c4714)
- agent: update bee system prompt (#58) (572572b), closes #55
- tools: improve Wikipedia parsing (#57) (513572f)
- llm: add missing protos to build ibm-vllm (#52) (6fe7bca)
- llm: add missing run context parameter for ibm-vllm (#54) (9ee7957)
- llm: bam chat preset typo (93b3167)
- llm: watsonx chat preset typo (f7d5128)
0.0.22 (2024-10-02)
- agent: prevent emitting tool output update during streaming (94ab4fe)
- agent: update bee system prompt, tools description (#48) (83fdb26)
- ibm-vllm: add llm adapter (ec4e455)
- ibm-vllm: update build process (4a73ac0)
- tools: wikipedia - remove records that did not pass the filtering criteria (46ce792)
0.0.21 (2024-10-02)
- agent: extend line prefix parser error messages (20efff2)
- code-interpreter: update configuration and readme (70280a9)
- code-interpreter: update to 0.0.11 (#41) (f136a09)
- llm: add llama3-8b to BAM chat preset (8f6be11)
- serializer: supports serialization for class with the same name (11cab93)
- tools: add langchain adapter (332edce), closes #42
- tools: rename GoogleCustomSearch to GoogleSearch (087351f)
- tools: update abort signal propagation (f718236)
0.0.20 (2024-09-27)
- code-interpreter: update to 0.0.7 (#37) (0392ff0)
- tool: add google custom search tool (#34) (ef839da)
- tools: update Wikipedia tool, remove links, extend interface (ee651c3)
- agent: update constraint decoding regex (8d47cae)
0.0.19 (2024-09-27)
- agent: add generic line prefix parser (#35) (2c39c6a)
- agent: add tools property to agent's metadata (59036b1)
- agent: improve invalid json parsing (da7c7e6)
- agent: improve parser typings (a86d85d)
- agent: switch to a new system prompt (#38) (9f41663)
- tool: add sql tool (#24) (719d80d)
- tool: OpenMeteo - add optional country parameter (a3612a5), closes #36
- tools: update Python tool description (b1f6104)
- agent: line prefix parser types (699ccac)
- llm: watsonx serialization (b1caa9f)
- tool: OpenMeteo - handle empty geocode result (424e620), closes #36
0.0.18 (2024-09-24)
- agent: update tool error handling (d403665)
- cache: extend instance abilities (cb7ccc9)
- llm: add bam adapter verbose example (7db2835)
- llm: add fromTemplate factory to the JSON driver (6d3b356)
- llm: add types for watsonx llm inference parameters (312626d)
- llm: make drivers serializable (05d6dbb)
- llm: set default parameters for groq (82daecd)
- llm: set default parameters for ollama (0eecb65)
- casting boolean envs (5994152)
0.0.17 (2024-09-16)
- agent: update prompt template schemas (5373835)
0.0.16 (2024-09-16)
- llm: export messagesToPromptFactory function (d534012)
- move llm drivers under the llm module (8ea5136)
- tool: update open meteo schema (8736e21)
0.0.15 (2024-09-13)
- agent: disable repetition checker by default (0a5f8ac)
- agent: parser ignores empty chunks (cbceb87)
- agent: propagate python file identifier to the system prompt (334e90f)
- groq: add llm adapter (#20) (5abb614)
- llm: add drivers (f38cee4)
- watsonx: add llama3.1-405b to the preset (8a93496)
- agent: update guided setting (19f11cf)
0.0.14 (2024-09-11)
- agent: add createdAt to newly created messages (9313f52)
0.0.13 (2024-09-11)
- agent: inject metadata to messages (a10d432)
- agent: make repetition checker disabled by default (4bc3a60)
- templates: add schema validation (5a5114d)
- templates: add support for defaults (8be1adb)
- tool: update openMeteo tool (5726691)
0.0.12 (2024-09-09)
- agent: add memory getter to the base class (bf985d1)
- agent: add runner config (36eff24)
- agent: premature stop on sequence repetition (d54eb2a)
- observability: propagate current instance to async scope (085d69a)
- bee-agent: rendering of empty lines in prompt (#18) (8aad96e)
- llm: stream method (9e47f1a)
- utils: prevent unhandledRejection (f74a4f5)
0.0.11 (2024-09-06)
- agent: unify user's input formatting (b6e4b02)
- agent: update runner memory setting (adb6a7e)
- memory: add sync mechanism to the TokenMemory (63975c9)
0.0.10 (2024-09-06)
- agent: allow to run with an empty prompt (8622e2d)
- execution: propagate runParams to RunContext (925f7d8)
- tool: add calculator tool (#10) (2af8af7)
- tool: update openMeteo default settings (ad4bb21)
0.0.9 (2024-09-04)
- agent: add template for tool not found error (c4ea1c8)
- custom-tool: add ssl support for code interpreter (#14) (7de84d1)
- example: add bee advanced example (775c21c)
- examples: use bee-agent-framework in imports (50f7d72)
- template: add support for adding mustache functions (6cda741)
- tool: improve arxiv (d6df723)
- utils: change ajv coerceTypes setting (f3489b1)
- watsonx: make second parameter in preset required (6d45901)
- agent: make templates property partial (01e63e4)
- agent: prevent wrapping the same error twice (19328cb)
0.0.8 (2024-09-03)
- agent: add support for overriding templates (58c89d9)
- tool: improve arxiv error handling (d4faf00)
- tool: stop using getters in custom tool (0e25450)
- tool: handle default values in JSON/zod schema (777c684)
0.0.7 (2024-09-02)
- agent: group iterations, add meta property (ec9fb96)
0.0.6 (2024-09-02)
- agent: extends bee success event (889da33)
- react: emit
toolErrorevent for an invalid input (cd33204) - remove modules index file (6c5f045)
- tools: update python tool output file prefix (716e74d)
0.0.5 (2024-08-30)
- bam: refactor chat preset (5f5fa05)
- python: add custom id support (d23d4f8)
- watsonx: add chat preset (f4797b8)
- watsonx: auto load accessToken/apiKey from env (369da92)
- example: fix code interpreter path (7f2f466)
0.0.4 (2024-08-29)
0.0.3 (2024-08-28)
- bam: add llama3 to preset (5d8df8c)
- react: improve max retries error message (54895f9)
- update .env.template (7ebad1f)
- env script handle existing file (51961af)
0.0.2 (2024-08-27)
- python: add ignoredFiles support to local storage (4c65b88)
- keep code interpreter related folders in examples (cc8ca40)
- watsonx: tokenization (399d8cb), closes #3
0.0.1 (2024-08-23)
- init project (7ebb60a)
0.0.2 (2024-08-27)
- python: add ignoredFiles support to local storage (4c65b88)
- keep code interpreter related folders in examples (cc8ca40)
- watsonx: tokenization (399d8cb), closes #3
- init project (7ebb60a)