* Update AzureChatOpenAI.ts - corrected reasoning description and default
- Description for reasoning effort only mentioned o1. Added o3.
- Changed reasoning effort default to medium as this is OpenAI's default / what users will most likely expect (https://platform.openai.com/docs/guides/reasoning)
* Update ChatOpenAI.ts - corrected reasoning description and default
- Description for reasoning effort only mentioned o1. Added o3.
- Changed reasoning effort default to medium as this is OpenAI's default / what users will most likely expect (https://platform.openai.com/docs/guides/reasoning)
* Update models.json - add specific model ID for o3-mini
- Added o3-mini-2025-01-31
- Updated "o3-mini" label to "o3-mini (latest)"
This allows the user to choose a specific model ID and avoid the risk of unexpected behavior if the "o3-mini" alias is updated.
* Late Chunking support for Jina Embeddings
* improved constructor handling. remove redundant code.
* remove unnessary function overrides, since late_chunking parameter only needs to be stored in the class.
* Expose dimensions parameter on Jina Embeddings
* fix linting issue
* update description text to refer to dimensions
* Update JinaAIEmbedding.ts
---------
Co-authored-by: Henry Heng <henryheng@flowiseai.com>
The previous code incorrectly assigned `toolInput` to an empty string when `customInput` was non-empty. Corrected the logic to ensure that `toolInput` is assigned `customInput` when `useQuestionFromChat` is false and `customInput` is non-empty.
* added: Neo4j database connectivity, Neo4j credentials, supports the usage of the GraphCypherQaChain node and modifies the FewShotPromptTemplate node to handle variables from the prefix field.
* Merge branch 'main' of github.com:FlowiseAI/Flowise into feature/graphragsupport
* revert pnpm-lock.yaml
* add: neo4j package
* Refactor GraphCypherQAChain: Update version to 1.0, remove memory input, and enhance prompt handling
- Changed version from 2.0 to 1.0.
- Removed the 'Memory' input parameter from the GraphCypherQAChain.
- Made 'cypherPrompt' optional and improved error handling for prompt validation.
- Updated the 'init' and 'run' methods to streamline input processing and response handling.
- Enhanced streaming response logic based on the 'returnDirect' flag.
* Refactor GraphCypherQAChain: Simplify imports and update init method signature
- Consolidated import statements for better readability.
- Removed the 'input' and 'options' parameters from the 'init' method, streamlining its signature to only accept 'nodeData'.
* add output, format final response, fix optional inputs
---------
Co-authored-by: Henry <hzj94@hotmail.com>
* feat: Enhance ConditionAgent with conversation history selection options
- Added a new parameter `conversationHistorySelection` to allow users to choose which messages from the conversation history to include in prompts.
- Options include: User Question, Last Conversation Message, All Conversation Messages, and Empty.
- Default selection is set to 'All Conversation Messages' for improved context management in sequential LLM and Agent nodes.
* Bump version from 2.0 to 3.0
* Update 'Require Approval' button description to clarify dependency on MemoryAgent
* Refine RequiredApproval message description
* Fix message description by removing unscaped character ('\')
* Break line
* docs: clarify and enhance the "Require Approval" description for Agent node
* chore: slight update of the description
* feat: Add option to disable conversation history
- Add new `disableConversationHistory` boolean parameter in LLMNodes.ts and Agent.ts to optionally skip including conversation history in prompts
- Fix potential error in Agent.ts when messages array is empty by adding null safety checks
- Improve memory efficiency by allowing stateless interactions when history isn't needed
* feat: add conversation history filtering options
Replace the disable conversation history feature with a more flexible filtering system that allows selecting:
- User question only
- Last message only
- All messages (default)
- No messages
This provides more granular control over conversation context management.
* chore: break lines
* chore: removed ending semi columns
* chore: fix eslint errors
* fix(sequentialagents): improve conversation history filtering logic
- Remove unnecessary state.messages check for user_question case
- Add proper null handling for last_message and all_messages cases
- Remove @ts-ignore comments with proper typing
* Update LLMNode.ts
* Update Agent.ts
---------
Co-authored-by: Henry Heng <henryheng@flowiseai.com>
Ali Saleh