Fix raptor resuable issue. (#4063)

Upgrades Document Layout Analysis model. (#4054)

Cache the result from llm for graphrag and raptor (#4051)

Fix chunk number error after re-parsing. (#4043)

Fix bugs in agent api and update api document (#3996)

Replaced md5 with xxhash64 for chunk id (#4009)

Try to reuse existing chunks (#3983)

Support iframe chatbot. (#3961)

organize chunks by document in the prompt (#3925)

api http return error (#3941)

Fix session API issues. (#3939)

Fix error message for image access. (#3936)

Add support to iframe chatbot (#3929)

Fix errors detected by Ruff (#3918)

Refactor UI text (#3911)

Fix: Fixed the issue that the agent list page failed to load #3827 (#3902)

Add api for list agents and agent seesions (#3835)

Refactor embedding batch_size (#3825)

Add pagerank to KB. (#3809)

Fix @ in model name issue. (#3821)

Fix the bug that the agent could not find the context (#3795)

Set Log level by env (#3798)

Fixed increase_usage for builtin models (#3748)

Feat: Scrolling knowledge base list and set the number of entries per page to 30 #3695 (#3712)

Revert "Feat: Scrolling knowledge base list #3695" (#3708)

Feat: Scrolling knowledge base list #3695 (#3703)

Fix bugs (#3535)

add input variables to begin component (#3498)

Add api for sessions and add max_tokens for tenant_llm (#3472)

Fix logs. Use dict.pop instead of del. Close #3473 (#3484)

Introduced beartype (#3460)

handle_task catch all exception (#3441)

Move settings initialization after module init phase (#3438)

refine the message of queuing a task (#3437)

Update version display on web UI (#3405)

Use consistent log file names, introduced initLogger (#3403)

fix bug about fetching knowledge graph (#3394)

adapt to lower case cohere (#3392)

Turn the relative date in question to absolute date (#3372)

Rework logging (#3358)

Fix bugs in api and add examples (#3353)

Integration with Infinity (#2894)

Miscellaneous edits to RAGFlow's UI (#3337)

Fix:#3230 When parsing a docx file using the Book parsing method, to_page is always -1, resulting in a block count of 0 even if parsing is successful (#3249)

add inputs to display to every components (#3242)

Remove dead code (#3210)

add keyword extraction time elapsed to the little lamp. (#3207)

add assistant to canvas chat history (#3201)

raise exception if embedding model not found (#3199)

add agent completion API (#3192)