v0.17.0

Summary

• AI chat: Implements Deep Research for agentic reasoning. To activate this, enable the Reasoning toggle under the Prompt Engine tab of your chat assistant dialogue.
• AI chat: Leverages Tavily-based web search to enhance contexts in agentic reasoning. To activate this, enter the correct Tavily API key under the Assistant Setting tab of your chat assistant dialogue.
• AI chat: Supports initiating a chat without specifying knowledge bases.
• AI chat: HTML files can also be previewed and referenced, in addition to PDF files.
• Dataset: Adds a Document parser dropdown menu to dataset configurations. This includes a DeepDoc model option, which is time-consuming, a much faster naive option (plain text), which skips DLR (Document Layout Recognition), OCR (Optimal Character Recognition), and TSR (Table Structure Recognition) tasks, and several currently experimental large model options.
• Agent component: (x) or a forward slash / can be used to insert available keys (variables) in the system prompt field of the Generate or Template component.
• Object storage: Supports using Aliyun OSS (Object Storage Service) as a file storage option.
• Models: Updates the supported model list for Tongyi-Qianwen, adding DeepSeek-specific models; adds ModelScope as a model provider.
• APIs: Document metadata can be updated through an API.

What's Changed

• More models for siliconflow. by @KevinHuSh in #4756
• Config chat share by @isthaison in #4700
• Fix Gemini chat issue. by @KevinHuSh in #4757
• optimize TenantLLMService.increase_usage for "can't update token usag… by @DavidSche in #4755
• Refresh Gemini model list. by @KevinHuSh in #4780
• Add a comment to valkey. by @KevinHuSh in #4783
• Fix divided by zero issue. by @KevinHuSh in #4784
• Increase ES update script length. by @KevinHuSh in #4785
• Fix: The requested interface timeout will cause the page to crash #4787 by @cike8899 in #4788
• Inner prompt parameter setting. by @KevinHuSh in #4806
• Feat: Add VariablePickerMenuPlugin to select variables in the prompt text box by menu #4764 by @cike8899 in #4765
• Feat: Fixed the issue where the prompt always displayed the initial value when switching between different generate operators #4764 by @cike8899 in #4808
• Feat: Remove begin's width from agent templates #4764 by @cike8899 in #4809
• Fix KG circle. by @KevinHuSh in #4823
• Fix: Fail to open console with Firefox #4816 by @cike8899 in #4838
• Reformat error message. by @KevinHuSh in #4829
• Feat: Modify the Preset configurations item style to distinguish it from other fields #4844 by @cike8899 in #4845
• Added a release notes by @writinwaters in #4848
• Feat: Add ChatCard #3221 by @cike8899 in #4852
• Make the update script shorter. by @KevinHuSh in #4854
• Feat: Add Sessions component #3221 by @cike8899 in #4865
• fix categorize agent input content not format error by @WANGRUI-ZB in #4842
• Feat: Add ChatBasicSetting component #3221 by @cike8899 in #4876
• Feat: Add hatPromptEngine component #3221 by @cike8899 in #4881
• Fix: Knowledge base page crashes when network connection is lost. #4894 by @cike8899 in #4895
• Changed elasticsearch image url by @yuzhichang in #4897
• fix: big xls file error by @SkyfireWXY in #4859
• Remove use of eval() from operators.py by @panzi in #4888
• Fix xinference chat role order issue. by @KevinHuSh in #4898
• Remove use of eval() from search.py by @panzi in #4887
• Display thinking for deepseek r1 by @KevinHuSh in #4904
• Feat: Add LlmSettingFieldItems component #3221 by @cike8899 in #4906
• Fix: After deleting all conversation lists, the chat input box can still be used for input. #4907 by @cike8899 in #4909
• Organized and add a list of large language models of Nvidia.v1.1 by @DiamondPoirier in #4910
• Add a list of large language models of deepseek and image2text models… by @DiamondPoirier in #4914
• Feat: Add ChatInput component #3221 by @cike8899 in #4915
• Fix DB assistant template error. by @KevinHuSh in #4925
• Feat: Display Think for Deepseek R1 model #4903 by @cike8899 in #4930
• Added a guide on setting metadata by @writinwaters in #4935
• Feat: Bind data to datasets page #3221 by @cike8899 in #4938
• Fixes KeyError: 'content' when using stream=False by @Peterson047 in #4944
• Feat: Add an id to the dataset testing route #3221 by @cike8899 in #4951
• Ignore tenant not found error while increasing token usage. by @KevinHuSh in #4950
• Fix hit cache error while raptoring. by @KevinHuSh in #4955
• format number float by @isthaison in #4954
• Feat: Jump from the chunk page to the dataset page #3221 by @cike8899 in #4961
• add gemini-2.0-flash-thinking-exp-01-21 by @isthaison in #4957
• Feat: Add ParsedPageCard component #3221 by @cike8899 in #4976
• Feat: Add background color to GraphRag configuration #4980 by @cike8899 in #4981
• Added PEP 723 metadata to download_deps.py by @yuzhichang in #4988
• Fix in keywords or question auto-generations. by @KevinHuSh in #5021
• Fix knowledge graph node not found (#4968) by @zhxlp in #4970
• Fix keyerror issue while rebuilding graph. by @KevinHuSh in #5022
• Optimize conversation when uploading attachments by @kuschzzp in #4964
• Feat: add LLM provider PPIO by @saikidev in #5013
• Fix window size of ES issue. by @KevinHuSh in #5026
• Rewrite Support specified language or language according to initial question by @isthaison in #4990
• Remove for KG extraction. by @KevinHuSh in #5027
• Fix: Cannot distinguish between export and import icons #5025 by @cike8899 in #5031
• Ignore exception of empty index. by @KevinHuSh in #5030
• Add another decrypt function. by @KevinHuSh in #5043
• Feat: Replace next-login-bg.svg #3221 by @cike8899 in #5046
• Fix: Starting the source code on Windows, the 'HTTP API' returns 404 by @hy89 in #5042
• Ignore exceptions when no index ahead. by @KevinHuSh in #5047
• Update agent session API, to support uploading files while create a new session by @flygithub in #5039
• update locale vi by @isthaison in #5035
• Specify img2text model by tag by @hrchu in #5063
• Remove for exeSql component. by @KevinHuSh in #5069
• add option Embed into webpage by @isthaison in #5065
• Added a graphrag guide by @writinwaters in #4978
• Fix: The max tokens defined by the tenant are not used (#4297) (#2817) by @zhxlp in #5066
• Feat: Add LanguageAbbreviation to simplify language resource files. #5065 by @cike8899 in #5072
• Feat: Extract the common parts of groupImage2TextOptions and groupOptionsByModelType #5063 by @cike8899 in #5074
• Fix: Chunk problem tag content cannot be displayed completely. #5076 by @cike8899 in #5077
• Feat: Add ChunkedResultPanel #3221 by @cike8899 in #5085
• Added an FAQ by @writinwaters in #5092
• Fix: PUT method does not work as expected with Invoke component by @flygithub in #5081
• Feat: Write the thinking style in the MarkdownContent layer #4930 by @cike8899 in #5091
• Fix: Use self.dataStore.indexExist in all_tags method of Dealer by @ubbg in #5108
• Fix docx table issue. by @KevinHuSh in #5117
• Fix too many clause while searching. by @KevinHuSh in #5119
• Fix: Option ineffective in Chat API by @hrchu in #5118
• fix: Ensure that the commands are executed in the correct directory s… by @qiufeihong2018 in #5089
• Feat: Add support for document meta fields update through api by @liwenju0 in #5120
• Remove redandent code. by @KevinHuSh in #5121
• Feat: Allow the Rewrite operator to connect to the Generate operator #1739 by @cike8899 in #5128
• Feat: Support preview of HTML files #5096 by @cike8899 in #5134
• Docs: Add note about docker volume deletion in README files，will be more novice-friendly by @liwenju0 in #5133
• Feat: Add insert variable icon in the header of prompt editor. #4764 by @cike8899 in #5142
• Fix: Improve message input handling with Shift+Enter support by @liwenju0 in #5129
• Iframe should support input variables by @gstrat88 in #5156
• Feat: Show formulas when answering, show reference labels in style, remove cursor flashing effect. #5173 by @cike8899 in #5174
• Fix: Normalize embedding model ID comparison across datasets by @liwenju0 in #5169
• Fix window size issue of ES. by @KevinHuSh in #5175
• Feat: Add reasoning item to chat configuration modal #5173 by @cike8899 in #5177
• Feat: Add agent page. #3221 by @cike8899 in #5179
• Feat: Add SearchPage component. #3221 by @cike8899 in #5184
• Optimized Recognizer.sort_X_firstly and Recognizer.sort_Y_firstly by @yuzhichang in #5182
• Feat: Add AgentTemplates component. #3221 by @cike8899 in #5194
• Apply agentic searching. by @KevinHuSh in #5196
• Fix session.ask return generator bug when stream=False on python sdk by @Vela-zz in #5209
• Support chat solo. by @KevinHuSh in #5218
• Feat: Chat without KB. #5216 by @cike8899 in #5217
• Remove v1 by @KevinHuSh in #5220
• Fix spelling errors by @DavidSche in #5224
• Fix empty question issue. by @KevinHuSh in #5225
• Feat: Add RAGFlowSelect component #3221 by @cike8899 in #5228
• Reuse loaded modules if possible by @yuzhichang in #5231
• Refine search query. by @KevinHuSh in #5235
• Fixed similarity on infinity by @yuzhichang in #5236
• Double check error of adding llm. by @KevinHuSh in #5237
• Added a guide on running a retrieval test, with and without knowledge graph by @writinwaters in #5200
• Support downloading models from ModelScope Community. by @yrk111222 in #5073
• Add configuration to choose default llm models by @Omarlsg98 in #5245
• Add support to boto3 default connection by @Omarlsg98 in #5246
• Feat: add partition of file uploads by @Omarlsg98 in #5248
• Refactor rerank model with dynamic batch processing and memory manage… by @liwenju0 in #5273
• Refine the error message. by @KevinHuSh in #5275
• Refine error message for re-rank model. by @KevinHuSh in #5278
• Let list_chat go without kb checking. by @KevinHuSh in #5280
• Remove content for Generate if it's not stream output. by @KevinHuSh in #5281
• Feat: Add PageHeader to DatasetWrapper #3221 by @cike8899 in #5284
• Tongyi adapts deepseek. by @KevinHuSh in #5285
• Feat: Disable Max_token by default #5283 by @cike8899 in #5290
• Added a prerequisite for ARM platforms by @writinwaters in #5295
• Feat: Add AgentSidebar #3221 by @cike8899 in #5296
• Fix base url missing for deepseek from Tongyi. by @KevinHuSh in #5294
• Optimize ocr by @yuzhichang in #5297
• Added a file size limit by @writinwaters in #5301
• Feat: Render operator menu by category. #3221 by @cike8899 in #5302
• Feat: Render agent details #3221 by @cike8899 in #5307
• Feat: Upload agent file #3221 by @cike8899 in #5311
• Replaced pypi.tuna.tsinghua.edu.cn with mirrors.aliyun.com/pypi by @yuzhichang in #5309
• Refine prompt of agentic search. by @KevinHuSh in #5312
• Feat: Add FormDrawer to agent page. #3221 by @cike8899 in #5323
• Fix: type-script new change by @isthaison in #5159
• Make node merging parallel. by @KevinHuSh in #5324
• Remove dup gb2312, solve currupt error. by @KevinHuSh in #5326
• Fix release.yml by @yuzhichang in #5327
• More robust community report. by @KevinHuSh in #5328
• Feat: Add DynamicVariableForm with shadcn-ui. #3221 by @cike8899 in #5336
• Resolve super class invokation error. by @KevinHuSh in #5337
• Fix: Due to the reference to tailwindcss, the height attribute setting of the image is invalid, resulting in an uneven model list #5339 by @cike8899 in #5340
• Fix: Prevent message sending during IME composition and block new submissions while waiting for a response by @jabee0228 in #5331
• Feat: Add Tavily Api Key to chat configuration modal. #5198 by @cike8899 in #5347
• Add tavily as web searh tool. by @KevinHuSh in #5349
• Fix: 'Chunk not found!' error in team-sharing knowledge base. by @zsdzl93 in #5361
• Fix FlagRerank max_length issue. by @KevinHuSh in #5366
• Added OpenAI-like completion api by @yongtenglei in #5351
• Feat: Add FormSheet. #3221 by @cike8899 in #5377
• Run keyword_extraction, question_proposal, content_tagging in thread pool by @yuzhichang in #5376
• Code refactor by @KevinHuSh in #5371
• Fix：validate knowledge base association before document upload by @liwenju0 in #5373
• DOCS: add OpenAI-compatible http and python api reference by @yongtenglei in #5374
• Fix:HTTP API -> Stop parsing documents(AttributeError: ‘list‘ object … by @k3sc in #5375
• Let parallism of RAPTOR controlable. by @KevinHuSh in #5379
• Feat: add VLLM by @yongtenglei in #5380
• Feat: Add DualRangeSlider #3221 by @cike8899 in #5386
• FIX: correct typo by @yongtenglei in #5387
• Fixed OpenAI compatibility stream [DONE] by @isthaison in #5389
• Feat: Support vLLM #4316 by @cike8899 in #5395
• Fix: correct wrong vLLM rerank model by @yongtenglei in #5399
• Miscelleneous editorial updates by @writinwaters in #5390
• Fix issue of ask API. by @KevinHuSh in #5400
• ci: remove may expand into attacker-controllable code by @yihong0618 in #5407
• Fix: issue of no id for /datasets/<dataset_id>/documents by @KevinHuSh in #5420
• Fix: drop useless ABC method by @yihong0618 in #5408
• Fix: string format error. by @KevinHuSh in #5422
• Fix: add image2text issue. by @KevinHuSh in #5431
• Fix: stream post body by @ssccinng in #5434
• Feat: Wrap DynamicVariableForm with Collapsible. #3221 by @cike8899 in #5440
• Feat: add data type invoke by @isthaison in #5126
• Feat: Accessing Alibaba Cloud OSS with Amazon S3 SDK by @hy89 in #5438
• Feat: Use shadcn-ui to build GenerateForm. #3221 by @cike8899 in #5449
• Fix: fix special delimiter parsing issue by @yongtenglei in #5448
• Feat: If the user is not logged in, jump to the login page by refreshing. by @cike8899 in #5451
• Fix: rm think if stream is Flase. by @KevinHuSh in #5458
• Feat: Modify the parsing method string to an enumeration type. #5467 by @cike8899 in #5468
• Fix: language selection display on the profile settings page by @hjlarry in #5459
• Refa: better service conf by @yihong0618 in #5471
• Refa: drop useless fasttext by @yihong0618 in #5470
• DOCS: amend docker image building page and more hints for mac users by @yongtenglei in #5461
• Fix: add ctrl+c signal for better exit by @yihong0618 in #5469
• Fix: enhance aliyun oss access with adding prefix path by @haifzhu in #5475
• Fix typos by @JinHai-CN in #5476
• Refa: better gitignore by @yihong0618 in #5473
• Fix table parser issue. by @KevinHuSh in #5482
• Fix typo and error by @JinHai-CN in #5479
• Feat: Put the configuration of different parsing methods into separate components. #5467 by @cike8899 in #5487
• Feat: Wrap MaxTokenNumber with DatasetConfigurationContainer. #5467 by @cike8899 in #5491
• Refine error message while embedding model error, by @KevinHuSh in #5490
• Feat: Hide the suffix of the large model name. #5433 by @cike8899 in #5494
• Feat: multiline text input for chat by @Omarlsg98 in #5317
• Feat: Add the Experimental text to the option of the large model of the Image2text type of LayoutRecognizeItem by @cike8899 in #5495
• Moved agent components into the agent folder by @writinwaters in #5496
• minor by @writinwaters in #5497
• DOC: for release. by @KevinHuSh in #5472

New Contributors

• @DavidSche made their first contribution in #4755
• @SkyfireWXY made their first contribution in #4859
• @DiamondPoirier made their first contribution in #4910
• @Peterson047 made their first contribution in #4944
• @zhxlp made their first contribution in #4970
• @saikidev made their first contribution in #5013
• @hy89 made their first contribution in #5042
• @flygithub made their first contribution in #5039
• @ubbg made their first contribution in #5108
• @qiufeihong2018 made their first contribution in #5089
• @gstrat88 made their first contribution in #5156
• @Vela-zz made their first contribution in #5209
• @yrk111222 made their first contribution in #5073
• @jabee0228 made their first contribution in #5331
• @zsdzl93 made their first contribution in #5361
• @yongtenglei made their first contribution in #5351
• @k3sc made their first contribution in #5375
• @yihong0618 made their first contribution in #5407
• @ssccinng made their first contribution in #5434
• @hjlarry made their first contribution in #5459
• @haifzhu made their first contribution in #5475

Full Changelog: v0.16.0...v0.17.0