- [bug fix] Ensure that StopEvent gets cleared from Context._in_progress["_done"] after a Workflow run (#17300)
- fix: add a timeout to langchain callback handler (#17296)
- tweak User vs tool in react prompts (#17273)
- refact: Refactor Document to be natively multimodal (#17204)
- fix: make ImageDocument derive from Document, backward compatible (#17259)
- fix: accept already base64-encoded data in ImageBlock (#17244)
- fix(metrics): fixed NDCG calculation and updated previous tests (#17236)
- fix: remove llama-index-legacy dependency in llama-index-core (#17231)
- Refined the default documentation generation for function tools (#17208)
- add support for voyageai >=0.3.0 (#17120)
- Introducting VoyageAI's new multimodal embeddings model (#17261)
- VoyageAI multimodal embedding, correction (#17284)
- Fixed import errors for experimental JSONalyzeQueryEngine (#17228)
- Add cypher corrector and allow graph schema filtering (#17223)
- Add timeout config to neo4j graph (#17267)
- Add text and embedding type to neo4j enhanced schema (#17289)
- add content blocks to anthropic (#17274)
- Do not send blank content to anthropic (#17278)
- Update anthropic type imports for v0.41.0 release (#17299)
- Fix Anthropic tokenizer protocol (fix by Devin) (#17201)
- Add Amazon bedrock guardrails (#17281)
- Add Amazon bedrock guardrails (#17281)
- Gemini 2.0 support (#17249)
- add tool call id/name to mistral chat messages (#17280)
- Adding llama 3.3-70b as function-calling-capable (#17253)
- fix openai message dicts for tool calls (#17254)
- Fix: TGI context window (#17252)
- handle more response types in anthropic multi modal llms (#17302)
- Support Confluence cookies (#17276)
- Parse "milvus_search_config" out of "vector_store_kwargs" (#17221)
- refactor and optimize milvus code (#17229)
- Handle empty retrieved Pinecone index values (#17242)
- feat: Add NOT filter condition to MetadataFilter and QdrantVectorStore (#17270)
- Add async support to weaviate vector store integration (#17220)
- Refined the default description generation for function tools (#17208)
- fix: relax pin on openai llm dependency (#17210)
- feat: integration on pinecone hosted rerankers (#17192)
- Add Scrapegraph tool integration (#17238)
- Update pgvector dependency to version 0.3.6 (#17195)
- Fix sync and async structured streaming (#17194)
- unpin pydantic to allow 2.8 or greater (#17193)
- Update core structured predict streaming, add ollama structured predict (#17188)
- bump tenacity dependency in llama-index-core (#17178)
- Add Verbose to Vectara
as_query_engine
(#17176)
- Update core structured predict streaming, add ollama structured predict (#17188)
- Fix message format for perplexity (#17182)
- Add possibility to use URI as doc id in WholeSiteReader (#17187)
- BUG FIX: llama-index-vectorstore-chromadb to work with chromadb v0.5.17 (#17184)
- cover SimpleDirectoryReader with unit tests (#17156)
- docs: rewrite openai image reasoning example without multimodal LLM (#17148)
- fix(metrics): fixed NDCG calculation and added comprehensive test cases (#17126)
- feat: improve ImageBlock (#17111)
- Remove forgotten print in ChatMemoryBuffer (#17114)
- [FIX] Move JSONalyzeQueryEngine to experimental (#17110)
- Unrestrict clip models to use (#17162)
- fix/openai-embbeding-retry (#17072)
- proper auth token in TEI (#17158)
- chore: fix httpx_client typo in LlamaCloudRetriever (#17101)
- fix: wrong project id variable in LlamaCloudRetriever (#17086)
- Adding AWS Nova models to Bedrock Converse (#17139)
- Ollama LLM: Added TypeError exception to
_get_response_token_counts
(#17150)
- changes in openai identification in url (#17161)
- Fix mem0 version check (#17159)
- fix: make OpenAIMultiModal work with new ChatMessage (#17138)
- Add AWS Bedrock Reranker (#17134)
- update doc id for unstructured reader (#17160)
- fix: use prepared statement in DuckDBRetriever (#17092)
- Create tables for pgvector regardless of schema status (#17100)
- make alpha not none in weaviate (#17163)
- Make Weaviate Vector Store integration work with complex properties (#17129)
- Add support for
IS_EMPTY
metadata filters to Weaviate Vector Store integration (#17128) - Make Weaviate Vector Store integration support nested metadata filtering (#17107)
- improve traceback logging for workflows (#17040)
- Initial version of checkpointing for Workflows (#17006)
- Fix base component tests (#17062)
- mark code splitter tests as optional (#17060)
- update contributing guide and dev deps (#17051)
- fix Handling of WorkflowDone exception (#17047)
- remove pyarrow from vertex deps (#16997)
- IBM watsonx.ai acomplete & achat integration (#17034)
- fixes tool_choice in certain scenarios for openai (#17058)
- Fix: update perplexity models, add unit tests and minor fixes (#17045)
- remove pyarrow from vertex deps (#16997)
- Add stream_chat and conditionally set AutoModelClass to MllamaForConditionalGeneration (#17031)
- Fix box api - mypy breakage (#17061)
- deeplake v3 backward compatibility (#17057)
- MongoDB Atlas: Adds search index commands and tests/examples of metadata filters (#15265)
- check if schema exists before doing rest of init during pgvector table creation (#17063)
- Pin pydantic to
<2.10
temporarily - feat[react-multimodal]: add gemini support for images in MultimodalReActAgentWorker (#16992)
- api_key typo in llama-cloud-index (#17001)
- Add support for latest gpt-4o model (#17015)
- zhipuai add stop param to api (#16996)
- Add support for mistral latest models (#16991)
- Fix: Correct typo in replicate multi modal package (#16998)
- add get_nodes to pinecone (#17007)
NOTE: Updating to v0.12.0 will require bumping every other llama-index-*
package! Every package has had a version bump. Only notable changes are below.
- Dropped python3.8 support, Unpinned numpy (#16973)
- Kg/dynamic pg triplet retrieval limit (#16928)
- Add ID support for LlamaCloudIndex & update from_documents logic, modernize apis (#16927)
- allow skipping waiting for ingestion when uploading file (#16934)
- add support for files endpoints (#16933)
- Add Custom Prompt Parameter (#16976)
- minor fix for messages/completion to prompt (#15729)
- Fix async streaming with bedrock converse (#16942)
- add vlm support (#16751)
- Permit passing params to Confluence client (#16961)
- Add base URL extraction method to GithubRepositoryReader (#16926)
- Allow passing in Weaviate vector store kwargs (#16954)
- Fix workflow timeout when streaming and exception happens (#16852)
- New function for core.multi_modal_llms.generic_utils (#16896)
- Added docstrings and unit tests for core.multimodal (#16872)
- actually setting num_workers = num_cpus for ingestion pipeline (#16803)
- skip model validation for nvdev embedding models (#16883)
- Add ModelScope embedding support (#16873)
- BREAKING: Update relationship table label nomenclature, existing graphs will need to be recreated (#16886)
- use new anthropic token counting api (#16909)
- Add inference profile support to Bedrock Converse and add Llama 3.1/3.2 (#16827)
- skip model validation for nvdev llm models (#16882)
- Fix modelscope 'pipeline' referenced before assignment (#16864)
- fix ignoring tool calls in additional kwargs of messages (#16764)
- add siliconflow llm class (#16861)
- fix top-n parameter in TEI reranker (#16884)
- add gitbook reader (#16862)
- feat: add relative file path google drive (#16907)
- Bump versions and update pypdf dependency to 5.1.0 (#16905)
- Add UserAgent header "llamaindex-python" for azure search (#16895)
- fix clickhouse init in vector store (#16903)
- bring back support for prompt templates in context chat engines (#16821)
- Fixed the JSON Format of Generated Sub-Question (double curly brackets) (#16820)
- markdown splitter improve metadata (#16789)
- fix empty index + generation synthesizer (#16785)
- Support for api_version and Azure AI model inference service (#16802)
- fix await-async-embeddings (#16790)
- add siliconflow embedding class (#16753)
- Hotfix: Chain Query Configuration (#16818)
- Add Anthropic Claude Haiku 3.5 to the list of supported Claude models (#16823)
- Support for api_version and Azure AI model inference service (#16802)
- Add Anthropic Claude Haiku 3.5 to the list of supported Claude models for bedrock and bedrock-converse integrations (#16825)
- Add Anthropic Claude Haiku 3.5 to the list of supported Claude models for bedrock and bedrock-converse integrations (#16825)
- More tolerant definition of LLMMetadata information (#16830)
- Fix abstract method signature error (#16809)
- remove beam search param for latest vllm (#16817)
- Add ColPali as reranker (#16829)
- add siliconflow rerank class (#16737)
- fix: add required_exts for one drive reader (#16822)
- Support breaking changes to filter syntax in latest chroma (#16806)
- support sparse embedding models, fix delete for serverless for pinecone (#16819)
- Fixed issue with default value set as None for workflow
ctx.get()
(#16756) - fix various issues with react agent streaming (#16755)
- add unit test for query pipeline (#16749)
- Fix _merge_ref_doc_kv_pairs duped for-loop (#16739)
- bugfix: determine if nodes is none when creating index (#16703)
- fixes LLMRerank default_parse_choice_select_answer_fn parsing issue (#16736)
- fix return type check on workflows (#16724)
- Fixing a verbose issue and making sql errors more informative (#16686)
- add siliconflow embedding class (#16753)
- Multi-Graph-Supported-FalkorDB (#16482)
- adding additional claude model name, for vertex AI (#16692)
- Added mistral large2 model id in bedrock (#16742)
- Improve Bedrock Tool Calling (#16723)
- add new sonnet3.5 to function calling bedrock converse models (#16702)
- update bedrock models (#16698)
- Added mistral large2 model id in bedrock (#16742)
- add new sonnet3.5 to function calling bedrock converse models (#16702)
- update bedrock models (#16698)
- Adding support to the new Aya-Expanse models from Cohere (#16733)
- DashScope llm support async (#16711)
- add nvidia/llama-3.2-nv-embedqa-1b-v1 to set of supported models (#16694)
- Pipeshift llama index integration (#16610)
- add Mem0 as a memory (#16708)
- Fix anthropic multimodal deps conflict, update models (#16699)
- feat: update Docling reader & node parser to Docling v2 (#16677)
- add nvidia/llama-3.2-nv-rerankqa-1b-v1 to set of supported models (#16695)
- add siliconflow rerank class (#16737)
- feat: update Docling reader & node parser to Docling v2 (#16677)
- feat: add permissions to one drive metadata (#16646)
- Add Managed Identity authentication support for Azure storage components (#16710)
- Add missing awaits in azure chat store (#16645)
- Add Managed Identity authentication support for Azure storage components (#16710)
- Add Managed Identity authentication support for Azure storage components (#16710)
- Add Managed Identity authentication support for Azure storage components (#16710)
- Makes the tool more compatible with the options, also for the future (#16676)
- Add Vectara Query Tool (#16722)
- Allow defining retrievable fields in Azure Vector Store (#16766)
- feat: add get_nodes azureai search (#16761)
- Added get_nodes() function in AISearch vector store (#16653)
- Fix querying for ID in AzureAISearchVectorStore (fixes delete_nodes by node_ids) (#16769)
- Fixed issue with persistence, rearranged and added new options to construction of HnswlibVectorStore (#16673)
- Init OpensearchVectorClient with
os_async_client
(#16767)
- chore: add embeddings on qdrant get_nodes return (#16760)
- add default ID if node ID is not provided (#16671)
- [actually nothing!]
- Add support for cohere multi-modal embeddings (#16667)
- support timeout param in litellmembedding (#16532)
- Make neo4j schema refresh configurable (#16651)
- fix: receive warnings from dbms server in neo4j queries (#16598)
- add chain postprocessing for vectara (#16627)
- update anthropic model names (#16643)
- fix: skip processing of choice.delta when it is None (#16636)
- Reka llamaindex integration (llm and multi-modal-llm) (#15753)
- Reka llamaindex integration (llm and multi-modal-llm) (#15753)
- Fix BUG where the api_key parameter is not set when calling DashScopeRerank (#16665)
- add retry logic to requests in cases where access token expires (#16662)
- fix missing dependency error for mongodb docstore (#16654)
- Add support for Couchbase for DocStore & IndexStore (#16509)
- Add support for Couchbase for DocStore & IndexStore (#16509)
- Add support for Couchbase for IndexStore (#16509)
- add missing dependency for mongodb kvstore (#16632)
- Fix drawing for HITL (#16624)
- Raise errors in instrumentation properly when handling asyncio futures (#16603)
- fix: pass params to VectorStoreQuery properly in PGRetriever (#16586)
- Fix structured predict type hints (#16585)
- Add async version of retry_on_exceptions_with_backoff utility (#16374)
- Refine CONTRIBUTING.md Documentation (#16548)
- Add embedding integration with Intel Gaudi in llama-index-embeddings-gaudi (#16521)
- Add OpenClip support through OpenVINO embedding (#16554)
- fix: remove warnings from dbms server in neo4j queries (#16598)
- IBM watsonx.ai Chat integration + function calling support (#16589)
- Add support for mistral latest models (#16571)
- Added check for fine-tuned models in function_calling method (#16568)
- allow passing in openai clients directly (#16560)
- Use Replicate streaming API (#16597)
- Add OpenVINO multimodal support (#16567)
- add zhipuai multi modal llm class (#16551)
- Use port of redirect uri in credential file to run local server in GoogleDocsReader (#16327)
- feat: postgres chat store Integration (#16557)
- Fix pydantic errors in upstash chat store (#16559)
- Add DiskANN for Azure Cosmos DB Mongo vector store (#16581)
- Hnswlib Vector Store integration (#16443)
- Add vector store integration of OceanBase (#16550)
- Added optional name of text field in Qdrant vector database. (#16576)
- Handle Empty Nodes List in PG Retrieval while adding text (#16447)
- Improved text2sql parsing (#16445)
- support timeout param in litellmembedding (#16532)
- add zhipuai embedding class (#16505)
- Add Memgraph Graph Store Integration (#16345)
- Add anthropic bedrock support (#16478)
- Fix Databricks structured_predict (#16527)
- Propagate default_headers in AnthropicMultiModal (#16496)
- Added Document360Reader. Contributed by the PLACE team. (#16305)
- Add Zyte serp integration (#16417)
- Added UpstageDocumentParseReader (#16099)
- Fix async methods in azure chat store (#16531)
- Fix format temp function (#16487)
- Add Support for Custom Metadata Keyword Suffix in Elasticsearch Integration (#16519)
- Add vector store integration for Nile (multi-tenant postgres) (#16437)
- Use efficient kNN filtering, fix filtering when input value is array of string (#16393)
- Various fixes for Oracle vector store integration (#16536)
- Update model name typo in Qdrant utils.py (#16494)
- fix timescale vector store class (#16539)
- Fixed issue for similarity score from Weaviate (#16489)
- Fix ChatMessage serialization with janky openai types (#16410)
- fix gemini embedding async method (#16369)
- Add llm integration with Intel Gaudi in llama-index-llms-gaudi (#16308)
- don't include tool calls if there was none (#16408)
- LlamaMultiModal class bug fix (#16413)
- MultiModal.HuggingFaceMultiModal: fix errors and README, add stream_complete (#16376)
- feat: add Docling reader and node parser (#16406)
- feat: add Docling reader and node parser (#16406)
- Add Zyte serp integration (#16417)
- fix(bug): fixed bug with ensuring the metadata_mapping value (#16431)
- Add ObjectBox Vector Store Integration (#16314)
- Treat non-dict tool JSON function arguments as empty (instead of unexpected raise) (#16316)
- Fixing instrumentation for workflows (#16290)
- (workaround) Suppress token detaching exception during workflow tracing (#16364)
- Raise warning instead of error when nodes have no content (#16354)
- Fix typo in BasePGRetriever causing graph context to not be added (#16357)
- adding vertex endpoint embedding (#16351)
- Adding support in FireworksAI for Meta 3.2 Models: 1b-instruct; 3b-instruct; 11b-vision; 90b-vision (#16349)
- Refactor OpenAI
update_tool_calls
(#16309)
- Add support for nested MetadataFilters and FilterOperator.IS_EMPTY (#16329)
- added
to_dict()
,from_dict()
and serializers for workflow context (#16250) - Ability to cancel workflow execution with
handler.cancel_run()
(#16320) - (breaking) Refactor
WorkflowHandler.run_step()
so user manually emits Event to start next step in workflow (#16277)
- Oraclevs integration (#16161)
- nudge-ft package and add an example for expanding your dataset (#16269)
- Add support for prompt caching for Anthropic LLM (#16270)
- Output token usage in raw data for Google Gemini LLMs (#16313)
- add 4o mini to azure openai models (#16335)
- Rremoving safety settings from generation config for Vertex AI models (#16337)
- LlamaIndex Multi_Modal_Llms Integration: Huggingface (#16133)
- rm extra print statements & replace create tmpfile function (#16291)
- Oraclevs integration (#16161)
- Add Zyte Web Reader (#16197)
- bump deps for latest bm25s version (#16339)
- Update BM25 retriever to use metadata (#16267)
- fix check for async client in redis chat store (#16321)
- Upstash Storage Chat Store Integration (#16237)
- milvus: always set self._collection (#16306)
- Fix milvus collection creation with index_config (#16165)
- Oracledb integration (#16161)
- Support TEXT_MATCH FilterOperator in Postgres Vector Store (#16304)
- Enhance insert Method in BaseIndex to Support Customizable Transformations (#16206)
- Ensure ChatMemoryBuffer's chat history never begins with a TOOL message (#16214)
- safe prompt helper string formatting (#16219)
- [Feature Request] Support max concurrent workflow_instance.run() executions (#16215)
- Workflows + Human In The Loop Dedicated Support (#16220)
- fix NoneType object error when passing in provided client (#16174)
- fix ollama chat missing
keep_alive
(#16182)
- Fix vertex init function (#16216)
- Add support for Mistral Multi modal LLM (#16191)
- Add pagination support for Jira Reader (#16226)
- Azure Cosmos DB Filtered Vector Search (#16175)
- Azure Cosmos DB Filtered Vector Search (#16175)
- Add missing JWT dependency (#16236)
- add option for string node representation during retireval in property graphs (#16100)
- improve markdown element node parser and structured prediction reliability (#16172)
- Fixed issue where Neptune was adding additional labels (#16137)
- Pass safety_settings to send_message methods to fix settings not being sent to API (#16153)
- upgrading box sdk to >= 1.5.0 #16169
- Async support for dynamodb (#16139)
- Async support for redis (#16139)
- Depend on AstraPy 1.5 and above for AstraDBVectorStore (#16164)
- Correct Pydantic warning(s) issed for llm base class (#16141)
- globally safe format prompt variables in strings with JSON (#15734)
- account for tools in prompt helper and response synthesizers (#16157)
- feat: add drive link to google drive reader metadata (#16156)
- Add required_exts option to SharePoint reader (#16152)
- Support user-defined schema in MilvusVectorStore (#16151)
- Use response synthesizer in context chat engines (#16017)
- Async chat memory operation (#16127)
- Sql query add option for markdown response (#16103)
- Add support for Path for SimpleDirectoryReader (#16108)
- Update chat message class for multi-modal (#15969)
- fix:
handler.stream_events()
doesn't yield StopEvent (#16115) - pass
hybrid_top_k
in vector retriever (#16105)
- fix elasticsearch embedding async function (#16083)
- feat: update JinaEmbedding for v3 release (#15971)
- Enhance Pandas Query Engine Output Processor (#16052)
- fix incorrect parameters in VertexAIIndex client (#16080)
- Add TopicNodeParser based on MedGraphRAG paper (#16131)
- Implement async for multi modal ollama (#16091)
- feat: add configurable base_url field in rerank (#16050)
- fix bug missing import for bytesio (#16096)
- Wordpress: Allow control of whether Pages and/or Posts are retrieved (#16128)
- Fix Issue 16071: wordpress requires username, password (#16072)
- fix hybrid search with latest lancedb client (#16057)
- Fix mongodb hybrid search top-k specs (#16105)
- context/result refactor for workflows (#16036)
- add sparse embedding abstraction (#16018)
- Fix Pydantic models numeric validation (#16008)
- Human in loop workflow example (#16011)
- opik integration (#16007)
- update llamacloud index with image nodes (#15996)
- Hotfix: Fix Citations Text (#16015)
- Fix: unnecessary warning issue in HuggingFace LLM when tokenizer is provided as argument (#16037)
- fix: new Data Connector adaption for DashVector (#16028)
- add quip reader (#16000)
- add fastembed sparse embeddings (#16018)
- Fix: get all documents from Elasticsearch KVStore (#16006)
- temporarily limit lancedb version (#16045)
- Implement
get_nodes()
on PGVectorStore (#16026)
- Add callback manager to retriever query engine from args (#15990)
- Do not pass system prompt from fn calling runner to fn calling worker (#15986)
- fix: Error when parsing react output if tool name contains non-English characters (#15956)
- Add four alibabacloud-aisearch llama-index integrations: rerank, node_parser, readers, embeddings (#15934)
- Add NUDGE Finetuning (#15954)
- update falkordb client (#15940)
- Add support for o1 openai models (#15979)
- force temp to 1.0 for o1 (#15983)
- Add four alibabacloud-aisearch llama-index integrations: rerank, node_parser, readers, embeddings (#15934)
- Add four alibabacloud-aisearch llama-index integrations: rerank, node_parser, readers, embeddings (#15934)
- Add four alibabacloud-aisearch llama-index integrations: rerank, node_parser, readers, embeddings (#15934)
- Differentiate sync and async calls in OpenSearchVectorClient (#15945)
- fix attribute error in PGVectorStore (#15961)
- add support for engine parameters (#15951)
- Catch nest_asyncio errors (#15975)
- feat: Add a retry policy config to workflow steps (#15757)
- Add doc id to Langchain format conversions (#15928)
- Add DynamoDBChatStore (#15917)
- Fix RagCLI pydantic error (#15931)
- add llama-index llms alibabacloud_aisearch integration (#15850)
- Make default mistral model support function calling with
large-latest
(#15906)
- Add InternalServerError to retry decorator (#15921)
- Update RankLLM with new rerankers (#15892)
- Adding vector store for Azure Cosmos DB NoSql (#14158)
- Fix error handling in sharepoint reader, fix error with download file (#15868)
- Adding support for MetadataFilters to WordLift Vector Store (#15905)
- Opensearch Serverless filtered query support using knn_score script (#15899)
- Make SentenceSplitter's secondary_chunking_regex optional (#15882)
- force openai structured output (#15706)
- fix assert error, add type ignore for streaming agents (#15887)
- Fix image document deserialization issue (#15857)
- Bug fix for KuzuPropertyGraphStore: Allow upserting relations even when chunks are absent (#15889)
- Removed unused llama-index-llms-anthropic dependency from Bedrock Converse (#15869)
- Fix PGVectorStore with latest pydantic, update pydantic imports (#15886)
- Add TablestoreVectorStore (#15657)
- add llama-deploy docs to docs builds (#15794)
- Add oreilly course cookbooks (#15845)
- Various bug fixes (#15836)
- Update ImageReader file loading logic (#15848)
- Various bug fixes (#15836)
- Refresh Opensearch index after delete operation (#15854)
- remove unneeded assert in property graph retriever (#15832)
- make simple property graphs serialize again (#15833)
- fix json schema for fastapi return types on core components (#15816)
- NVIDIA llm: Add Completion for starcoder models (#15802)
- add ollama response usage (#15773)
- fix pydantic v2 validation errors (#15800)
- fix: convert Document id from int to string in DiscordReader (#15806)
- Add MariaDB vector store integration package (#15564)
- Add mypy to core (#14883)
- Fix incorrect instrumentation fields/types (#15752)
- FunctionCallingAgent sources bug + light wrapper to create agent (#15783)
- Add text to sql advanced workflow nb (#15775)
- fix: remove context after streaming workflow to enable streaming again (#15776)
- Fix chat memory persisting and loading methods to use correct JSON format (#15545)
- Fix
_example_type
class var being read as private attr with Pydantic V2 (#15758)
- add dimensions param to LiteLLMEmbedding, fix a bug that prevents reading vars from env (#15770)
- Bugfix upstage embedding when initializing the UpstageEmbedding class (#15767)
- Fix Sagemaker Field required issue (#15778)
- fix relations upsert with special chars (#15769)
- Add native vector index support for neo4j lpg and fix vector filters (#15759)
- fix: GitHub Models metadata retrieval (#15747)
- Update
base.py
to fixself
issues (#15729)
- add ollama response usage (#15773)
- Fix Sagemaker Field required issue (#15778)
- Support image type detection without knowing the file name (#15763)
- feat: implement get_nodes for MilvusVectorStore (#15696)
- fix: tencentvectordb inconsistent attribute name (#15733)
- refact: merge Context and Session to simplify the workflows api (#15709)
- chore: stop using deprecated
ctx.data
in workflows docs (#15716) - fix: stop streaming workflow events when a step raises (#15714)
- Fix llm_chat_callback for multimodal llms (#15700)
- chore: Increase unit tests coverage for the workflow package (#15691)
- fix SimpleVectorStore.from_persist_dir() behaviour (#15534)
- fix json serialization for azure embeddings (#15724)
- Add KuzuPropertyGraphStore (#15678)
- added new User Defined Function reranker (#15546)
- Fix
random_seed
type in mistral llm (#15701)
- Add function/tool calling support to nvidia llm (#15359)
- bump ollama client deps for multimodal llm (#15702)
- Fix: Firecrawl scraping url response (#15720)
- Adding Not Diamond to llama_index (#15703)
- MMR in Milvus vector stores (#15634)
- feat: implement get_nodes for MilvusVectorStore (#15696)
- fix tool schemas generation for pydantic v2 to handle nested models (#15679)
- feat: support default values for nested workflows (#15660)
- feat: allow FunctionTool with just an async fn (#15638)
- feat: Allow streaming events from steps (#15488)
- fix auto-retriever pydantic indent error (#15648)
- Implement Router Query Engine example using workflows (#15635)
- Add multi step query engine example using workflows (#15438)
- start traces for llm-level operations (#15542)
- Pass callback_manager to init in CodeSplitter from_defaults (#15585)
- Add Xinference Embedding Class (#15579)
- Integrations: AI21 function calling Support (#15622)
- Added support for anthropic models through GCP Vertex AI (#15661)
- Implement Cerebras Integration (#15665)
- fix downloaded nim endpoint path (#15645)
- fix llama-index-postprocessor-nvidia-rerank tests (#15643)
- add xinference rerank class (#15639)
- fix set output fields in AlibabaCloudOpenSearchConfig (#15562)
- Upgrade azure-search-documents to 2024-07-01 GA API and Add Support for Scalar and Binary Quantization in Index Creation (#15650)
- Neo4j Vector Store: Make Embedding Dimension Check Optional (#15628)
- Change the default consistency level of Milvus (#15577)
- Fix the ElasticsearchStore key error (#15631)
- Replacing client-side docs search with algolia (#15574)
- Add docs on extending workflows (#15573)
- rename method for nested workflows to add_workflows (#15596)
- chore: fix @step usage in the core codebase (#15588)
- Modify the validate function in ReflectionWorkflow example notebook to use pydantic model_validate_json method (#15567)
- feature: allow concurrent runs of the same workflow instance (#15568)
- docs: remove redundant pass_context=True from docs and examples (#15571)
- fix openai embeddings with pydantic v2 (#15576)
- bump voyage ai embedding client dep (#15595)
- Vertex LLM: Correctly add function calling part to prompt (#15569)
- Vertex LLM: Remove manual setting of message content to Function Calling (#15586)
- removed deprecated
ServiceContext
-- using this now will print an error with a link to the migration guide - removed deprecated
LLMPredictor
-- using this now will print an error, any existing LLM is a drop-in replacement - made
pandas
an optional dependency
- bumped the minor version of every package to account for the new version of
llama-index-core
- remove nested progress bars in base element node parser (#15550)
- Adding exhaustive docs for workflows (#15556)
- Adding multi-strategy workflow with reflection notebook example (#15445)
- remove openai dep from core (#15527)
- Improve token counter to handle more response types (#15501)
- feat: Allow using step decorator without parentheses (#15540)
- feat: workflow services (aka nested workflows) (#15325)
- Remove requirement to specify "allowed_query_fields" parameter when using "cypher_validator" in TextToCypher retriever (#15506)
- fix mistral embeddings usage (#15508)
- use ollama client for embeddings (#15478)
- support static input shape for openvino embedding and reranker (#15521)
- Added code to expose structured schema for Neptune (#15507)
- Integration: AI21 Tools support (#15518)
- Support token counting for llama-index integration with bedrock (#15491)
- feat: add tool calling support for achat cohere (#15539)
- Adding gigachat LLM support (#15313)
- Fix incorrect type in OpenAI token usage report (#15524)
- allow streaming token counts for openai (#15548)
- add truncate support (#15490)
- Update to 0.2.0, remove old code (#15533)
- update default model to nvidia/nv-rerankqa-mistral-4b-v3 (#15543)
- Fixing the issues in loading file paths from bitbucket (#15311)
- enhance google drive reader for improved functionality and usability (#15512)
- check and sanitize remote reader urls (#15494)
- fix: setting IDF modifier in QdrantVectorStore for sparse vectors (#15538)
- avoid nltk 3.9 since its broken (#15473)
- docs: openllmetry now uses instrumentation (#15443)
- Fix LangChainDeprecationWarning (#15397)
- Add get/set API to the Context and make it coroutine-safe (#15152)
- docs: Cleanlab's cookbook (#15352)
- pass kwargs in
async_add()
for vector stores (#15333) - escape json in structured llm (#15404)
- docs: Add JSONAlyze Query Engine using workflows cookbook (#15408)
- Add GigaChat embedding (#15278)
- feat: Integrating Azure OpenAI Finetuning (#15297)
- Exposed NeptuneQueryException and added additional debug information (#15448)
- Fixed issue #15414 and added ability to do partial matchfor Neptune Analytics (#15415)
- Use backticks to escape label (#15324)
- feat: add tool calling for cohere (#15144)
- Ports over LongRAGPack, Corrective RAG Pack, and Self-Discover Pack to Workflows (#15160)
- Ports over LongRAGPack, Corrective RAG Pack, and Self-Discover Pack to Workflows (#15160)
- Ports over LongRAGPack, Corrective RAG Pack, and Self-Discover Pack to Workflows (#15160)
- Enhance PreprocessReader (#15302)
- Temporarily revert nltk dependency due to latest version being removed from pypi
- Add citation query engine with workflows example (#15372)
- bug: Semantic double merging splitter creates chunks larger thank chunk size (#15188)
- feat: make
send_event()
in workflows assign the target step (#15259) - make all workflow events accessible like mappings (#15310)
- Add BGEM3Index (#15197)
- update HF's completion_to_prompt (#15354)
- Wrapper for SambaNova (Sambaverse and SambaStudio) with Llama-index (#15220)
- Update code_hierarchy.py adding php support (#15145)
- fix bug when calling llama-index-postprocessor-dashscope-rerank (#15358)
- Box refactor: Box File to Llama-Index Document adaptor (#15314)
- GCSReader: Implementing ResourcesReaderMixin and FileSystemReaderMixin (#15365)
- Box refactor: Box File to Llama-Index Document adaptor (#15314)
- Box tools for AI Agents (#15236)
- Check if hnsw index exists (#15287)
- chore: bump nltk version (#15277)
- Box tools for AI Agents (#15236)
- feat: add default_headers to Gemini multi-model (#15296)
- chore: stop using ServiceContext from the clickhouse integration (#15300)
- chore: remove ServiceContext usage from experimental package (#15301)
- fix: MarvinMetadataExtractor functionality and apply async support (#15247)
- chore: bump black version (#15288)
- chore: bump nltk version (#15277)
- chore: bump nltk version (#15277)
- chore: bump nltk version (#15277)
- chore: bump nltk version (#15277)
- chore: bump nltk version (#15277)
- chore: bump nltk version (#15277)
- chore: bump nltk version (#15277)
- Feature: added multiprocessing for creating hf embedddings (#15260)
- fix: children nodes not carrying metadata from source nodes (#15254)
- Workflows: fix the validation error in the decorator (#15252)
- fix: strip '''sql (Markdown SQL code snippet) in SQL Retriever (#15235)
- Remove usage of ServiceContext in Colbert integration (#15249)
- feat: Support Milvus collection properties (#15241)
- Update models supported by Cleanlab TLM (#15240)
- add generation prompt to HF chat template (#15239)
- add generation prompt to HF chat template (#15239)
- Neo4jPropertyGraphStore.get() check for id prop (#15228)
- Fix fs.open path type (#15226)
- add num_workers in workflow decorator to resolve step concurrancy issue (#15210)
- Sub Question Query Engine as workflow notebook example (#15209)
- Add Llamatrace to workflow notebooks (#15186)
- Use node hash instead of node text to match nodes in fusion retriever (#15172)
- handle mistral v1.0 client (#15229)
- Fix relik extractor skip error (#15225)
- handle mistral v1.0 client (#15229)
- Add neo4j generic node label (#15191)
- Allow for images in Anthropic messages (#15227)
- handle mistral v1.0 client (#15229)
- Update Mixture Of Agents llamapack with workflows (#15232)
- Fixed slack client ref in ToolSpec (#15202)
- feat: Allow None metadata filter by using IS_EMPTY operator (#15167)
- fix: use parent source node to node relationships if possible during node parsing (#15182)
- Use node hash instead of node text to match nodes in fusion retriever (#15172)
- Neo4j property graph client side batching (#15179)
- PropertyGraphStore support for Amazon Neptune (#15126)
- feat: add default_headers to Gemini model (#15141)
- OpenAI: Support new strict functionality in tool param (#15177)
- Add support for full MetadataFilters in Opensearch (#15176)
- feat: Allow None metadata filter by using IS_EMPTY operator (#15167)
- Add support for fields projection and update sample Notebook (#15140)
- Tweaks to workflow docs (document
.send_event()
, expand examples) (#15154) - Create context manager to instrument event and span tags (#15116)
- keyval index store index store updated to accept custom collection suffix (#15134)
- make workflow context able to collect multiples of the same event (#15153)
- Fix
__str__
method for AsyncStreamingResponse (#15131)
- feat(integration): add a global handler for Literal AI (#15064)
- Add relik kg constructor (#15123)
- fix neo4j property graph relation properties when querying (#15068)
- feat: add default_headers to Fireworks llm (#15150)
- Fix: Gemini 1.0 Pro Vision has been official deprecated, switch default model to gemini-1.5-flash (#15000)
- Add LLM for AlibabaCloud PaiEas (#14983)
- Fix Predibase Integration for HuggingFace-hosted fine-tuned adapters (#15130)
- update
StartEvent
usage to allow for dot notation attribute access (#15124) - Add GraphRAGV2 notebook (#15119)
- Fixed minor bug in DynamicLLMPathExtractor as well as default output parsers not working (#15085)
- update typing for workflow timeouts (#15102)
- fix(sql_wrapper): dont mention foreign keys when there is none (#14998)
- fix neo4j retrieving relation properties (#15111) (#15108)
- Update base.py to use @atexit for cleanup (#15047)
- bump pinecone client version deps (#15121)
- Handle nested MetadataFilters for Redis vector store (#15093)
- Update WordLift Vector Store to use new client package (#15045)
- Introduce
Workflow
s for event-driven orchestration (#15067) - Added feature to context chat engine allowing previous chunks to be inserted into the current context window (#14889)
- MLflow Integration added to docs (#14977)
- docs(literalai): add Literal AI integration to documentation (#15023)
- expand span coverage for query pipeline (#14997)
- make re-raising error skip constructor during
asyncio_run()
(#14970)
- Add proper async embedding support
- add support for textembed embedding (#14968)
- initial implementation FalkorDBPropertyGraphStore (#14936)
- Fix: Azure AI inference integration support for tools (#15044)
- Updates to Default model for support for function calling (#15046)
- toggle for ollama function calling (#14972)
- Add function calling for Ollama (#14948)
- update to OpenLLM 0.6 (#14935)
- Adds a LlamaPack that implements LongRAG (#14916)
- Support for Re-Ranker via Text Embedding Interface (#15063)
- confluence reader sort auth parameters priority (#14905)
- UnstructuredReader use filename as ID (#14946)
- Add GitLab reader integration (#15030)
- Fix issue with average ratings being a float vs an int (#15070)
- use proper stemmer in bm25 tokenize (#14965)
- Fix issue with deleting non-existent index (#14949)
- disable embeddings for sparse strategy (#15032)
- Update default sparse encoder for Hybrid search (#15019)
- Enhance MilvusVectorStore with flexible index management for overwriting (#15058)
- Adds option to construct PGVectorStore with a HNSW index (#15024)
- Fix: Token counter expecting response.raw as dict, got ChatCompletionChunk (#14937)
- Return proper tool outputs per agent step instead of all (#14885)
- Minor bug fixes to async structured streaming (#14925)
- fireworks ai llama3.1 support (#14914)
- Add claude 3.5 sonnet to multi modal llms (#14932)
- 🐞 fix(integrations): BM25Retriever persist missing arg similarity_top_k (#14933)
- Llamaindex retriever for Vertex AI Search (#14913)
- Improved
deeplake.get_nodes()
performance (#14920)
- Bugfix: Don't pass empty list of embeddings to elasticsearch store when using sparse strategy (#14918)
- Add vector store integration of lindorm (#14623)
- feat: allow to limit how many elements retrieve (qdrant) (#14904)
- Add an optional parameter similarity_score to VectorContextRetrieve… (#14831)
- add property extraction (using property names and optional descriptions) for KGs (#14707)
- able to attach output classes to LLMs (#14747)
- Add streaming for tool calling / structured extraction (#14759)
- fix from removing private variables when copying/pickling (#14860)
- Fix empty array being send to vector store in ingestion pipeline (#14859)
- optimize ingestion pipeline deduping (#14858)
- Add an optional parameter similarity_score to VectorContextRetriever (#14831)
- Bugfix: AzureOpenAI may fail with custom azure_ad_token_provider (#14869)
- feat: ✨ Implement async functionality in BedrockConverse (#14326)
- make some dependencies optional
- bump langchain version in integration (#14879)
- Bugfix: ollama streaming response (#14830)
- align deps (#14850)
- update notion reader to handle duplicate pages, database+page ids (#14861)
- Implements delete_nodes() and clear() for Weviate, Opensearch, Milvus, Postgres, and Pinecone Vector Stores (#14800)
- MongoDB Atlas Vector Search: Enhanced Metadata Filtering (#14856)
- Implements delete_nodes() and clear() for Weviate, Opensearch, Milvus, Postgres, and Pinecone Vector Stores (#14800)
- Implements delete_nodes() and clear() for Weviate, Opensearch, Milvus, Postgres, and Pinecone Vector Stores (#14800)
- Implements delete_nodes() and clear() for Weviate, Opensearch, Milvus, Postgres, and Pinecone Vector Stores (#14800)
- Implements delete_nodes() and clear() for Weviate, Opensearch, Milvus, Postgres, and Pinecone Vector Stores (#14800)
- Fixing the issue where the _apply_node_postprocessors function needs QueryBundle (#14839)
- Add Context-Only Response Synthesizer (#14439)
- Fix AgentRunner AgentRunStepStartEvent dispatch (#14828)
- Improve output format system prompt in ReAct agent (#14814)
- Remove double curly replacing from output parser utils (#14735)
- Update simple_summarize.py (#14714)
- chore: read AZURE_POOL_MANAGEMENT_ENDPOINT from env vars (#14732)
- Azure AI Inference integration (#14672)
- Azure AI Inference integration (#14672)
- feat: ✨ Implement async functionality in BedrockConverse (#14326)
- Add new integration for YandexGPT Embedding Model (#14313)
- Update docstring for gmailtoolspec's search_messages tool (#14840)
- add support for nvidia/nv-rerankqa-mistral-4b-v3 (#14844)
- Fix OpenAI Embedding async client bug (#14835)
- Fix Azure OpenAI LLM and Embedding async client bug (#14833)
- Fix Azure OpenAI LLM and Embedding async client bug (#14833)
- Add support for gpt-4o-mini (#14820)
- Add support for gpt-4o-mini (#14820)
- Add support for mistralai nemo model (#14819)
- Fix bug when sanitize is used in neo4j property graph (#14812)
- Add filter to get_triples in neo4j (#14811)
- feat: add nested filters for azureaisearch (#14795)
- feat: Add NOT IN filter for Qdrant vector store (#14791)
- feat: add azureaisearch supported conditions (#14787)
- feat: azureaisearch support collection string (#14712)
- Fix OpenWeatherMapToolSpec.forecast_tommorrow_at_location (#14745)
- follow odata.nextLink (#14708)
- Adds Quantization option to QdrantVectorStore (#14740)
- feat: improve azureai search deleting (#14693)
- fix: tools are required for attachments in openai api (#14609)
- new integration
- fix fastembed python version (#14710)
- Various docs updates
- Add user configurations for Cleanlab LLM integration (#14676)
- race between concurrent pptx readers over a single temp filename (#14686)
- changes to Exa search tool getting started and example notebook (#14690)
- fix: update operator logic for simple vector store filter (#14674)
- Add AgentOps integration (#13935)
- chore: update required python version in Qdrant fastembed package (#14677)
- Bump version llama-index-embeddings-huggingface-optimum-intel (#14670)
- Added support for custom index settings (#14655)
- Initial release
- Fix #14637 Llamaindex managed Vertex AI index needs to be updated. (#14641)
- fix unstructured import in simple file reader (#14642)
- fix handling react usage in
llm.predict_and_call
for llama-agents (#14556) - add the missing arg verbose when
ReActAgent
callingsuper().__init__
(#14565) - fix
llama-index-core\llama_index\core\node_parser\text\utils.py
error when use IngestionPipeline parallel (#14560) - deprecate
KnowledgeGraphIndex
, tweak docs (#14575) - Fix
ChatSummaryMemoryBuffer
fails to summary chat history with tool callings (#14563) - Added
DynamicLLMPathExtractor
for Entity Detection With a Schema inferred by LLMs on the fly (#14566) - add cloud document converter (#14608)
- fix KnowledgeGraphIndex arg 'kg_triple_extract_template' typo error (#14619)
- Fix: Update
UnstructuredElementNodeParser
due to change in unstructured (#14606) - Update ReAct Step to solve issue with incomplete generation (#14587)
- Conditions logging to promptlayer on successful request (#14632)
- Add integration embeddings databricks (#14590)
- Fix MessageRole import from the wrong package in AI21 Package (#14596)
- handle empty response in Bedrock AnthropicProvider (#14479)
- add claude 3.5 sonnet support to Bedrock InvokeAPI (#14594)
- Fix Bedrock Converse's tool use blocks, when there are multiple consecutive function calls (#14386)
- add optimum intel with ipex backend to llama-index-integration (#14553)
- add baidu-qianfan llm (#14414)
- fix: crash LLMMetadata in model name lookup (#14569)
- Remove hf embeddings dep from text-embeddings-inference (#14592)
- update yi llm context_window (#14578)
- add fs arg to PandasExcelReader.load_data (#14554)
- UnstructuredReader enhancements (#14390)
- nit: firecrawl fixes for creating documents (#14579)
- Update BM25Retriever to use newer (and faster) bm25s library #(14581)
- refactor: Don't swallow exceptions from Qdrant collection_exists (#14564)
- add support for qdrant bm42, setting sparse + dense configs (#14577)
- fix file reader path bug on windows (#14537)
- follow up with kwargs propagation in colbert index due to change in parent class (#14522)
- deprecate query pipeline agent in favor of FnAgentWorker (#14525O)
- support latest version of arize #14526
- Add support for LiteLLM Proxy Server for embeddings (#14523)
- Adding device choice from sentence_transformers (#14546)
- Fixed ordering of returned nodes on vector queries (#14461)
- handle empty response in Bedrock AnthropicProvider (#14479)
- Fix Bedrock Converse's join_two_dicts function when a new string kwarg is added (#14548)
- Add upstage tokenizer and token counting method (#14502)
- Fix bug with getting object name for blobs (#14547)
- Pandas excel reader load data fix for appending documents (#14501)
- Add Iceberg Reader integration to LLamaIndex (#14477)
- Added retries (#14488)
- add
list_databases
method (#14488)
- Enhance SlackReader to fetch Channel IDs from Channel Names/Patterns (#14429)
- Add API url to firecrawl reader (#14452)
- fix score in nodes returned by the BM25 retriever (#14495)
- add async methods to azure ai search (#14496)
- Kdbai rest compatible (#14511)
- Adds Hybrid and Full-Text Search to MongoDBAtlasVectorSearch (#14490)
- fixed issue with function calling llms and empty tool calls (#14453)
- Fix ChatMessage not considered as stringable in query pipeline (#14378)
- Update schema llm path extractor to also take a list of valid triples (#14357)
- Pass the kwargs on when
build_index_from_nodes
(#14341)
- Add Alibaba Cloud dashscope agent (#14318)
- Add MetadataFilters to neo4j_property_graph (#14362)
- add known context lengths for hosted models (#14436)
- update available models (#14409)
- Better error handling for invalid API token (#14440)
- Integrate Yi model (#14353)
- Creates Data Loader for Google Chat (#14397)
- Invalidate s3fs cache in S3Reader (#14441)
- Add StructuredDataReader support for xlsx, csv, json and jsonl (#14369)
- Integrating a new tool called jina search (#14317)
- Update Astra DB vector store to use modern astrapy library (#14407)
- Fix the index accessing of ids of chroma get (#14434)
- Implemented delete_nodes() and clear() in deeplake vector store (#14457)
- Implemented get_nodes() in deeplake vector store (#14388)
- Add support for dynamic metadata fields in Elasticsearch index creation (#14431)
- Kdbai version compatible (#14402)
- added dead simple
FnAgentWorker
for custom agents (#14329) - Pass the kwargs on when build_index_from_nodes (#14341)
- make async utils a bit more robust to nested async (#14356)
- every llm is a chat model (#14334)
- added possibility to run local embedding model in RAG evaluation packages (#14352)
- Improvements to
llama-cloud
and client dependencies (#14254)
- Improve the interface and client interactions in
LlamaCloudIndex
(#14254)
- add claude sonnet 3.5 to bedrock converse (#14306)
- set default context size (#14293)
- add api_key alias on upstage llm and embeddings (#14233)
- Optimized inserts (#14321)
- azure_table_storage params bug (#14182)
- Add neo4j client method (#14314)
- Improve efficiency of average precision (#14260)
- add crewai + llamaindex cookbook (#14266)
- Add mimetype field to TextNode (#14279)
- Improve IBM watsonx.ai docs (#14271)
- Updated frontpage of docs, added agents guide, and more (#14089)
- Add support for claude 3.5 (#14277)
- Implement Bedrock Converse API for function calling (#14055)
- added average precision as a retrieval metric (#14189)
- added
.show_jupyter_graph()
method visualizing default simple graph_store in jupyter notebooks (#14104) - corrected the behaviour of nltk file lookup (#14040)
- Added helper args to generate_qa_pairs (#14054)
- Add new chunking semantic chunking method: double-pass merging (#13629)
- enable stepwise execution of query pipelines (#14117)
- Replace tenacity upper limit by only rejecting 8.4.0 (#14218)
- propagate error_on_no_tool_call kwarg in
llm.predict_and_call()
(#14253) - in query pipeline, avoid casting nodes as strings and use
get_content()
instead (#14242) - Fix NLSQLTableQueryEngine response metadata (#14169)
- do not overwrite relations in default simple property graph (#14244)
- Enable selecting Intel GPU for ipex embedding integrations (#14214)
- add mixedbread ai integration (#14161)
- Add default node property to neo4j upsert relations (#14095)
- Added re-ranking into the PostgresML Managed Index (#14134)
- use async AI21 client for async methods (#14193)
- Added (fake) async calls to avoid errors (#14241)
- Add function calling to deep infra llm (#14127)
- Enable selecting Intel GPU for ipex embedding integrations (#14214)
- add command r support oci genai (#14080)
- Prem AI Templates Llama Index support (#14105)
- Integrate You.com conversational APIs (#14207)
- Add metadata field "collection_name" to SimpleMongoReader (#14245)
- add marker-pdf reader (#14099)
- Added upstage as a reader (#13415)
- add mixedbread ai integration (#14161)
- LanceDB: code cleanup, minor updates (#14077)
- add option to customize default OpenSearch Client and Engine (#14249)
- Fix Pin tenacity and numpy in core (#14203)
- Add precision and recall metrics (#14170)
- Enable Function calling and agent runner for Vertex AI (#14088)
- Fix for batch_gather (#14162)
- Remove sentence-transformers dependency from HuggingFace utils package (#14204)
- Add MistralAI Finetuning API support (#14101)
- Update MistralAI (#14199)
- fix: 🐛 Fix Bedrock Converse' pyproject.toml for the PyPI release (#14197)
- Use typical include llama_index/ (#14196)
- Feature/azure_table_storage (#14182)
- add support for nvidia/nv-embed-v1 (https://huggingface.co/nvidia/NV-Embed-v1) (#14194)
- add news retriever (#13934)
- Fixes a bug where there is a missing await. (#14177)
- Restore Nomic Embed einops dependency (#14176)
- Changing BM25Retriever _retrieve to use numpy methods (#14015)
- Add missing @llm_chat_callback() to Gemini.stream_chat (#14166)
- Enable Function calling and agent runner for Vertex AI (#14088)
- feat: support VectorStoreQueryMode.TEXT_SEARCH on OpenSearch VectorStore (#14153)
- Fix parsing sql query.py (#14109)
- Implement NDCG metric (#14100)
- Fixed System Prompts for Structured Generation (#14026)
- Split HuggingFace embeddings in HuggingFace API and TextGenerationInference packages (#14013)
- Add PandasExcelReader class for parsing excel files (#13991)
- feat: add spans to ingestion pipeline (#14062)
- Fix Qdrant nodes (#14149)
- Fixes TypeError: sequence item : expected str instance, int found
- feat: Add Managed Index for LlamaIndex on Vertex AI for RAG (#13626)
- Feature/add command r support oci genai (#14080)
- MilvusVectorStore: always include text_key in output_fields (#14076)
- Add Mixture Of Agents paper implementation (#14112)
- Split HuggingFace embeddings in HuggingFace API and TextGenerationInference packages (#14013)
- Split HuggingFace embeddings in HuggingFace API and TextGenerationInference packages (#14013)
- Split HuggingFace embeddings in HuggingFace API and TextGenerationInference packages (#14013)
- Split HuggingFace embeddings in HuggingFace API and TextGenerationInference packages (#14013)
- Feat: IBM watsonx.ai llm and embeddings integration (#13600)
- Feat: IBM watsonx.ai llm and embeddings integration (#13600)
- Feat: IBM watsonx.ai llm and embeddings integration (#13600)
- Fix to milvus filter enum parsing (#14111)
- fix anthropic llm calls (#14108)
- Wrong mongo name was used instead of Postgres (#14107)
- Remove unnecessary excluded from fields in Bedrock embedding (#14085)
- Feature/added trust remote code (#14102)
- nit: fix for pandas excel reader (#14086)
- Update anthropic dependency to 0.26.2 minimum version (#14091)
- Add support for Llama 3 Instruct prompt format (#14072)
- Implement Bedrock Converse API for function calling (#14055)
- fix/postgres-metadata-in-filter-single-elem (#14035)
- Add PandasExcelReader class for parsing excel files (#13991)
- Update dependency of llama-index-embeddings-ipex-llm
- Add api key as field in Gemini Embedding (#14061)
- Expand milvus vector store filter options (#13961)
- Add WEBP and GIF to supported image types for SimpleDirectoryReader (#14038)
- refactor: add spans to abstractmethods via mixin (#14003)
- Adding streaming support for SQLAutoVectorQueryEngine (#13947)
- add option to specify embed_model to NLSQLTableQueryEngine (#14006)
- add spans for multimodal LLMs (#13966)
- change to compact in auto prev next (#13940)
- feat: add exception events for streaming errors (#13917)
- feat: add spans for tools (#13916)
- Fix error when using azure_ad without setting the API key (#13970)
- add Jina Embeddings MultiModal (#13861)
- Add Nomic multi modal embeddings (#13920)
- ensure cypher returns list before iterating (#13938)
- Add AI21 Labs Jamba-Instruct Support (#14030)
- fix(deepinfrallm): default max_tokens (#13998)
- correct
__del__()
Vllm (#14053)
- Add ZenGuard llamapack (#13959)
- fix how class attributes are set in google drive reader (#14022)
- Add Google Maps Text Search Reader (#13884)
- Jira personal access token with hosted instances (#13890)
- set document ids when loading (#14000)
- Add DuckDBRetriever (#13929)
- Add inclusion filter to chromadb (#14010)
- Fix LanceDBVectorStore
add()
logic (#13993)
- Support all filter operators for Milvus vector store (#13745)
- Broaden SQLAlchemy support in llama-index-vector-stores-postgres to 1.4+ (#13936)
- Qdrant: Create payload index for
doc_id
(#14001)
- use default UUIDs when possible for property graph index vector stores (#13886)
- avoid empty or duplicate inserts in property graph index (#13891)
- Fix cur depth for
get_rel_map
in simple property graph store (#13888) - (bandaid) disable instrumentation from logging generators (#13901)
- Add backwards compatibility to Dispatcher.get_dispatch_event() method (#13895)
- Fix: Incorrect naming of acreate_plan in StructuredPlannerAgent (#13879)
- Handle cases where type is missing (neo4j property graph) (#13875)
- Rename
Neo4jPGStore
toNeo4jPropertyGraphStore
(with backward compat) (#13891)
- Improve the retry mechanism of OpenAI (#13878)
- AsyncWebPageReader: made it actually async; it was exhibiting blocking behavior (#13897)
- Fix/OpenSearch filter logic (#13804)
- Allow proper setting of the vector store in property graph index (#13816)
- fix imports in langchain bridge (#13871)
- NebulaGraph support for PropertyGraphStore (#13816)
- fix fireworks imports in langchain llm (#13871)
- feat(openllm): 0.5 sdk integrations update (#13848)
- Update SDK compatibility (#13836)
- Fixed a bug with tokens causing an infinite loop in GoogleDriveReader (#13863)
- pass embeddings from index to property graph retriever (#13843)
- protect instrumentation event/span handlers from each other (#13823)
- add missing events for completion streaming (#13824)
- missing callback_manager.on_event_end when there is exception (#13825)
- use
model
kwarg for model name for gemini (#13791)
- Add mistral code model (#13807)
- update mistral codestral with fill in middle endpoint (#13810)
- 0.5 integrations update (#13848)
- Safety setting for Pydantic Error for Vertex Integration (#13817)
- handle path objects in smart pdf reader (#13847)
- Added
PropertyGraphIndex
and other supporting abstractions. See the full guide for more details (#13747) - Updated
AutoPrevNextNodePostprocessor
to allow passing in response mode and LLM (#13771) - fix type handling with return direct (#13776)
- Correct the method name to
_aget_retrieved_ids_and_texts
in retrievval evaluator (#13765) - fix: QueryTransformComponent incorrect call
self._query_transform
(#13756) - implement more filters for
SimpleVectorStoreIndex
(#13365)
- Added support for Bedrock Titan Embeddings v2 (#13580)
- add Oracle Cloud Infrastructure (OCI) Generative AI (#13631)
- Expose "safe_serialization" parameter from AutoModel (#11939)
- Added
Neo4jPGStore
for property graph support (#13747)
- Added dashscope managed index (#13378)
- add Oracle Cloud Infrastructure (OCI) Generative AI (#13631)
- fix undefined variable (#13768)
- SecGPT - LlamaIndex Integration #13127
- Add Hologres vector db (#13619)
- Remove FlagEmbedding as Milvus's dependency (#13767) Unify the collection construction regardless of the value of enable_sparse (#13773)
- refactor to put helper methods inside class definition (#13749)
- Add VectorMemory and SimpleComposableMemory (#13352)
- Improve MarkdownReader to ignore headers in code blocks (#13694)
- proper async element node parsers (#13698)
- return only the message content in function calling worker (#13677)
- nit: fix multimodal query engine to use metadata (#13712)
- Add notebook with workaround for lengthy tool descriptions and QueryPlanTool (#13701)
- Improve device selection (#13644)
- Add the PostgresML Managed Index (#13623)
- Added chat engine, streaming, factual consistency score, and more (#13639)
- Add Integration for DeepInfra LLM Models (#13652)
- add GPU support for llama-index-llm-ipex-llm (#13691)
- lmstudio integration (#13557)
- Use aiter_lines function to iterate over lines in ollama integration (#13699)
- Added safety_settings parameter for gemini (#13568)
- VoyageAI reranking bug fix (#13622)
- Add missing return (#13720)
- Add Scrapfly Web Loader (#13654)
- fix bug with delete and special chars (#13651)
- Try-catch in case the ._client attribute is not present (#13681)
- Enabling streaming in BaseSQLTableQueryEngine (#13599)
- Fix nonetype errors in relational node parsers (#13615)
- feat(instrumentation): new spans for ALL llms (#13565)
- Properly Limit the number of generated questions (#13596)
- Pass 'exclude_llm_metadata_keys' and 'exclude_embed_metadata_keys' in element Node Parsers (#13567)
- Add batch mode to QueryPipeline (#13203)
- Improve SentenceEmbeddingOptimizer to respect Settings.embed_model (#13514)
- ReAct output parser robustness changes (#13459)
- fix for pydantic tool calling with a single argument (#13522)
- Avoid unexpected error when stream chat doesn't yield (#13422)
- Implement local Nomic Embed with the inference_mode parameter (#13607)
- Deprecate
mode()
in favor of__init__(base_url=...)
(#13572) - add snowflake/arctic-embed-l support (#13555)
- update how retries get triggered for openai (#13608)
- Integrations: upstage LLM and Embeddings (#13193)
- feat: add gemini new models to multimodal LLM and regular (#13539)
- fix: enable tool use (#13566)
- Add support for lmstudio integration (#13557)
- Deprecate
mode()
in favor of__init__(base_url=...)
(#13572)
- update how retries get triggered for openai (#13608)
- Add Unify LLM Support (#12921)
- Integrations: upstage LLM and Embeddings (#13193)
- Adding Support for MedLM Models (#11911)
- Add dashscope rerank for postprocessor (#13353)
- Deprecate
mode()
in favor of__init__(base_url=...)
(#13572)
- SimpleMongoReader should allow optional fields in metadata (#13575)
- fix: (ArxivReader) set exclude_hidden to False when reading data from hidden directory (#13578)
- fix: sec_filings header when making requests to sec.gov #13548
- Added firecrawl search mode (#13560)
- Updated Browserbase web reader (#13535)
- added Cassandra database tool spec for agents (#13423)
- Allow querying AzureAISearch without non-null metadata field (#13531)
- Integrate VectorStore from Elasticsearch client (#13291)
- Fix the filter expression construction of Milvus vector store (#13591)
- Disconnect when deleted (#13611)
- Added the WordLift Vector Store (#13028)
- Add image_documents at call time for
MultiModalLLMCompletionProgram
(#13467) - fix RuntimeError by switching to asyncio from threading (#13486)
- Add support for prompt kwarg (#13405)
- VectorStore -> BasePydanticVectorStore (#13439)
- fix: user_message does not exist bug (#13432)
- import missing response type (#13382)
- add
CallbackManager
toMultiModalLLM
(#13400)
- Remove "Truncate" parameter from Bedrock Cohere invoke model request (#13442)
- Trafilatura kwargs and progress bar for trafilatura web reader (#13454)
- Fix #9522 - SQLAlchemy warning when using hybrid search (#13476)
- Fix #9522 - SQLAlchemy warning when using hybrid search (#13476)
- update UpTrain Callback Handler to support new Upgratin eval schema (#13479)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- VectorStore -> BasePydanticVectorStore (#13439)
- bump rerank versions (#13465)
- bump rerank versions (#13465)
- bump rerank versions (#13465)
- bump rerank versions (#13465)
- bump rerank versions (#13465)
- bump rerank versions (#13465)
- bump rerank versions (#13465)
- bump rerank versions (#13465)
- bump rerank versions (#13465)
- bump rerank versions (#13465)
- gpt-4o support (#13463)
- gpt-4o support (#13463)
- fix the RAG fusion pipeline (#13413)
- fix: update OpenAIAssistantAgent to use attachments (#13341)
- new embeddings integration (#13323)
- new llm integration (#13231)
- fix: Corrected connection parameters in connections.connect() (#13448)
- fix AzureAiSearchVectorStore metadata f-string (#13435)
- adds Unit and Integration tests for MongoDBAtlasVectorSearch (#12854)
- update llama-index-llms-huggingface dependency (#13420)
- new vector store integration
- Implement async methods in RedisKVStore (#12943)
- pin llama-index-llms-cohere dependency (#13417)
- pin cohere dependency (#13417)
- fix indexing issue and runtime error message (#13414)
- fix Cohere Rerank bug (#13410)
- fix retriever integration (#13409)
- new tool
- fix missing authorized_user_info check on GoogleDriveReader (#13394)
- await Firestore's AsyncDocumentReference (#13386)
- add dynamic model listing support (#13398)
- add start_char_idx and end_char_idx with MarkdownElementParser (#13377)
- use handlers from global default (#13368)
- Initial release (#13128)
- Call Cohere RAG inference with documents argument (#13196)
- update add method decode utf-8 (#13194)
- Initial release (#13286)
- update tool to use updated api/sdk (#13373)
- Update to weaviate client v4 (#13229)
- fix bug where PDFReader ignores extra_info (#13369)
- Add sync httpx client support (#13370)
- Add sync httpx client support (#13370)
- Add missing openai model token context (#13337)
- Add fail_on_http_error (#13366)
- Add attribution tag for pinecone (#13329)
- set default max_tokens to 1024 (#13371)
- Fix hiddent temp directory issue for arxiv reader (#13351)
- fix truncate passing aget_query_embedding and get_text_embedding (#13367)
- Add llama-3 models (#13336)
- Add CRITIC and reflection agent integrations (#13108)
- fix
from_defaults()
erasing summary memory buffer history (#13325) - use existing async event loop instead of
asyncio.run()
in core (#13309) - fix async streaming from query engine in condense question chat engine (#13306)
- Handle ValueError in extract_table_summaries in element node parsers (#13318)
- Handle llm properly for QASummaryQueryEngineBuilder and RouterQueryEngine (#13281)
- expand instrumentation payloads (#13302)
- Fix Bug in sql join statement missing schema (#13277)
- add encoding_type parameters in JinaEmbedding class (#13172)
- fix encoding type access in JinaEmbeddings (#13315)
- add nvidia nim embeddings support (#13177)
- Fix async issue when streaming with Mistral AI (#13292)
- add nvidia nim llm support (#13176)
- add nvidia nim rerank support (#13178)
- Update MarkdownReader to parse text before first header (#13327)
- feat: Spider Web Loader (#13200)
- Add VectorStore integration for Vespa (#13213)
- Add support for Vertex AI Vector Search as Vector Store (#13186)
- remove error ignoring during chat engine streaming (#13160)
- add structured planning agent (#13149)
- update base class for planner agent (#13228)
- Fix: Error when parse file using SimpleFileNodeParser and file's extension doesn't in FILE_NODE_PARSERS (#13156)
- add matching
source_node.node_id
verification to node parsers (#13109) - Retrieval Metrics: Updating HitRate and MRR for Evaluation@K documents retrieved. Also adding RR as a separate metric (#12997)
- Add chat summary memory buffer (#13155)
- ZillizCloudPipelineIndex accepts flexible params to create pipelines (#10134, #10112)
- Add tool usage support with text-generation-inference integration from Hugging Face (#12471)
- Add streaming for maritalk (#13207)
- Integrate mistral.rs LLM (#13105)
- mymagicai api update (#13148)
- Streaming Support for Nvidia's Triton Integration (#13135)
- added async support to ollama llms (#13150)
- Exclude access control metadata keys from LLMs and embeddings - SharePoint Reader (#13184)
- feat: Browserbase Web Reader (#12877)
- Added YouTube Metadata Reader (#12975)
- fix redis kvstore key that was in bytes (#13201)
- Respect filter condition for Azure AI Search (#13215)
- small bump for new chroma client version (#13158)
- Adding Firestore Vector Store (#12048)
- small fix to returned IDs after
add()
(#12515)
- Add hybrid retrieval mode to MilvusVectorStore (#13122)
- parameterize queries in pgvector store (#13199)
- add agent_worker.as_agent() (#13061)
- Use Bedrock cohere character limit (#13126)
- Change default value for attendees to empty list (#13134)
- Skip index creation error when index already exists (#13085)
- Fix datetime for google calendar create_event api (#13132)
- Merge multiple prompts into one (#13131)
- Use MetadataFilters in LlamaCloud Retriever (#13117)
- Fix kuzu integration .execute() calls (#13100)
- Maintenance update to keep up to date with lantern builds (#13116)
- Corrected wrong output type for
OutputKeys.from_keys()
(#13086) - add run_jobs to aws base embedding (#13096)
- allow user to customize the keyword extractor prompt template (#13083)
- (CondenseQuestionChatEngine) Do not condense the question if there's no conversation history (#13069)
- QueryPlanTool: Execute tool calls in subsequent (dependent) nodes in the query plan (#13047)
- Fix for fusion retriever sometime return Nonetype query(s) before similarity search (#13112)
- Support llama-index-embeddings-ipex-llm for Intel GPUs (#13097)
- Fix bug in raft dataset generator - multiple system prompts (#12751)
- Add access control related metadata to SharePoint reader (#13067)
- Nested metadata filter support (#13113)
- Nested metadata filter support (#13113)
- fix async streaming response from query engine (#12953)
- enforce uuid in element node parsers (#12951)
- add function calling LLM program (#12980)
- make the PydanticSingleSelector work with async api (#12964)
- fix query pipeline's arun_with_intermediates (#13002)
- Add COA Agent integration (#13043)
- Official LATs agent integration (#13031)
- Add LLMCompiler Agent Integration (#13044)
- Add the ability to pass custom headers to Anthropic LLM requests (#12819)
- Adding claude 3 opus to BedRock integration (#13033)
- Add new Llama 3 and Mixtral 8x22b model into Llama Index for Fireworks (#12970)
- Fix AsyncOpenAI "RuntimeError: Event loop is closed bug" when instances of AsyncOpenAI are rapidly created & destroyed (#12946)
- Don't retry on all OpenAI APIStatusError exceptions - just InternalServerError (#12947)
- Updated IBM watsonx foundation models (#12973)
- Return the parent node if the query node is not present (#12983)
- fixed bug when function is defined twice (#12941)
- dding support for streaming partial instances of Pydantic output class in OpenAIPydanticProgram (#13021)
- add reader for openapi files (#12998)
- Avoid infinite loop when not handled exception is raised (#12963)
- Improve whole site reader to remove duplicate links (#12977)
- Fix Bedrock KB retriever to use query bundle (#12910)
- Integrating AWS DocumentDB as a vector storage method (#12217)
- Fix databricks vector search metadata (#12999)
- Neo4j metadata filtering support (#12923)
- Fix error querying PineconeVectorStore using sparse query mode (#12967)
- Many fixes for async and checking if collection exists (#12916)
- Adds the index deletion functionality to the WeviateVectoreStore (#12993)
- Add intermediate outputs to QueryPipeline (#12683)
- Fix show progress causing results to be out of order (#12897)
- add OR filter condition support to simple vector store (#12823)
- improved custom agent init (#12824)
- fix pipeline load without docstore (#12808)
- Use async
_aprocess_actions
in_arun_step_stream
(#12846) - provide the exception to the StreamChatErrorEvent (#12879)
- fix bug in load and search tool spec (#12902)
- Expose azure_ad_token_provider argument to support token expiration (#12818)
- Add httpx_async_client option (#12896)
- add ipex-llm embedding integration (#12740)
- add octoai embeddings (#12857)
- Expose azure_ad_token_provider argument to support token expiration (#12818)
- add support for loading "low-bit format" model to IpexLLM integration (#12785)
- support
open-mixtral-8x22b
(#12894)
- added LATS agent pack (#12735)
- Use passed in metadata for documents (#12844)
- added Firecrawl Web Loader (#12825)
- use batch insertions into Milvus vector store (#12837)
- add vearch to vector stores (#10972)
- BREAKING Moved
PandasQueryEngine
andPandasInstruction
parser tollama-index-experimental
(#12419)- new install:
pip install -U llama-index-experimental
- new import:
from llama_index.experimental.query_engine import PandasQueryEngine
- new install:
- Fixed some core dependencies to make python3.12 work nicely (#12762)
- update async utils
run_jobs()
to include tqdm description (#12812) - Refactor kvdocstore delete methods (#12681)
- Support for Mistral Large from Bedrock (#12804)
- Added OpenVino LLMs (#12639)
- Update LlamaIndex-Predibase Integration to latest API (#12736)
- Enable choice of either Predibase-hosted or HuggingFace-hosted fine-tuned adapters in LlamaIndex-Predibase integration (#12789)
- Modernize GuardrailsOutputParser (#12676)
- Chain-of-Abstraction Agent Pack (#12757)
- Fixed issue with chunking multi-byte characters (#12715)
- Fix bug in raft dataset generator - multiple system prompts (#12751)
- Add openvino rerank support (#12688)
- convert to Path in docx reader if input path str (#12807)
- make pip check work for optional pdf packages (#12758)
- wrong doc id when using default s3 endpoint in S3Reader (#12803)
- Add Amazon Bedrock knowledge base integration as retriever (#12737)
- Add mongodb atlas bm25 retriever (#12519)
- fix message serialization in redis chat store (#12802)
- Relax dependency version to accept astrapy
1.*
(#12792)
- Add support for Couchbase as a Vector Store (#12680)
- Fix elasticsearch hybrid rrf window_size (#12695)
- Added support to retrieve metadata fields from milvus (#12626)
- Modernize redis vector store, use redisvl (#12386)
- refactor: Switch default Qdrant sparse encoder (#12512)
- Support indented code block fences in markdown node parser (#12393)
- Pass in output parser to guideline evaluator (#12646)
- Added example of query pipeline + memory (#12654)
- Add missing node postprocessor in CondensePlusContextChatEngine async mode (#12663)
- Added
return_direct
option to tools /tool metadata (#12587) - Add retry for batch eval runner (#12647)
- Thread-safe instrumentation (#12638)
- Coroutine-safe instrumentation Spans #12589
- Add in-memory loading for non-default filesystems in PDFReader (#12659)
- Remove redundant tokenizer call in sentence splitter (#12655)
- Add SynthesizeComponent import to shortcut imports (#12655)
- Improved truncation in SimpleSummarize (#12655)
- adding err handling in eval_utils default_parser for correctness (#12624)
- Add async_postprocess_nodes at RankGPT Postprocessor Nodes (#12620)
- Fix MarkdownNodeParser ref_doc_id (#12615)
- Added initial support for openvino embeddings (#12643)
- add anthropic tool calling (#12591)
- add ipex-llm integration (#12322)
- add more data types support to ipex-llm llm integration (#12635)
- Proper PrivateAttr usage in OpenLLM (#12655)
- Bumped anthropic dep version (#12655)
- bump generativeai dep (#12645)
- Add streaming support for DenseXRetrievalPack (#12607)
- Improve efficiency of MongoDB reader (#12664)
- Added multilingual support for the Wikipedia reader (#12616)
- remove invalid chars from default collection name (#12672)
- Added support to retrieve metadata fields from milvus (#12626)
- Bug fix - Similarity metric is always IP for MilvusVectorStore (#12611)
- Update imports for message thread typing (#12437)
- Fix for pydantic query engine outputs being blank (#12469)
- Add span_id attribute to Events (instrumentation) (#12417)
- Fix RedisDocstore node retrieval from docs property (#12324)
- Add node-postprocessors to retriever_tool (#12415)
- FLAREInstructQueryEngine : delegating retriever api if the query engine supports it (#12503)
- Make chat message to dict safer (#12526)
- fix check in batch eval runner for multi-kwargs (#12563)
- Fixes agent_react_multimodal_step.py bug with partial args (#12566)
- Added support to load clip model from local file path (#12577)
- text embedding integration: Cloudflare Workers AI (#12446)
- Fix pydantic issue in class definition (#12469)
- Small typo fix in QA generation prompt (#12470)
- Replace redis driver with FalkorDB driver (#12434)
- Add ability to pass custom HTTP headers to Anthropic client (#12558)
- Add support for Cohere Command R+ model (#12581)
- Integrations with DataBricks LLM API (#12432)
- Updated Watsonx foundation models (#12493)
- Updated base model name on watsonx integration #12491
- Add RankGPT support inside RankLLM (#12475)
- Use recursive strategy by default for SharePoint (#12557)
- Readability web page reader fix playwright async api bug (#12520)
- small
to_list
fix (#12515)
- Add support for Neptune Analytics as a Vector Store (#12423)
- fix(postgres): numeric metadata filters (#12583)
- pass proper query bundle in QueryFusionRetriever (#12387)
- Update llama_parse_json_element.py to fix error on lists (#12402)
- Add node postprocessors to retriever tool (#12415)
- Fix bug where user specified llm is not respected in fallback logic in element node parsers(#12403)
- log proper LLM response key for async callback manager events (#12421)
- Deduplicate the two built-in react system prompts; Also make it read from a Markdown file (#12307)
- fix bug in BatchEvalRunner for multi-evaluator eval_kwargs_lists (#12418)
- add the callback manager event for vector store index insert_nodes (#12443)
- fixes an issue with serializing chat messages into chat stores when they contain pydantic API objects (#12394)
- fixes an issue with slow memory.get() operation (caused by multiple calls to get_all()) (#12394)
- fixes an issue where an agent+tool message pair is cut from the memory (#12394)
- Added
FnNodeMapping
for object index (#12391) - Make object mapping optional / hidden for object index (#12391)
- Make object index easier to create from existing vector db (#12391)
- When LLM failed to follow the react response template, tell it so #12300
- Bump cohere version to 5.1.1 (#12279)
- add Intel Extension for Transformers embedding model (#12410)
- make neo4j query insensitive (#12337)
- Bump cohere version to 5.1.1 (#12279)
- add ipex-llm integration (#12322)
- Fix litellm ChatMessage role validation error (#12449)
- Use
FunctionCallingLLM
base class in OpenAI (#12227)
- Fix llama-index-core dep (#12374)
- Bump cohere version to 5.1.1 (#12279)
- Added RankLLM rerank (#12296)
- RankLLM fixes (#12399)
- Fixed bug with path names (#12366)
- Add AnalyticDB VectorStore (#12230)
- Fixed typo in imports/readme (#12370)
- add
in
filter operator for qdrant (#12376)
- Add score to NodeWithScore in KnowledgeGraphQueryEngine (#12326)
- Batch eval runner fixes (#12302)
- Added support for binary / quantized embeddings (#12321)
- add support for custom endpoints to MistralAI (#12328)
- Fix RedisDocstore node retrieval from docs property (#12324)
- pretty prints in
LlamaDebugHandler
(#12216) - stricter interpreter constraints on pandas query engine (#12278)
- PandasQueryEngine can now execute 'pd.*' functions (#12240)
- delete proper metadata in docstore delete function (#12276)
- improved openai agent parsing function hook (#12062)
- add raise_on_error flag for SimpleDirectoryReader (#12263)
- remove un-caught openai import in core (#12262)
- Fix download_llama_dataset and download_llama_pack (#12273)
- Implement EvalQueryEngineTool (#11679)
- Expand instrumenation Span coverage for AgentRunner (#12249)
- Adding concept of function calling agent/llm (mistral supported for now) (#12222, )
- Use
sentence-transformers
as a backend (#12277)
- Added voyageai as a reranker (#12111)
- Added google cloud storage reader (#12259)
- Support for different drives (#12146)
- Remove unnecessary PyDrive dependency from Google Drive Reader (#12257)
- added readme.com reader (#12246)
- added pack for RAFT (#12275)
- Added
(a)predict_and_call()
function to base LLM class + openai + mistralai (#12188) - fixed bug with
wait()
in async agent streaming (#12187)
- Added alephalpha embeddings (#12149)
- Added alephalpha LLM (#12149)
- fixed bug with
wait()
in async agent streaming (#12187)
- fixed import errors in docugami reader (#12154)
- fix PDFReader for remote fs (#12186)
- Updated docs backend from sphinx to mkdocs, added ALL api reference, some light re-org, better search (#11301)
- Added async loading to
BaseReader
class (although its fake async for now) (#12156) - Fix path implementation for non-local FS in
SimpleDirectoryReader
(#12141) - add args/kwargs to spans, payloads for retrieval events, in instrumentation (#12147)
- [react agent] Upon exception, say so, so that Agent can correct itself (#12137)
- Added rate limit handling (#12127)
- Add Amazon Neptune Support as Graph Store (#12097)
- fix VllmServer to work without CUDA-required vllm core (#12003)
- Use S3FS in S3Reader (#12061)
- Added proper kvstore dep (#12157)
- Added proper kvstore dep (#12157)
- fix unclosed session in es add function #12135
- Add support for
KDBAIVectorStore
(#11967)
- Lazy init for async elements StreamingAgentChatResponse (#12116)
- Fix streaming generators get bug by SynthesisEndEvent (#12092)
- CLIP embedding more models (#12063)
- Add
num_workers
to summary module (#)
- Fix datetime fields (#12112)
- Add ability to select time period of posts/messages (#12078)
- Add api version/base api as optional for open ai embedding (#12091)
- Add node postprocessing to network retriever (#12027)
- Add privacy-safe networks demo (#12027)
- Chore: bumps min version of langfuse dep (#12077)
- Chore: bumps google-generativeai dep (#12085)
- Chore: bumps google-generativeai dep (#12085)
- Chore: bumps google-generativeai dep (#12085)
- Chore: bumps google-generativeai dep (#12085)
- Chore: bumps google-generativeai dep (#12085)
- Chore: bumps google-generativeai dep (#12085)
- New integration (#12068)
- Fix - Google Drive Issue of not loading same name files (#12022)
- Adding Metadata Filtering support for UpstashVectorStore (#12054)
- Fix: prevent RaptorPack infinite recursion (#12008)
- Fix(OptimumEmbedding): removing token_type_ids causing ONNX validation issues
- Fix: Anthropic LLM merge consecutive messages with same role (#12013)
- DiffPrivacy ICL Pack - OpenAI Completion LLMs (#11881)
- Remove llama_hub_url keyword from download_llama_dataset of command (#12038)
- New
instrumentation
module for observability (#11831) - Allow passing in LLM for
CitationQueryEngine
(#11914) - Updated keyval docstore to allow changing suffix in addition to namespace (#11873)
- Add (some) async streaming support to query_engine #11949
- Fixed embedding type for query texts (#11901)
- Support for premai embeddings (#11954)
- Added support for network retrievers (#11800)
- Added support for haiku (#11916)
- Fixed import error for ChatMessage (#11902)
- added gpt-35-turbo-0125 for AZURE_TURBO_MODELS (#11956)
- fixed error with nontype in logprobs (#11967)
- Support for premai llm (#11954)
- Support for solar as an LLM class (#11710)
- Add support for medlm in vertex (#11911)
- added README files and query string for google drive reader (#11724)
- Updated ImageReader to add
plain_text
option to trigger pytesseract (#11913)
- use pure requests to reduce deps, simplify code (#11924)
- use pure requests to reduce deps, simplify code (#11924)
- Allow changing suffix for mongodb docstore (#11873)
- Support for databricks vector search as a vector store (#10754)
- (re)implement proper delete (#11959)
- Fixes for IN filters and OR text search (#11872, #11927)
- Removed chroma as a bundled dep to reduce
llama-index
deps
- Introduce retries for rate limits in
OpenAI
llm class (#11867) - Added table comments to SQL table schemas in
SQLDatabase
(#11774) - Added
LogProb
type toChatResponse
object (#11795) - Introduced
LabelledSimpleDataset
(#11805) - Fixed insert
IndexNode
objects with unserializable objects (#11836) - Fixed stream chat type error when writing response to history in
CondenseQuestionChatEngine
(#11856) - Improve post-processing for json query engine (#11862)
- Fixed async kwarg error (#11822)
- Fixed pydantic import (#11765)
- Properly close connection after verifying connectivity (#11821)
- Add support for new
command-r
model (#11852)
- Fixed streaming decoding with special tokens (#11807)
- Added support for latest and open models (#11792)
- Fixed small bug when passing in the API get for stock news (#11772)
- Slimmed down chroma deps (#11775)
- Fixes for deleting (#11825)
- Support for nested metadata filters (#11778)
- Update import path for callback handler (#11754)
- Ensure
LoadAndSearchToolSpec
loads document objects (#11733) - Fixed bug for no nodes in
QueryFusionRetriever
(#11759) - Allow using different runtime kwargs for different evaluators in
BatchEvalRunner
(#11727) - Fixed issues with fsspec +
SimpleDirectoryReader
(#11665) - Remove
asyncio.run()
requirement from guideline evaluator (#11719)
- Update voyage embeddings to use proper clients (#11721)
- Fixed issues with vectara query engine in non-summary mode (#11668)
- Add
return_output
option for json output with query and response (#11761)
- Added support for a
CodeHiearchyAgentPack
that allows for agentic traversal of a codebase (#10671)
- Added a new llama-pack for citations + chat with cohere (#11697)
- Prevent forced
flush()
on document add (#11734)
- Small typo in metadata column name (#11751)
- Initial support for TiDB vector store (#11635)
- Small fix for
int
fields in metadata filters (#11742)
New format! Going to try out reporting changes per package.
- Update mappings for
upgrade
command (#11699)
- add
relative_score
anddist_based_score
toQueryFusionRetriever
(#11667) - check for
none
in async agent queue (#11669) - allow refine template for
BaseSQLTableQueryEngine
(#11378) - update mappings for llama-packs (#11699)
- fixed index error for extracting rel texts in KG index (#11695)
- return proper response types from synthesizer when no nodes (#11701)
- Inherit metadata to summaries in DocumentSummaryIndex (#11671)
- Inherit callback manager in sql query engines (#11662)
- Fixed bug with agent streaming not being written to chat history (#11675)
- Fixed a small bug with
none
deltas when streaming a function call with an agent (#11713)
- Added support for new multi-modal models
haiku
andsonnet
(#11656)
- Added a new llama-pack for hierarchical agents + finance chat (#11387)
- Added support for checking if NLTK files are already downloaded (#11676)
- Use the metadata passed in when creating documents (#11626)
- Update wording in warning message (#11702)
- Avoid calling
nest_asyncio.apply()
in code to avoid confusing errors for users (#11707)
- Catch RPC errors (#11657)
- Anthropic support for new models (#11623, #11612)
- Easier creation of chat prompts (#11583)
- Added a raptor retriever llama-pack (#11527)
- Improve batch cohere embeddings through bedrock (#11572)
- Added support for vertex AI embeddings (#11561)
- Ensure order in async embeddings generation (#11562)
- Fixed empty metadata for csv reader (#11563)
- Serializable fix for composable retrievers (#11617)
- Fixed milvus metadata filter support (#11566)
- FIxed pydantic import in clickhouse vector store (#11631)
- Fixed system prompts for gemini/vertext-gemini (#11511)
- Added FeishuWikiReader (#11491)
- Added videodb retriever integration (#11463)
- Added async to opensearch vector store (#11513)
- New LangFuse one-click callback handler (#11324)
- Fixed deadlock issue with async chat streaming (#11548)
- Improved hidden file check in SimpleDirectoryReader (#11496)
- Fixed null values in document metadata when using SimpleDirectoryReader (#11501)
- Fix for sqlite utils in jsonalyze query engine (#11519)
- Added base url and timeout to ollama multimodal LLM (#11526)
- Updated duplicate handling in query fusion retriever (#11542)
- Fixed bug in kg indexx struct updating (#11475)
- Released llama-index-networks (#11413)
- Jina reranker (#11291)
- Added DuckDuckGo agent search tool (#11386)
- helper functions for chatml (#10272)
- added brave search tool for agents (#11468)
- Added Friendli LLM integration (#11384)
- metadata only queries for chromadb (#11328)
- Fixed inheriting llm callback in synthesizers (#11404)
- Catch delete error in milvus (#11315)
- Fixed pinecone kwargs issue (#11422)
- Supabase metadata filtering fix (#11428)
- api base fix in gemini embeddings (#11393)
- fix elasticsearch vector store await (#11438)
- vllm server cuda fix (#11442)
- fix for passing LLM to context chat engine (#11444)
- set input types for cohere embeddings (#11288)
- default value for azure ad token (#10377)
- added back prompt mixin for react agent (#10610)
- fixed system roles for gemini (#11481)
- fixed mean agg pooling returning numpy float values (#11458)
- improved json path parsing for JSONQueryEngine (#9097)
- Added a llama-pack for KodaRetriever, for on-the-fly alpha tuning (#11311)
- Added support for
mistral-large
(#11398) - Last token pooling mode for huggingface embeddings models like SFR-Embedding-Mistral (#11373)
- Added fsspec support to SimpleDirectoryReader (#11303)
- Fixed an issue with context window + prompt helper (#11379)
- Moved OpenSearch vector store to BasePydanticVectorStore (#11400)
- Fixed function calling in fireworks LLM (#11363)
- Made cohere embedding types more automatic (#11288)
- Improve function calling in react agent (#11280)
- Fixed MockLLM imports (#11376)
- Added
llama-index-postprocessor-colbert-rerank
package (#11057) MyMagicAI
LLM (#11263)MariaTalk
LLM (#10925)- Add retries to github reader (#10980)
- Added FireworksAI embedding and LLM modules (#10959)
- Fixed string formatting in weaviate (#11294)
- Fixed off-by-one error in semantic splitter (#11295)
- Fixed
download_llama_pack
for multiple files (#11272) - Removed
BUILD
files from packages (#11267) - Loosened python version reqs for all packages (#11267)
- Fixed args issue with chromadb (#11104)
- Fixed multi-modal LLM for async acomplete (#11064)
- Fixed issue with llamaindex-cli imports (#11068)
I'm still a bit wonky with our publishing process -- apologies. This is just a version bump to ensure the changes that were supposed to happen in 0.10.9 actually did get published. (AF)
- add llama-index-cli dependency
- Added Self-Discover llamapack (#10951)
- Fixed linting in CICD (#10945)
- Fixed using remote graph stores (#10971)
- Added missing LLM kwarg in NoText response synthesizer (#10971)
- Fixed openai import in rankgpt (#10971)
- Fixed resolving model name to string in openai embeddings (#10971)
- Off by one error in sentence window node parser (#10971)
First, apologies for missing the changelog the last few versions. Trying to figure out the best process with 400+ packages.
At some point, each package will have a dedicated changelog.
But for now, onto the "master" changelog.
- Added
NomicHFEmbedding
(#10762) - Added
MinioReader
(#10744)
- Various fixes for clickhouse vector store (#10799)
- Fix index name in neo4j vector store (#10749)
- Fixes to sagemaker embeddings (#10778)
- Fixed performance issues when splitting nodes (#10766)
- Fix non-float values in reranker + b25 (#10930)
- OpenAI-agent should be a dep of openai program (#10930)
- Add missing shortcut imports for query pipeline components (#10930)
- Fix NLTK and tiktoken not being bundled properly with core (#10930)
- Add back
llama_index.core.__version__
(#10930)
- Fixed passing in LLM to
as_chat_engine
(#10605) - Fixed system prompt formatting for anthropic (#10603)
- Fixed elasticsearch vector store error on
__version__
(#10656) - Fixed import on openai pydantic program (#10657)
- Added client back to opensearch vector store exports (#10660)
- Fixed bug in SimpleDirectoryReader not using file loaders properly (#10655)
- Added lazy LLM initialization to RankGPT (#10648)
- Fixed bedrock embedding
from_credentials
passing ing the model name (#10640) - Added back recent changes to TelegramReader (#10625)
- Several changes are introduced. See the full blog post for complete details.
- Add back deprecated API for BedrockEmbdding (#10581)
Last patch before v0.10!
- add conditional links to query pipeline (#10520)
- refactor conditional links + add to cookbook (#10544)
- agent + query pipeline cleanups (#10563)
- Add sleep to fix lag in chat stream (#10339)
- OllamaMultiModal kwargs (#10541)
- Update Ingestion Pipeline to handle empty documents (#10543)
- Fixing minor spelling error (#10516)
- fix elasticsearch async check (#10549)
- Docs/update slack demo colab (#10534)
- Adding the possibility to use the IN operator for PGVectorStore (#10547)
- fix agent reset (#10562)
- Fix MD duplicated Node id from multiple docs (#10564)
- Update pooling strategy for embedding models (#10536)
- Add Multimodal Video RAG example (#10530)
- Add SECURITY.md (#10531)
- Move agent module guide up one-level (#10519)
- Deeplake fixes (#10529)
- Add Cohere section for llamaindex (#10523)
- Fix md element (#10510)
- Upgraded deeplake vector database to use BasePydanticVectorStore (#10504)
- Fix MD parser for inconsistency tables (#10488)
- Fix ImportError for pypdf in MetadataExtractionSEC.ipynb (#10491)
- Refactor: add AgentRunner.from_llm method (#10452)
- Support custom prompt formatting for non-chat LLMS (#10466)
- Bump cryptography from 41.0.7 to 42.0.0 (#10467)
- Add persist and load method for Colbert Index (#10477)
- Allow custom agent to take in user inputs (#10450)
- remove exporter from arize-phoenix global callback handler (#10465)
- Fixing Dashscope qwen llm bug (#10471)
- Fix: calling AWS Bedrock models (#10443)
- Update Azure AI Search (fka Azure Cognitive Search) vector store integration to latest client SDK 11.4.0 stable + updating jupyter notebook sample (#10416)
- fix some imports (#10485)
- ollama vision cookbook (#10438)
- Support Gemini "transport" configuration (#10457)
- Add Upstash Vector (#10451)
- Add multi-modal ollama (#10434)
- update base class for astradb (#10435)
- Add Async support for Base nodes parser (#10418)
- Add support for
gpt-3.5-turbo-0125
(#10412) - Added
create-llama
support to rag cli (#10405)
- Fixed minor bugs in lance-db vector store (#10404)
- Fixed streaming bug in ollama (#10407)
- Nomic Embedding (#10388)
- Dashvector support sparse vector (#10386)
- Table QA with MarkDownParser and Benchmarking (#10382)
- Simple web page reader (#10395)
- fix full node content in KeywordExtractor (#10398)
- Improve and fix bugs for MarkdownElementNodeParser (#10340)
- Fixed and improve Perplexity support for new models (#10319)
- Ensure system_prompt is passed to Perplexity LLM (#10326)
- Extended BaseRetrievalEvaluator to include an optional PostProcessor (#10321)
- Support for new GPT Turbo Models (#10291)
- Support Multiple docs for Sentence Transformer Fine tuning(#10297)
- Marvin imports fixed (#9864)
- Support for new OpenAI v3 embedding models (#10279)
- Extra checks on sparse embeddings for qdrant (#10275)
- Added a RAG CLI utility (#10193)
- Added a textai vector store (#10240)
- Added a Postgresql based docstore and index store (#10233)
- specify tool spec in tool specs (#10263)
- Fixed serialization error in ollama chat (#10230)
- Added missing fields to
SentenceTransformerRerank
(#10225) - Fixed title extraction (#10209, #10226)
- nit: make chainable output parser more exposed in library/docs (#10262)
- 🐛 summary index not carrying over excluded metadata keys (#10259)
- Added support for
SageMakerEmbedding
(#10207)
- Fix duplicated
file_id
on openai assistant (#10223) - Fix circular dependencies for programs (#10222)
- Run
TitleExtractor
on groups of nodes from the same parent document (#10209) - Improve vectara auto-retrieval (#10195)
beautifulsoup4
dependency to new optional extrahtml
(#10156)- make
BaseNode.hash
an@property
(#10163) - Neutrino (#10150)
- feat: JSONalyze Query Engine (#10067)
- [wip] add custom hybrid retriever notebook (#10164)
- add from_collection method to ChromaVectorStore class (#10167)
- CLI experiment v0: ask (#10168)
- make react agent prompts more editable (#10154)
- Add agent query pipeline (#10180)
- Update supabase vecs metadata filter function to support multiple fields (#10133)
- Bugfix/code improvement for LanceDB integration (#10144)
beautifulsoup4
optional dependency (#10156)- Fix qdrant aquery hybrid search (#10159)
- make hash a @property (#10163)
- fix: bug on poetry install of llama-index[postgres] (#10171)
- [doc] update jaguar vector store documentation (#10179)
- Remove use of not-launched finish_message (#10188)
- Updates to Lantern vector stores docs (#10192)
- fix typo in multi_document_agents.ipynb (#10196)
- Added SageMakerEndpointLLM (#10140)
- Added support for Qdrant filters (#10136)
- Update bedrock utils for Claude 2:1 (#10139)
- BugFix: deadlocks using multiprocessing (#10125)
- Added RankGPT as a postprocessor (#10054)
- Ensure backwards compatibility with new Pinecone client version bifucation (#9995)
- Recursive retriever all the things (#10019)
- BugFix: When using markdown element parser on a table containing comma (#9926)
- extend auto-retrieval notebook (#10065)
- Updated the Attribute name in llm_generators (#10070)
- jaguar vector store add text_tag to add_kwargs in add() (#10057)
- added query-time row retrieval + fix nits with query pipeline over structured data (#10061)
- ReActive Agents w/ Context + updated stale link (#10058)
- Added selectors and routers to query pipeline (#9979)
- Added sparse-only search to qdrant vector store (#10041)
- Added Tonic evaluators (#10000)
- Adding async support to firestore docstore (#9983)
- Implement mongodb docstore
put_all
method (#10014)
- Properly truncate sql results based on
max_string_length
(#10015) - Fixed
node.resolve_image()
for base64 strings (#10026) - Fixed cohere system prompt role (#10020)
- Remove redundant token counting operation in SentenceSplitter (#10053)
- Implements a Node Parser using embeddings for Semantic Splitting (#9988)
- Add Anyscale Embedding model support (#9470)
- nit: fix pandas get prompt (#10001)
- Fix: Token counting bug (#9912)
- Bump jinja2 from 3.1.2 to 3.1.3 (#9997)
- Fix corner case for qdrant hybrid search (#9993)
- Bugfix: sphinx generation errors (#9944)
- Fix:
language
used before assignment inCodeSplitter
(#9987) - fix inconsistent name "text_parser" in section "Use a Text Splitter… (#9980)
- 🐛 fixing batch size (#9982)
- add auto-async execution to query pipelines (#9967)
- 🐛 fixing init (#9977)
- Parallel Loading with SimpleDirectoryReader (#9965)
- do not force delete an index in milvus (#9974)
- Added support for together.ai models (#9962)
- Added support for batch redis/firestore kvstores, async firestore kvstore (#9827)
- Parallelize
IngestionPipeline.run()
(#9920) - Added new query pipeline components: function, argpack, kwargpack (#9952)
- Updated optional langchain imports to avoid warnings (#9964)
- Raise an error if empty nodes are embedded (#9953)
- Added support for Nvidia TenorRT LLM (#9842)
- Allow
tool_choice
to be set during agent construction (#9924) - Added streaming support for
QueryPipeline
(#9919)
- Set consistent doc-ids for llama-index readers (#9923, #9916)
- Remove unneeded model inputs for HuggingFaceEmbedding (#9922)
- Propagate
tool_choice
flag to downstream APIs (#9901) - Add
chat_store_key
to chat memoryfrom_defaults()
(#9928)
- add query pipeline (#9908)
- Feature: Azure Multi Modal (fixes: #9471) (#9843)
- add postgres docker (#9906)
- Vectara auto_retriever (#9865)
- Redis Chat Store support (#9880)
- move more classes to core (#9871)
- Propagate
tool_choice
flag to downstream APIs (#9901) - filter out negative indexes from faiss query (#9907)
- added NE filter for qdrant payloads (#9897)
- Fix incorrect id assignment in MyScale query result (#9900)
- Qdrant Text Match Filter (#9895)
- Fusion top k for hybrid search (#9894)
- Fix (#9867) sync_to_async to avoid blocking during asynchronous calls (#9869)
- A single node passed into compute_scores returns as a float (#9866)
- Remove extra linting steps (#9878)
- add vectara links (#9886)
- Added a
BaseChatStore
andSimpleChatStore
abstraction for dedicated chat memory storage (#9863) - Enable custom
tree_sitter
parser to be passed intoCodeSplitter
(#9845) - Created a
BaseAutoRetriever
base class, to allow other retrievers to extend to auto modes (#9846) - Added support for Nvidia Triton LLM (#9488)
- Added
DeepEval
one-click observability (#9801)
- Updated the guidance integration to work with the latest version (#9830)
- Made text storage optional for doctores/ingestion pipeline (#9847)
- Added missing
sphinx-automodapi
dependency for docs (#9852) - Return actual node ids in weaviate query results (#9854)
- Added prompt formatting to LangChainLLM (#9844)
- Added concurrancy limits for dataset generation (#9779)
- New
deepeval
one-click observability handler (#9801) - Added jaguar vector store (#9754)
- Add beta multimodal ReAct agent (#9807)
- Changed default batch size for OpenAI embeddings to 100 (#9805)
- Use batch size properly for qdrant upserts (#9814)
_verify_source_safety
uses AST, not regexes, for proper safety checks (#9789)- use provided LLM in element node parsers (#9776)
- updated legacy vectordb loading function to be more robust (#9773)
- Use provided http client in AzureOpenAI (#9772)
- Add reranker for BEIR evaluation (#9743)
- Add Pathway integration. (#9719)
- custom agents implementation + notebook (#9746)
- fix beam search for vllm: add missing parameter (#9741)
- Fix alpha for hrbrid search (#9742)
- fix token counter (#9744)
- BM25 tokenizer lowercase (#9745)
- docs: fixes qdrant_hybrid.ipynb typos (#9729)
- make llm completion program more general (#9731)
- Refactor MM Vector store and Index for empty collection (#9717)
- Adding IF statement to check for Schema using "Select" (#9712)
- allow skipping module loading in
download_module
anddownload_llama_pack
(#9734)
- Added
.iter_data()
method toSimpleDirectoryReader
(#9658) - Added async support to
Ollama
LLM (#9689) - Expanding pinecone filter support for
in
andnot in
(#9683)
- Improve BM25Retriever performance (#9675)
- Improved qdrant hybrid search error handling (#9707)
- Fixed
None
handling inChromaVectorStore
(#9697) - Fixed postgres schema creation if not existing (#9712)
- Added zilliz cloud as a managed index (#9605)
- Bedrock client and LLM fixes (#9671, #9646)
- Added
insert_batch_size
to limit number of embeddings held in memory when creating an index, defaults to 2048 (#9630) - Improve auto-retrieval (#9647)
- Configurable Node ID Generating Function (#9574)
- Introduced action input parser (#9575)
- qdrant sparse vector support (#9644)
- Introduced upserts and delete in ingestion pipeline (#9643)
- Add Zilliz Cloud Pipeline as a Managed Index (#9605)
- Add support for Google Gemini models via VertexAI (#9624)
- support allowing additional metadata filters on autoretriever (#9662)
- Fix pip install commands in LM Format Enforcer notebooks (#9648)
- Fixing some more links and documentations (#9633)
- some bedrock nits and fixes (#9646)
- new llama datasets
LabelledEvaluatorDataset
&LabelledPairwiseEvaluatorDataset
(#9531)
- multi-doc auto-retrieval guide (#9631)
- fix(vllm): make Vllm's 'complete' method behave the same as other LLM class (#9634)
- FIx Doc links and other documentation issue (#9632)
- [example] adding user feedback (#9601)
- FEATURE: Cohere ReRank Relevancy Metric for Retrieval Eval (#9495)
- Fix Gemini Chat Mode (#9599)
- Fixed
types-protobuf
from being a primary dependency (#9595) - Adding an optional auth token to the TextEmbeddingInference class (#9606)
- fix: out of index get latest tool call (#9608)
- fix(azure_openai.py): add missing return to subclass override (#9598)
- fix mix up b/w 'formatted' and 'format' params for ollama api call (#9594)
- agent refactor: step-wise execution (#9584)
- Add OpenRouter, with Mixtral demo (#9464)
- Add hybrid search to neo4j vector store (#9530)
- Add support for auth service accounts for Google Semantic Retriever (#9545)
- Fixed missing
default=None
forLLM.system_prompt
(#9504) - Fix #9580 : Incorporate metadata properly (#9582)
- Integrations: Gradient[Embeddings,LLM] - sdk-upgrade (#9528)
- Add mixtral 8x7b model to anyscale available models (#9573)
- Gemini Model Checks (#9563)
- Update OpenAI fine-tuning with latest changes (#9564)
- fix/Reintroduce
WHERE
filter to the Sparse Query for PgVectorStore (#9529) - Update Ollama API to ollama v0.1.16 (#9558)
- ollama: strip invalid
formatted
option (#9555) - add a device in optimum push #9541 (#9554)
- Title vs content difference for Gemini Embedding (#9547)
- fix pydantic fields to float (#9542)
- Added full support for Google Gemini text+vision models (#9452)
- Added new Google Semantic Retriever (#9440)
- added
from_existing()
method + async support to OpenAI assistants (#9367)
- Fixed huggingface LLM system prompt and messages to prompt (#9463)
- Fixed ollama additional kwargs usage (#9455)
- Add MistralAI LLM (#9444)
- Add MistralAI Embeddings (#9441)
- Add
Ollama
Embedding class (#9341) - Add
FlagEmbeddingReranker
for reranking (#9285) - feat: PgVectorStore support advanced metadata filtering (#9377)
- Added
sql_only
parameter to SQL query engines to avoid executing SQL (#9422)
- Feat/PgVector Support custom hnsw.ef_search and ivfflat.probes (#9420)
- fix F1 score definition, update copyright year (#9424)
- Change more than one image input for Replicate Multi-modal models from error to warning (#9360)
- Removed GPT-Licensed
aiostream
dependency (#9403) - Fix result of BedrockEmbedding with Cohere model (#9396)
- Only capture valid tool names in react agent (#9412)
- Fixed
top_k
being multiplied by 10 in azure cosmos (#9438) - Fixed hybrid search for OpenSearch (#9430)
- Updated the base
LLM
interface to matchLLMPredictor
(#9388) - Deprecated
LLMPredictor
(#9388)
- Added batch prediction support for
LabelledRagDataset
(#9332)
- Fixed save and load for faiss vector store (#9330)
- Added an option
reuse_client
to openai/azure to help with async timeouts. Set toFalse
to see improvements (#9301) - Added support for
vLLM
llm (#9257) - Add support for python 3.12 (#9304)
- Support for
claude-2.1
model name (#9275)
- Fix embedding format for bedrock cohere embeddings (#9265)
- Use
delete_kwargs
for filtering in weaviate vector store (#9300) - Fixed automatic qdrant client construction (#9267)
- Make
reference_contexts
optional inLabelledRagDataset
(#9266) - Re-organize
download
module (#9253) - Added document management to ingestion pipeline (#9135)
- Add docs for
LabelledRagDataset
(#9228) - Add submission template notebook and other doc updates for
LabelledRagDataset
(#9273)
- Convert numpy to list for
InstructorEmbedding
(#9255)
- Advanced Metadata filter for vector stores (#9216)
- Amazon Bedrock Embeddings New models (#9222)
- Added PromptLayer callback integration (#9190)
- Reuse file ids for
OpenAIAssistant
(#9125)
- Deprecate ExactMatchFilter (#9216)
- Add new abstractions for
LlamaDataset
's (#9165) - Add metadata filtering and MMR mode support for
AstraDBVectorStore
(#9193) - Allowing newest
scikit-learn
versions (#9213)
- Added
LocalAI
demo and began deprecation cycle (#9151) - Deprecate
QueryResponseDataset
andDatasetGenerator
ofevaluation
module (#9165)
- Fix bug in
download_utils.py
with pointing to wrong repo (#9215) - Use
azure_deployment
kwarg inAzureOpenAILLM
(#9174) - Fix similarity score return for
AstraDBVectorStore
Integration (#9193)
- Add
persist
andpersist_from_dir
methods toObjectIndex
that are able to support it (#9064) - Added async metadata extraction + pipeline support (#9121)
- Added back support for start/end char idx in nodes (#9143)
- Fix for some kwargs not being set properly in global service context (#9137)
- Small fix for
memory.get()
when system/prefix messages are large (#9149) - Minor fixes for global service context (#9137)
- Add support for
PGVectoRsStore
(#9087) - Enforcing
requests>=2.31
for security, while unpinningurllib3
(#9108)
- Increased default memory token limit for context chat engine (#9123)
- Added system prompt to
CondensePlusContextChatEngine
that gets prepended to thecontext_prompt
(#9123) - Fixed bug in
CondensePlusContextChatEngine
not using chat history properly (#9129)
- Added
default_headers
argument to openai LLMs (#9090) - Added support for
download_llama_pack()
and LlamaPack integrations - Added support for
llamaindex-cli
command line tool
- store normalize as bool for huggingface embedding (#9089)
- Fixed bug with AzureOpenAI logic for inferring if stream chunk is a tool call (#9018)
FastEmbed
embeddings provider (#9043)- More precise testing of
OpenAILike
(#9026) - Added callback manager to each retriever (#8871)
- Ability to bypass
max_tokens
inference withOpenAILike
(#9032)
- Fixed bug in formatting chat prompt templates when estimating chunk sizes (#9025)
- Sandboxed Pandas execution, remediate CVE-2023-39662 (#8890)
- Restored
mypy
for Python 3.8 (#9031) - Loosened
dataclasses-json
version range, and removes unnecessaryjinja2
extra frompandas
(#9042)
- Added
CondensePlusContextChatEngine
(#8949)
- Fixed bug with
OpenAIAgent
inserting errors into chat history (#9000) - Fixed various bugs with LiteLLM and the new OpenAI client (#9003)
- Added context window attribute to perplexity llm (#9012)
- Add
node_parser
attribute back to service context (#9013) - Refactor MM retriever classes (#8998)
- Fix TextNode instantiation on SupabaseVectorIndexDemo (#8994)
- Add perplexity LLM integration (#8734)
- Fix token counting for new openai client (#8981)
- Fix small pydantic bug in postgres vector db (#8962)
- Fixed
chunk_overlap
anddoc_id
bugs inHierarchicalNodeParser
(#8983)
- Added new notebook guide for Multi-Modal Rag Evaluation (#8945)
- Added
MultiModalRelevancyEvaluator
, andMultiModalFaithfulnessEvaluator
(#8945)
- Added Cohere Reranker fine-tuning (#8859)
- Support for custom httpx client in
AzureOpenAI
LLM (#8920)
- Fixed issue with
set_global_service_context
not propagating settings (#8940) - Fixed issue with building index with Google Palm embeddings (#8936)
- Fixed small issue with parsing ImageDocuments/Nodes that have no text (#8938)
- Fixed issue with large data inserts in Astra DB (#8937)
- Optimize
QueryEngineTool
for agents (#8933)
- New
IngestionPipeline
concept for ingesting and transforming data - Data ingestion and transforms are now automatically cached
- Updated interface for node parsing/text splitting/metadata extraction modules
- Changes to the default tokenizer, as well as customizing the tokenizer
- Packaging/Installation changes with PyPi (reduced bloat, new install options)
- More predictable and consistent import paths
- Plus, in beta: MultiModal RAG Modules for handling text and images!
- Find more details at:
https://medium.com/@llama_index/719f03282945
- Increase max weaivate delete size to max of 10,000 (#8887)
- Final pickling remnant fix (#8902)
- Fixed bug in loading pickled objects (#8880)
- Fix
custom_path
vscustom_dir
indownload_loader
(#8865)
- openai assistant agent + advanced retrieval cookbook (#8863)
- add retrieval API benchmark (#8850)
- Add JinaEmbedding class (#8704)
- Improved default timeouts/retries for OpenAI (#8819)
- Add back key validation for OpenAI (#8819)
- Disable automatic LLM/Embedding model downloads, give informative error (#8819)
- fix openai assistant tool creation + retrieval notebook (#8862)
- Quick fix Replicate MultiModal example (#8861)
- fix: paths treated as hidden (#8860)
- fix Replicate multi-modal LLM + notebook (#8854)
- Feature/citation metadata (#8722)
- Fix ImageNode type from NodeWithScore for SimpleMultiModalQueryEngine (#8844)
- Advanced Multi Modal Retrieval Example and docs (#8822, #8823)
- Fix retriever node postprocessors for
CitationQueryEngine
(#8818) - Fix
cannot pickle 'builtins.CoreBPE' object
in most scenarios (#8835)
- Support parallel function calling with new OpenAI client in
OpenAIPydanticProgram
(#8793)
- Fix bug in pydantic programs with new OpenAI client (#8793)
- Fixed bug with un-listable fsspec objects (#8795)
OpenAIAgent
parallel function calling (#8738)
- Properly supporting Hugging Face recommended model (#8784)
- Fixed missing import for
embeddings.__all__
(#8779)
- Use
tool_choice
overfunction_call
andtool
overfunctions
inOpenAI(LLM)
(#8738) - Deprecate
to_openai_function
in favor ofto_openai_tool
(#8738)
OpenAIAgent
parallel function calling (#8738)- Add AI assistant agent (#8735)
- OpenAI GPT4v Abstraction (#8719)
- Add support for
Lantern
VectorStore (#8714)
- Fix returning zero nodes in elastic search vector store (#8746)
- Add try/except for
SimpleDirectoryReader
loop to avoid crashing on a single document (#8744) - Fix for
deployment_name
in async embeddings (#8748)
- added native sync and async client support for the lasted
openai
client package (#8712) - added support for
AzureOpenAIEmbedding
(#8712)
- Fixed errors about "no host supplied" with
download_loader
(#8723)
OpenAIEmbedding
no longer supports azure, moved into theAzureOpenAIEmbedding
class (#8712)
- add new devday models (#8713)
- moved
max_docs
parameter from constructor tolazy_load_data()
forSimpleMongoReader
(#8686)
- [experimental] Hyperparameter tuner (#8687)
- Fix typo error in CohereAIModelName class: cohere light models was missing v3 (#8684)
- Update deeplake.py (#8683)
- prompt optimization guide (#8659)
- VoyageEmbedding (#8634)
- Multilingual support for
YoutubeTranscriptReader
(#8673) - emotion prompt guide (#8674)
- Adds mistral 7b instruct v0.1 to available anyscale models (#8652)
- Make pgvector's setup (extension, schema, and table creation) optional (#8656)
- Allow init of stores_text variable for Pinecone vector store (#8633)
- fix: azure ad support (#8667)
- Fix nltk bug in multi-threaded environments (#8668)
- Fix google colab link in cohereai notebook (#8677)
- passing max_tokens to the
Cohere
llm (#8672)
- Deepmemory support (#8625)
- Add CohereAI embeddings (#8650)
- Add Azure AD (Microsoft Entra ID) support (#8667)
- Add
lm-format-enforcer
integration for structured output (#8601) - Google Vertex Support (#8626)
- Add
VoyageAIEmbedding
integration (#8634) - Add fine-tuning evaluator notebooks (#8596)
- Add
SingleStoreDB
integration (#7991) - Add support for ChromaDB PersistentClient (#8582)
- Add DataStax Astra DB support (#8609)
- Update dataType in Weaviate (#8608)
- In Knowledge Graph Index with hybrid retriever_mode,
- return the nodes found by keyword search when 'No Relationship found'
- Fix exceed context length error in chat engines (#8530)
- Retrieve actual content of all the triplets from KG (#8579)
- Return the nodes found by Keywords when no relationship is found by embeddings in hybrid retriever_mode in
KnowledgeGraphIndex
(#8575) - Optimize content of retriever tool and minor bug fix (#8588)
- Add Amazon
BedrockEmbedding
(#8550) - Moves
HuggingFaceEmbedding
to center onPooling
enum for pooling (#8467) - Add IBM WatsonX LLM support (#8587)
- [Bug] Patch Clarifai classes (#8529)
- fix retries for bedrock llm (#8528)
- Fix : VectorStore’s QueryResult always returns saved Node as TextNode (#8521)
- Added default file_metadata to get basic metadata that many postprocessors use, for SimpleDirectoryReader (#8486)
- Handle metadata with None values in chromadb (#8584)
- allow prompts to take in functions with
function_mappings
(#8548) - add advanced prompt + "prompt engineering for RAG" notebook (#8555)
- Leverage Replicate API for serving LLaVa modal (#8539)
- Update pull request template with google colab support inclusion (#8525)
- notebook showing how to fine-tune llama2 on structured outputs (#8540)
- added GradientAIFineTuningHandler
- added pydantic_program_mode to ServiceContext
- Initialize MultiModal Retrieval using LlamaIndex (#8507)
- Add missing import to
ChatEngine
usage pattern.md
doc (#8518) - 🐛 fixed async add (#8531)
- fix: add the needed CondenseQuestionChatEngine import in the usage_pa… (#8518)
- Add import LongLLMLinguaPostprocessor for LongLLMLingua.ipynb (#8519)
- Docs refactor (#8500)
An overhaul of the docs organization. Major changes
- Added a big new "understanding" section
- Added a big new "optimizing" section
- Overhauled Getting Started content
- Categorized and moved module guides to a single section
- Add longllmlingua (#8485)
- Add google colab support for notebooks (#7560)
- Adapt Cassandra VectorStore constructor DB connection through cassio.init (#8255)
- Allow configuration of service context and storage context in managed index (#8487)
- Add Llava MultiModal QA examples for Tesla 10k RAG (#8271)
- fix bug streaming on react chat agent not working as expected (#8459)
- patch: add selected result to response metadata for router query engines, fix bug (#8483)
- add Jina AI embeddings notebook + huggingface embedding fix (#8478)
- add
is_chat_model
to replicate (#8469) - Brought back
toml-sort
topre-commit
(#8267) - Added
LocationConstraint
for localtest_s3_kvstore
(#8263)
- Expose prompts in different modules (query engines, synthesizers, and more) (#8275)
- New LLM integrations
- Support for Hugging Face Inference API's
conversational
,text_generation
, andfeature_extraction
endpoints viahuggingface_hub[inference]
(#8098) - Add Amazon Bedrock LLMs (#8223)
- Add AI21 Labs LLMs (#8233)
- Add OpenAILike LLM class for OpenAI-compatible api servers (#7973)
- Support for Hugging Face Inference API's
- New / updated vector store integrations
- Add DashVector (#7772)
- Add Tencent VectorDB (#8173)
- Add option for custom Postgres schema on PGVectorStore instead of only allowing public schema (#8080)
- Add Gradient fine tuning engine (#8208)
- docs(FAQ): frequently asked questions (#8249)
- Fix inconsistencies with
ReActAgent.stream_chat
(#8147) - Deprecate some functions for GuardrailsOutputParser (#8016)
- Simplify dependencies (#8236)
- Bug fixes for LiteLLM (#7885)
- Update for Predibase LLM (#8211)
- Add
DELETE
for MyScale vector store (#8159) - Add SQL Retriever (#8197)
- add semantic kernel document format (#8226)
- Improve MyScale Hybrid Search and Add
DELETE
for MyScale vector store (#8159)
- Fixed additional kwargs in ReActAgent.from_tools() (#8206)
- Fixed missing spaces in prompt templates (#8190)
- Remove auto-download of llama2-13B on exception (#8225)
- add response synthesis to text-to-SQL (#8196)
- Added support for
LLMRailsEmbedding
(#8169) - Inferring MPS device with PyTorch (#8195)
- Consolidated query/text prepending (#8189)
- Add fine-tuning router support + embedding selector (#8174)
- add more document converters (#8156)
- Add normalization to huggingface embeddings (#8145)
- Improve MyScale Hybrid Search (#8159)
- Fixed duplicate
FORMAT_STR
being inside prompt (#8171) - Added: support for output_kwargs={'max_colwidth': xx} for PandasQueryEngine (#8110)
- Minor fix in the description for an argument in cohere llm (#8163)
- Fix Firestore client info (#8166)
- Added support for fine-tuning cross encoders (#7705)
- Added
QueryFusionRetriever
for merging multiple retrievers + query augmentation (#8100) - Added
nb-clean
topre-commit
to minimize PR diffs (#8108) - Support for
TextEmbeddingInference
embeddings (#8122)
- Improved the
BM25Retriever
interface to acceptBaseNode
objects (#8096) - Fixed bug with
BM25Retriever
tokenizer not working as expected (#8096) - Brought mypy to pass in Python 3.8 (#8107)
ReActAgent
adding missingsuper().__init__
call (#8125)
- add pgvector sql query engine (#8087)
- Added HoneyHive one-click observability (#7944)
- Add support for both SQLAlchemy V1 and V2 (#8060)
- Moves
codespell
topre-commit
(#8040) - Added
prettier
for autoformatting extensions besides.py
(#8072)
- Fixed forgotten f-str in
HuggingFaceLLM
(#8075) - Relaxed numpy/panadas reqs
- Added support for
GradientEmbedding
embed models (#8050)
- added
messages_to_prompt
kwarg toHuggingFaceLLM
(#8054) - improved selection and sql parsing for open-source models (#8054)
- fixed bug when agents hallucinate too many kwargs for a tool (#8054)
- improved prompts and debugging for selection+question generation (#8056)
LocalAI
more intuitive module-level var names (#8028)- Enable
codespell
for markdown docs (#7972) - add unstructured table element node parser (#8036)
- Add: Async upserting for Qdrant vector store (#7968)
- Add cohere llm (#8023)
- Parse multi-line outputs in react agent answers (#8029)
- Add properly named kwargs to keyword
as_retriever
calls (#8011) - Updating Reference to RAGAS LlamaIndex Integration (#8035)
- Vectara bugfix (#8032)
- Fix: ChromaVectorStore can attempt to add in excess of chromadb batch… (#8019)
- Fix get_content method in Mbox reader (#8012)
- Apply kwarg filters in WeaviateVectorStore (#8017)
- Avoid ZeroDivisionError (#8027)
LocalAI
intuitive module-level var names (#8028)- zep/fix: imports & typing (#8030)
- refactor: use
str.join
(#8020) - use proper metadata str for node parsing (#7987)
- You.com retriever (#8024)
- Pull fields from mongodb into metadata with
metadata_names
argument (#8001) - Simplified
LocalAI.__init__
preserving the same behaviors (#7982)
- Use longest metadata string for metadata aware text splitting (#7987)
- Handle lists of strings in mongodb reader (#8002)
- Removes
OpenAI.class_type
as it was dead code (#7983) - Fixing
HuggingFaceLLM.device_map
type hint (#7989)
- Added support for
Clarifai
LLM (#7967) - Add support for function fine-tuning (#7971)
- Update document summary index (#7815)
- change default retrieval mode to embedding
- embed summaries into vector store by default at indexing time (instead of calculating embedding on the fly)
- support configuring top k in llm retriever
- Added support for pydantic object outputs with query engines (#7893)
ClarifaiEmbedding
class added for embedding support (#7940)- Markdown node parser, flat file reader and simple file node parser (#7863)
- Added support for mongdb atlas
$vectorSearch
(#7866)
- Adds support for using message metadata in discord reader (#7906)
- Fix
LocalAI
chat capability withoutmax_tokens
(#7942) - Added
codespell
for automated checking (#7941) ruff
modernization and autofixes (#7889)- Implement own SQLDatabase class (#7929)
- Update LlamaCPP context_params property (#7945)
- fix duplicate embedding (#7949)
- Adds
codespell
tool for enforcing good spelling (#7941) - Supporting
mypy
local usage withvenv
(#7952) - Vectara - minor update (#7954)
- Avoiding
pydantic
reinstalls in CI (#7956) - move tree_sitter_languages into data_requirements.txt (#7955)
- Add
cache_okay
param toPGVectorStore
to help suppress TSVector warnings (#7950)
- Updated
KeywordNodePostprocessor
to use spacy to support more languages (#7894) LocalAI
supporting global or per-query/chat/completions
vs/completions
(#7921)- Added notebook on using REBEL + Wikipedia filtering for knowledge graphs (#7919)
- Added support for
ElasticsearchEmbedding
(#7914)
- Supporting
LocalAI
LLMs (#7913) - Validations protecting against misconfigured chunk sizes (#7917)
- Simplify NL SQL response to SQL parsing, with expanded NL SQL prompt (#7868)
- Improve vector store retrieval speed for vectordb integrations (#7876)
- Added replacing {{ and }}, and fixed JSON parsing recursion (#7888)
- Nice-ified JSON decoding error (#7891)
- Nice-ified SQL error from LLM not providing SQL (#7900)
- Nice-ified
ImportError
forHuggingFaceLLM
(#7904) - eval fixes: fix dataset response generation, add score to evaluators (#7915)
- add "build RAG from scratch notebook" - OSS/local (#7864)
- Fix elasticsearch hybrid scoring (#7852)
- Replace
get_color_mapping
andprint_text
Langchain dependency with internal implementation (#7845) - Fix async streaming with azure (#7856)
- Avoid
NotImplementedError()
in sub question generator (#7855) - Patch predibase initialization (#7859)
- Bumped min langchain version and changed prompt imports from langchain (#7862)
- Fix dropping textnodes in recursive retriever (#7840)
- share callback_manager between agent and its llm when callback_manager is None (#7844)
- fix pandas query engine (#7847)
- Added
Konko
LLM support (#7775) - Add before/after context sentence (#7821)
- EverlyAI integration with LlamaIndex through OpenAI library (#7820)
- add Arize Phoenix tracer to global handlers (#7835)
- Normalize scores returned from ElasticSearch vector store (#7792)
- Fixed
refresh_ref_docs()
bug with order of operations (#7664) - Delay postgresql connection for
PGVectorStore
until actually needed (#7793) - Fix KeyError in delete method of
SimpleVectorStore
related to metadata filters (#7829) - Fix KeyError in delete method of
SimpleVectorStore
related to metadata filters (#7831) - Addressing PyYAML import error (#7784)
- ElasticsearchStore: Update User-Agent + Add example docker compose (#7832)
StorageContext.persist
supportingPath
(#7783)- Update ollama.py (#7839)
- fix bug for self._session_pool (#7834)
- add pairwise evaluator + benchmark auto-merging retriever (#7810)
- Minor cleanup in embedding class (#7813)
- Misc updates to
OpenAIEmbedding
(#7811)
- Added native support for
HuggingFaceEmbedding
,InstructorEmbedding
, andOptimumEmbedding
(#7795) - Added metadata filtering and hybrid search to MyScale vector store (#7780)
- Allowing custom text field name for Milvus (#7790)
- Add support for
vector_store_query_mode
toVectorIndexAutoRetriever
(#7797)
- Update
LanceDBVectorStore
to handle score and distance (#7754) - Pass LLM to
memory_cls
inCondenseQuestionChatEngine
(#7785)
- add pydantic metadata extractor (#7778)
- Allow users to set the embedding dimensions in azure cognitive vector store (#7734)
- Add semantic similarity evaluator (#7770)
- 📝docs: Update Chatbot Tutorial and Notebook (#7767)
- Fixed response synthesizers with empty nodes (#7773)
- Fix
NotImplementedError
in auto vector retriever (#7764) - Multiple kwargs values in "KnowledgeGraphQueryEngine" bug-fix (#7763)
- Allow setting azure cognitive search dimensionality (#7734)
- Pass service context to index for dataset generator (#7748)
- Fix output parsers for selector templates (#7774)
- Update Chatbot_SEC.ipynb (#7711)
- linter/typechecker-friendly improvements to cassandra test (#7771)
- Expose debug option of
PgVectorStore
(#7776) - llms/openai: fix Azure OpenAI by considering
prompt_filter_results
field (#7755)
- Add support for
gpt-3.5-turbo-instruct
(#7729) - Add support for
TimescaleVectorStore
(#7727) - Added
LongContextReorder
for lost-in-the-middle issues (#7719) - Add retrieval evals (#7738)
- Added node post-processors to async context chat engine (#7731)
- Added unique index name for postgres tsv column (#7741)
- Fix langchain import error for embeddings (#7714)
- Added metadata filtering to the base simple vector store (#7564)
- add low-level router guide (#7708)
- Add CustomQueryEngine class (#7703)
- Fix context window metadata in lite-llm (#7696)
- Add CorrectnessEvaluator (#7661)
- Added support for
Ollama
LLMs (#7635) - Added
HWPReader
(#7672) - Simplified portkey LLM interface (#7669)
- Added async operation support to
ElasticsearchStore
vector store (#7613) - Added support for
LiteLLM
(#7600) - Added batch evaluation runner (#7692)
- Avoid
NotImplementedError
for async langchain embeddings (#7668) - Imrpoved reliability of LLM selectors (#7678)
- Fixed
query_wrapper_prompt
andsystem_prompt
for output parsers and completion models (#7678) - Fixed node attribute inheritance in citation query engine (#7675)
- Refactor and update
BaseEvaluator
interface to be more consistent (#7661)- Use
evaluate
function for generic input - Use
evaluate_response
function withResponse
objects from llama index query engine
- Use
- Update existing evaluators with more explicit naming
ResponseEvaluator
->FaithfulnessEvaluator
QueryResponseEvaluator
->RelevancyEvaluator
- old names are kept as class aliases for backwards compatibility
- add low-level tutorial section (#7673)
- default delta should be a dict (#7665)
- better query wrapper logic on LLMPredictor (#7667)
- add non-linear embedding adapter (#7658)
- Add "finetune + RAG" evaluation to knowledge fine-tuning notebook (#7643)
- Fixed chunk-overlap for sentence splitter (#7590)
- Added
AGENT_STEP
callback event type (#7652)
- Allowed
simple
mode to work withas_chat_engine()
(#7637) - Fixed index error in azure streaming (#7646)
- Removed
pdb
from llama-cpp (#7651)
- guide: fine-tuning to memorize knowledge (#7626)
- added ability to customize prompt template for eval modules (#7626)
- Properly detect
llama-cpp-python
version for loading the default GGML or GGUFllama2-chat-13b
model (#7616) - Pass in
summary_template
properly withRetrieverQueryEngine.from_args()
(#7621) - Fix span types in wandb callback (#7631)
- Make sure context and system prompt is included in prompt for first chat for llama2 (#7597)
- Avoid negative chunk size error in refine process (#7607)
- Fix relationships for small documents in hierarchical node parser (#7611)
- Update Anyscale Endpoints integration with full streaming and async support (#7602)
- Better support of passing credentials as LLM constructor args in
OpenAI
,AzureOpenAI
, andAnyscale
(#7602)
- Update milvus vector store to support filters and dynamic schemas (#7286)
- See the updated notebook for usage
- Added NLTK to core dependencies to support the default sentence splitter (#7606)
- Added support for ElasticSearch Vector Store (#7543)
- Fixed small
_index
bug inElasticSearchReader
(#7570) - Fixed bug with prompt helper settings in global service contexts (#7576)
- Remove newlines from openai embeddings again (#7588)
- Fixed small bug with setting
query_wrapper_prompt
in the service context (#7585)
- Clean up vector store interface to use
BaseNode
instead ofNodeWithEmbedding
- For majority of users, this is a no-op change
- For users directly operating with the
VectorStore
abstraction and manually constructingNodeWithEmbedding
objects, this is a minor breaking change. UseTextNode
withembedding
set directly, instead ofNodeWithEmbedding
.
- add embedding adapter fine-tuning engine + guide (#7565)
- Added support for Azure Cognitive Search vector store (#7469)
- Support delete in supabase (#6951)
- Added support for Espilla vector store (#7539)
- Added support for AnyScale LLM (#7497)
- Default to user-configurable top-k in
VectorIndexAutoRetriever
(#7556) - Catch validation errors for structured responses (#7523)
- Fix streaming refine template (#7561)
- Added Portkey LLM integration (#7508)
- Support postgres/pgvector hybrid search (#7501)
- upgrade recursive retriever node reference notebook (#7537)
- replace list index with summary index (#7478)
- rename list index to summary index part 2 (#7531)
- add agent finetuning guide (#7526)
- Make (some) loaders serializable (#7498)
- add node references to recursive retrieval (#7522)
- Raise informative error when metadata is too large during splitting (#7513)
- Allow langchain splitter in simple node parser (#7517)
- fix link to Marvin notebook in docs (#7504)
- Ensure metadata is not
None
inSimpleWebPageReader
(#7499) - Fixed KGIndex visualization (#7493)
- Improved empty response in KG Index (#7493)
- Added support for
MarvinEntityExtractor
metadata extractor (#7438) - Added a url_metadata callback to SimpleWebPageReader (#7445)
- Expanded callback logging events (#7472)
- Only convert newlines to spaces for text 001 embedding models in OpenAI (#7484)
- Fix
KnowledgeGraphRagRetriever
for non-nebula indexes (#7488) - Support defined embedding dimension in
PGVectorStore
(#7491) - Greatly improved similarity calculation speed for the base vector store (#7494)
- feat: non-kg heterogeneous graph support in Graph RAG (#7459)
- rag guide (#7480)
- Improve openai fine-tuned model parsing (#7474)
- doing some code de-duplication (#7468)
- support both str and templates for query_wrapper_prompt in HF LLMs (#7473)
- Add embedding finetuning (#7452)
- Added support for RunGPT LLM (#7401)
- Integration guide and notebook with DeepEval (#7425)
- Added
VectorIndex
andVectaraRetriever
as a managed index (#7440) - Added support for
to_tool_list
to detect and use async functions (#7282)
- add openai finetuning class (#7442)
- Service Context to/from dict (#7395)
- add finetuning guide (#7429)
- Add example how to run FalkorDB docker (#7441)
- Update root.md to use get_response_synthesizer expected type. (#7437)
- Bugfix MonsterAPI Pydantic version v2/v1 support. Doc Update (#7432)
- AutoMergingRetriever (#7420)
- Added support for
MonsterLLM
using MonsterAPI (#7343) - Support comments fields in NebulaGraphStore and int type VID (#7402)
- Added configurable endpoint for DynamoDB (#6777)
- Add structured answer filtering for Refine response synthesizer (#7317)
- Use
utf-8
for json file reader (#7390) - Fix entity extractor initialization (#7407)
- Added support for FalkorDB/RedisGraph graph store (#7346)
- Added directed sub-graph RAG (#7378)
- Added support for
BM25Retriever
(#7342)
- Added
max_tokens
toXinference
LLM (#7372) - Support cache dir creation in multithreaded apps (#7365)
- Ensure temperature is a float for openai (#7382)
- Remove duplicate subjects in knowledge graph retriever (#7378)
- Added support for both pydantic v1 and v2 to allow other apps to move forward (#7394)
- Refactor prompt template (#7319)
- Use
BasePromptTemplate
for generic typing - Use
PromptTemplate
,ChatPromptTemplate
,SelectorPromptTemplate
as core implementations - Use
LangchainPromptTemplate
for compatibility with Langchain prompt templates - Fully replace specific prompt classes (e.g.
SummaryPrompt
) with genericBasePromptTemplate
for typing in codebase. - Keep
Prompt
as an alias forPromptTemplate
for backwards compatibility. - BREAKING CHANGE: remove support for
Prompt.from_langchain_prompt
, please usetemplate=LangchainPromptTemplate(lc_template)
instead.
- Use
OpenAIFineTuningHandler
for collecting LLM inputs/outputs for OpenAI fine tuning (#7367)
- Add support for
claude-instant-1.2
(#7369)
- Support fine-tuned OpenAI models (#7364)
- Added support for Cassandra vector store (#6784)
- Support pydantic fields in tool functions (#7348)
- Fix infinite looping with forced function call in
OpenAIAgent
(#7363)
- auto vs. recursive retriever notebook (#7353)
- Reader and Vector Store for BagelDB with example notebooks (#7311)
- Use service context for intermediate index in retry source query engine (#7341)
- temp fix for prompt helper + chat models (#7350)
- Properly skip unit-tests when packages not installed (#7351)
- Added FireStore docstore/index store support (#7305)
- add recursive agent notebook (#7330)
- Fix Azure pydantic error (#7329)
- fix callback trace ids (make them a context var) (#7331)
- Awadb Vector Store (#7291)
- Fix bug in OpenAI llm temperature type
- Expose a system prompt/query wrapper prompt in the service context for open-source LLMs (#6647)
- Changed default MyScale index format to
MSTG
(#7288) - Added tracing to chat engines/agents (#7304)
- move LLM and embeddings to pydantic (#7289)
- Fix sentence splitter bug (#7303)
- Fix sentence splitter infinite loop (#7295)
- Improve SQL Query parsing (#7283)
- Fix loading embed_model from global service context (#7284)
- Limit langchain version until we migrate to pydantic v2 (#7297)
- Added Knowledge Graph RAG Retriever (#7204)
- accept
api_key
kwarg in OpenAI LLM class constructor (#7263) - Fix to create separate queue instances for separate instances of
StreamingAgentChatResponse
(#7264)
- Added support for Rockset as a vector store (#7111)
- Fixed bug in service context definition that could disable LLM (#7261)
- Enable the LLM or embedding model to be disabled by setting to
None
in the service context (#7255) - Resolve nearly any huggingface embedding model using the
embed_model="local:<model_name>"
syntax (#7255) - Async tool-calling support (#7239)
- Updated supabase kwargs for add and query (#7103)
- Small tweak to default prompts to allow for more general purpose queries (#7254)
- Make callback manager optional for
CustomLLM
+ docs update (#7257)
- feat: add node_postprocessors to ContextChatEngine (#7232)
- add ensemble query engine tutorial (#7247)
- Allow EMPTY keys for Fastchat/local OpenAI API endpoints (#7224)
- Added "LLAMA_INDEX_CACHE_DIR" to control cached files (#7233)
- Default to pydantic selectors when possible (#7154, #7223)
- Remove the need for langchain wrappers on
embed_model
in the service context (#7157) - Metadata extractors take an
LLM
object now, in addition toLLMPredictor
(#7202) - Added local mode + fallback to llama.cpp + llama2 (#7200)
- Added local fallback for embeddings to
BAAI/bge-small-en
(#7200) - Added
SentenceWindowNodeParser
+MetadataReplacementPostProcessor
(#7211)
- Change default LLM to gpt-3.5-turbo from text-davinci-003 (#7223)
- Change prompts for compact/refine/tree_summarize to work better with gpt-3.5-turbo (#7150, #7179, #7223)
- Increase default LLM temperature to 0.1 (#7180)
- Reverted #7223 changes to defaults (#7235)
- Default to pydantic selectors when possible (#7154, #7223)
- Remove the need for langchain wrappers on
embed_model
in the service context (#7157) - Metadata extractors take an
LLM
object now, in addition toLLMPredictor
(#7202) - Added local mode + fallback to llama.cpp + llama2 (#7200)
- Added local fallback for embeddings to
BAAI/bge-small-en
(#7200) - Added
SentenceWindowNodeParser
+MetadataReplacementPostProcessor
(#7211)
- Change default LLM to gpt-3.5-turbo from text-davinci-003 (#7223)
- Change prompts for compact/refine/tree_summarize to work better with gpt-3.5-turbo (#7150, #7179, #7223)
- Increase default LLM temperature to 0.1 (#7180)
- docs: Improvements to Mendable Search (#7220)
- Refactor openai agent (#7077)
- Use
1 - cosine_distance
for pgvector/postgres vector db (#7217) - fix metadata formatting and extraction (#7216)
- fix(readers): Fix non-ASCII JSON Reader bug (#7086)
- Chore: change PgVectorStore variable name from
sim
todistance
for clarity (#7226)
- Fixed metadata formatting with custom tempalates and inheritance (#7216)
- Add "one click observability" page to docs (#7183)
- Added Xorbits inference for local deployments (#7151)
- Added Zep vector store integration (#7203)
- feat/zep vectorstore (#7203)
- Update the default
EntityExtractor
model (#7209) - Make
ChatMemoryBuffer
pickleable (#7205) - Refactored
BaseOpenAIAgent
(#7077)
- add ensemble retriever notebook (#7190)
- DOCS: added local llama2 notebook (#7146)
- Fix for
AttributeError: 'OpenAIAgent' object has no attribute 'callback_manager'
by calling super constructor withinBaseOpenAIAgent
- Remove backticks from nebula queries (#7192)
- Added an
EntityExtractor
for metadata extraction (#7163)
- add router module docs (#7171)
- add retriever router (#7166)
- Added a
RouterRetriever
for routing queries to specific retrievers (#7166)
- Fix for issue where having multiple concurrent streamed responses from
OpenAIAgent
would result in interleaving of tokens across each response stream. (#7164) - fix llms callbacks issue (args[0] error) (#7165)
- Added metadata filtering to weaviate (#7130)
- Added token counting (and all callbacks) to agents and streaming (#7122)
- Added
to/from_string
andto/from_dict
methods to memory objects (#7128) - Include columns comments from db tables in table info for SQL queries (#7124)
- Add Neo4j support (#7122)
- Added
Azure AD
validation support to theAzureOpenAI
class (#7127) - add
flush=True
when printing agent/chat engine response stream (#7129) - Added
Azure AD
support to theAzureOpenAI
class (#7127) - Update LLM question generator prompt to mention JSON markdown (#7105)
- Fixed
astream_chat
in chat engines (#7139)
- Update
ReActAgent
to support memory modules (minor breaking change since the constructor takesmemory
instead ofchat_history
, but the mainfrom_tools
method remains backward compatible.) (#7116) - Update
ReActAgent
to support streaming (#7119) - Added Neo4j graph store and query engine integrations (#7122)
- add object streaming (#7117)
- Chat source nodes (#7078)
- anthropic api key customization (#7082)
- Fix broken link to API reference in Contributor Docs (#7080)
- Update vector store docs (#7076)
- Update comment (#7073)
- Added HotpotQADistractor benchmark evaluator (#7034)
- Add metadata filter and delete support for LanceDB (#7048)
- Use MetadataFilters in opensearch (#7005)
- Added support for
KuzuGraphStore
(#6970) - Added
kg_triplet_extract_fn
to customize how KGs are built (#7068)
- Fix string formatting in context chat engine (#7050)
- Fixed tracing for async events (#7052)
- Less strict triplet extraction for KGs (#7059)
- Add configurable limit to KG data retrieved (#7059)
- Nebula connection improvements (#7059)
- Bug fix in building source nodes for agent response (#7067)
- Support function calling api for AzureOpenAI (#7041)
- tune prompt to get rid of KeyError in SubQ engine (#7039)
- Fix validation of Azure OpenAI keys (#7042)
- Added
kwargs
toComposableGraph
for the underlying query engines (#6990) - Validate openai key on init (#6940)
- Added async embeddings and async RetrieverQueryEngine (#6587)
- Added async
aquery
andasync_add
to PGVectorStore (#7031) - Added
.source_nodes
attribute to chat engine and agent responses (#7029) - Added
OpenInferenceCallback
for storing generation data in OpenInference format (#6998)
- Fix achat memory initialization for data agents (#7000)
- Add
print_response_stream()
to agengt/chat engine response class (#7018)
- Fix achat memory initialization for data agents (#7000)
- Add
print_response_stream()
to agengt/chat engine response class (#7018)
- Default to pydantic question generation when possible for sub-question query engine (#6979)
- Fix returned order of messages in large chat memory (#6979)
- Added a
SentenceTransformerRerank
node post-processor for fast local re-ranking (#6934) - Add numpy support for evaluating queries in pandas query engine (#6935)
- Add metadata filtering support for Postgres Vector Storage integration (#6968)
- Proper llama2 support for agents and query engines (#6969)
- Added
model_name
to LLMMetadata (#6911) - Fallback to retriever service context in query engines (#6911)
- Fixed
as_chat_engine()
ValueError with extra kwargs (#6971
- Add support for Replicate LLM (vicuna & llama 2!)
- fix streaming for condense chat engine (#6958)
- Add support for chroma v0.4.0 (#6937)
- Log embedding vectors to callback manager (#6962)
- add more robust embedding timeouts (#6779)
- improved connection session management on postgres vector store (#6843)
- specify
embed_model="local"
to use default local embbeddings in the service context (#6806) - Add async
acall
endpoint toBasePydanticProgram
(defaults to sync version). Implement forOpenAIPydanticProgram
- fix null metadata for searching existing vector dbs (#6912)
- add module guide docs for
SimpleDirectoryReader
(#6916) - make sure
CondenseQuestionChatEngine
streaming chat endpoints work even if not explicitly settingstreaming=True
in the underlying query engine.
- Added embedding speed benchmark (#6876)
- Added BEIR retrieval benchmark (#6825)
- remove toctrees from deprecated_terms (#6895)
- Relax typing dependencies (#6879)
- docs: modification to evaluation notebook (#6840)
- raise error if the model does not support functions (#6896)
- fix(bench embeddings): bug not taking into account string length (#6899)x
- Improved milvus consistency support and output fields support (#6452)
- Added support for knowledge graph querying w/ cypyer+nebula (#6642)
- Added
Document.example()
to create documents for fast prototyping (#6739) - Replace react chat engine to use native reactive agent (#6870)
- chore: added a help message to makefile (#6861)
- Fixed support for using SQLTableSchema context_str attribute (#6891)
- Added sources to agent/chat engine responses (#6854)
- Added basic chat buffer memory to agents / chat engines (#6857)
- Adding load and search tool (#6871)
- Add simple agent benchmark (#6869)
- add agent docs (#6866)
- add react agent (#6865)
- Replace react chat engine with native react agent (#6870)
- Set default chat mode to "best": use openai agent when possible, otherwise use react agent (#6870)
- Fixed support for legacy vector store metadata (#6867)
- fix chroma notebook in docs (#6872)
- update LC embeddings docs (#6868)
- Add
Anthropic
LLM implementation (#6855)
- Fix indexing error in
SentenceEmbeddingOptimizer
(#6850) - fix doc for custom embedding model (#6851)
- fix(silent error): Add validation to
SimpleDirectoryReader
(#6819) - Fix link in docs (#6833)
- Fixes Azure gpt-35-turbo model not recognized (#6828)
- Update Chatbot_SEC.ipynb (#6808)
- Rename leftover original name to LlamaIndex (#6792)
- patch nested traces of the same type (#6791)
MetadataExtractor
- Documnent Metadata Augmentation via LLM-based feature extractors (#6764)
- fixed passing in query bundle to node postprocessors (#6780)
- fixed error in callback manager with nested traces (#6791)
- Sub question query engine returns source nodes of sub questions in the callback manager (#6745)
- trulens integration (#6741)
- Add sources to subquestion engine (#6745)
- Added/Fixed streaming support to simple and condense chat engines (#6717)
- fixed
response_mode="no_text"
response synthesizer (#6755) - fixed error setting
num_output
andcontext_window
in service context (#6766) - Fix missing as_query_engine() in tutorial (#6747)
- Fixed variable sql_query_engine in the notebook (#6778)
- fix required function fields (#6761)
- Remove usage of stop token in Prompt, SQL gen (#6782)
- Support Azure OpenAI (#6718)
- Support prefix messages (e.g. system prompt) in chat engine and OpenAI agent (#6723)
- Added
CBEventType.SUB_QUESTIONS
event type for tracking sub question queries/responses (#6716)
- Fix HF LLM output error (#6737)
- Add system message support for langchain message templates (#6743)
- Fixed applying node-postprocessors (#6749)
- Add missing
CustomLLM
import underllama_index.llms
(#6752) - fix(typo):
get_transformer_tokenizer_fn
(#6729) - feat(formatting):
black[jupyter]
(#6732) - fix(test):
test_optimizer_chinese
(#6730)
- Streaming support for OpenAI agents (#6694)
- add recursive retriever + notebook example (#6682)
- Index creation progress bars (#6583)
- Improved chat refine template (#6645)
- Change
BaseOpenAIAgent
to usellama_index.llms.OpenAI
. Adjustchat_history
to useList[ChatMessage]]
as type. - Remove (previously deprecated)
llama_index.langchain_helpers.chain_wrapper
module. - Remove (previously deprecated)
llama_index.token_counter.token_counter
module. See migration guide for more details on new callback based token counting. - Remove
ChatGPTLLMPredictor
andHuggingFaceLLMPredictor
. See migration guide for more details on replacements. - Remove support for setting
cache
viaLLMPredictor
constructor. - Update
BaseChatEngine
interface:- adjust
chat_history
to useList[ChatMessage]]
as type - expose
chat_history
state as a property - support overriding
chat_history
inchat
andachat
endpoints
- adjust
- Remove deprecated arguments for
PromptHelper
:max_input_size
,embedding_limit
,max_chunk_overlap
- Update all notebooks to use native openai integration (#6696)
- add optional tqdm progress during index creation (#6583)
- Added async support for "compact" and "refine" response modes (#6590)
- [feature]add transformer tokenize functionalities for optimizer (chinese) (#6659)
- Add simple benchmark for vector store (#6670)
- Introduce
llama_index.llms
module, with newLLM
interface, andOpenAI
,HuggingFaceLLM
,LangChainLLM
implementations. (#6615) - Evaporate pydantic program (#6666)
- Improve metadata/node storage and retrieval for RedisVectorStore (#6678)
- Fixed node vs. document filtering in vector stores (#6677)
- add context retrieval agent notebook link to docs (#6660)
- Allow null values for the 'image' property in the ImageNode class and se… (#6661)
- Fix broken links in docs (#6669)
- update milvus to store node content (#6667)
- add context augmented openai agent (#6655)
- Redis support for index stores and docstores (#6575)
- DuckDB + SQL query engine notebook (#6628)
- add notebook showcasing deplot data loader (#6638)
- More robust JSON parsing from LLM for
SelectionOutputParser
(#6610) - bring our loaders back in line with llama-hub (#6630)
- Remove usage of SQLStructStoreIndex in notebooks (#6585)
- MD reader: remove html tags and leave linebreaks alone (#6618)
- bump min langchain version to latest version (#6632)
- Fix metadata column name in postgres vector store (#6622)
- Postgres metadata fixes (#6626, #6634)
- fixed links to dataloaders in contribution.md (#6636)
- fix: typo in docs in creating custom_llm huggingface example (#6639)
- Updated SelectionOutputParser to handle JSON objects and arrays (#6610)
- Fixed docstring argument typo (#6652)
- refactor structured output + pydantic programs (#6604)
- Fix serialization for OpenSearch vector stores (#6612)
- patch docs relationships (#6606)
- Bug fix for ignoring directories while parsing git repo (#4196)
- updated Chroma notebook (#6572)
- Backport old node name (#6614)
- Add the ability to change chroma implementation (#6601)
- Patch imports for Document obj for backwards compatibility (#6597)
- New
TextNode
/Document
object classes based on pydantic (#6586) TextNode
/Document
objects support metadata customization (metadata templates, exclude metadata from LLM or embeddings) (#6586)- Nodes no longer require flat metadata dictionaries, unless the vector store you use requires it (#6586)
- use
NLTK_DATA
env var to control NLTK download location (#6579) - [discord] save author as metadata in group_conversations.py (#6592)
- bs4 -> beautifulsoup4 in requirements (#6582)
- negate euclidean distance (#6564)
- add df output parser notebook link to docs (#6581)
Node
has been renamed toTextNode
and is imported fromllama_index.schema
(#6586)TextNode
andDocument
must be instantiated with kwargs:Document(text=text)
(#6586)TextNode
(fkaNode
) has aid_
ornode_id
property, rather thandoc_id
(#6586)TextNode
andDocument
have a metadata property, which replaces the extra_info property (#6586)TextNode
no longer has anode_info
property (start/end indexes are accessed directly withstart/end_char_idx
attributes) (#6586)
- Add typesense vector store (#6561)
- add df output parser (#6576)
- Track langchain dependency via bridge module. (#6573)
- add object index (#6548)
- add SQL Schema Node Mapping + SQLTableRetrieverQueryEngine + obj index fixes (#6569)
- sql refactor (NLSQLTableQueryEngine) (#6529)
- Update vector_stores.md (#6562)
- Minor
BaseResponseBuilder
interface cleanup (#6557) - Refactor TreeSummarize (#6550)
- properly convert weaviate distance to score (#6545)
- refactor tree summarize and fix bug to not truncate context (#6550)
- fix custom KG retrieval notebook nits (#6551)
- multi-selector support in router query engine (#6518)
- pydantic selector support in router query engine using OpenAI function calling API (#6518)
- streaming response support in
CondenseQuestionChatEngine
andSimpleChatEngine
(#6524) - metadata filtering support in
QdrantVectorStore
(#6476) - add
PGVectorStore
to support postgres with pgvector (#6190)
- better error handling in the mbox reader (#6248)
- Fix blank similarity score when using weaviate (#6512)
- fix for sorted nodes in
PrevNextNodePostprocessor
(#6048)
- Refactor PandasQueryEngine to take in df directly, deprecate PandasIndex (#6527)
- query planning tool with OpenAI Function API (#6520)
- docs: example of kg+vector index (#6497)
- Set context window sizes for Cohere and AI21(J2 model) (#6485)
- add default input size for Cohere and AI21 (#6485)
- docs: replace comma with colon in dict object (#6439)
- extra space in prompt and error message update (#6443)
- [Issue 6417] Fix prompt_templates docs page (#6499)
- Rip out monkey patch and update model to context window mapping (#6490)
- New OpenAI Agent + Query Engine Cookbook (#6496)
- allow recursive data extraction (pydantic program) (#6503)
- update mongo interface (#6501)
- fixes that we forgot to include for openai pydantic program (#6503) (#6504)
- Fix github pics in Airbyte notebook (#6493)
- Add node doc_id filtering to weaviate (#6467)
- New
TokenCountingCallback
to customize and track embedding, prompt, and completion token usage (#6440) - OpenAI Retrieval Function Agent (#6491)
- Deprecated current token tracking (llm predictor and embed model will no longer track tokens in the future, please use the
TokenCountingCallback
(#6440) - Add maximal marginal relevance to the Simple Vector Store, which can be enabled as a query mode (#6446)
as_chat_engine
properly inherits the current service context (#6470)- Use namespace when deleting from pinecone (#6475)
- Fix paths when using fsspec on windows (#3778)
- Fix for using custom file readers in
SimpleDirectoryReader
(#6477) - Edit MMR Notebook (#6486)
- FLARE fixes (#6484)
- Add OpenAIAgent and tutorial notebook for "build your own agent" (#6461)
- Add OpenAIPydanticProgram (#6462)
- Fix citation engine import (#6456)
- Added FLARE query engine (#6419).
- Added better support for vector store with existing data (e.g. allow configurable text key) for Pinecone and Weaviate. (#6393)
- Support batched upsert for Pineone (#6393)
- Added initial guidance integration. Added
GuidancePydanticProgram
for generic structured output generation andGuidanceQuestionGenerator
for generating sub-questions inSubQuestionQueryEngine
(#6246).
- Remove hardcoded chunk size for citation query engine (#6408)
- Mongo demo improvements (#6406)
- Fix notebook (#6418)
- Cleanup RetryQuery notebook (#6381)
- Added
SQLJoinQueryEngine
(generalization ofSQLAutoVectorQueryEngine
) (#6265) - Added support for graph stores under the hood, and initial support for Nebula KG. More docs coming soon! (#2581)
- Added guideline evaluator to allow llm to provide feedback based on user guidelines (#4664)
- Added support for MongoDB Vector stores to enable Atlas knnbeta search (#6379)
- Added new CitationQueryEngine for inline citations of sources in response text (#6239)
- Fixed bug with
delete_ref_doc
not removing all metadata from the docstore (#6192) - FIxed bug with loading existing QDrantVectorStore (#6230)
- Added changelog officially to github repo (#6191)
- SimpleDirectoryReader has new
filename_as_id
flag to automatically set the doc_id (useful forrefresh_ref_docs()
) - DocArray vector store integration
- Tair vector store integration
- Weights and Biases callback handler for tracing and versioning indexes
- Can initialize indexes directly from a vector store:
index = VectorStoreIndex.from_vector_store(vector_store=vector_store)
- Fixed multimodal notebook
- Updated/fixed the SQL tutorial in the docs
- Minor docs updates
- Added github pull-requset templates
- Added github issue-forms
- Added new JSONQueryEngine that uses JSON schema to deliver more accurate JSON query answers
- Metadata support for redis vector-store
- Added Supabase vector store integration
- Fixed typo in text-to-sql prompt
- Removed GPT prefix from indexes (old imports/names are still supported though)
- Major docs updates, brought important modules to the top level
- Added agent tool abstraction for llama-hub data loaders
- Minor doc updates
- Added
Discover LlamaIndex
video series to the tutorials docs section - Minor docs updates