langchain[patch]: fix ChatVertexAI streaming #14369

efriis · 2023-12-07T00:25:45Z

No description provided.

vercel · 2023-12-07T00:25:46Z

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
langchain	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Dec 7, 2023 5:44pm

* feat: Increased compatibility with new and old versions for dalle (#14222) - **Description:** Increased compatibility with all versions openai for dalle, This pr add support for openai version from 0 ~ 1.3. * Adds "NIN" metadata filter for pgvector to all checking for set absence (#14205) This PR adds support for metadata filters of the form: `{"filter": {"key": { "NIN" : ["list", "of", "values"]}}}` "IN" is already supported, so this is a quick & related update to add "NIN" * Fixed a typo in smart_llm prompt (#13052)  * Feature: GitLab url from ENV (#14221)  --------- Co-authored-by: Erick Friis <erick@langchain.dev> * info sql tool remove whitespaces in table names (#13712) Remove whitespaces from the input of the ListSQLDatabaseTool for better support. for example, the input "table1,table2,table3" will throw an exception whiteout the change although it's a valid input. --------- Co-authored-by: Harrison Chase <hw.chase.17@gmail.com> * Updated integration with Clarifai python SDK functions (#13671) Description : Updated the functions with new Clarifai python SDK. Enabled initialisation of Clarifai class with model URL. Updated docs with new functions examples. * OpenAIEmbeddings: retry_min_seconds/retry_max_seconds parameters (#13138) - **Description:** new parameters in OpenAIEmbeddings() constructor (retry_min_seconds and retry_max_seconds) that allow parametrization by the user of the former min_seconds and max_seconds that were hidden in _create_retry_decorator() and _async_retry_decorator() - **Issue:** #9298, #12986 - **Dependencies:** none - **Tag maintainer:** @hwchase17 - **Twitter handle:** @adumont make format ✅ make lint ✅ make test ✅ Co-authored-by: Harrison Chase <hw.chase.17@gmail.com> * Amadeus toolkit minor update (#13002) - update `Amadeus` toolkit with ability to switch Amadeus environments - update minor code explanations --------- Co-authored-by: MinjiK <minji.kim@amadeus.com> * Demonstrate use of get_buffer_string (#13013) **Description** The docs for creating a RAG chain with Memory [currently use a manual lambda](https://python.langchain.com/docs/expression_language/cookbook/retrieval#with-memory-and-returning-source-documents) to format chat history messages. [There exists a helper method within the codebase](https://github.com/langchain-ai/langchain/blob/master/libs/langchain/langchain/schema/messages.py#L14C15-L14C15) to perform this task so I've updated the documentation to demonstrate its usage Also worth noting that the current documented method of using the included `_format_chat_history ` function actually results in an error: ``` TypeError: 'HumanMessage' object is not subscriptable ``` --------- Co-authored-by: Harrison Chase <hw.chase.17@gmail.com> * Fix typo in lcel example for rerank in doc (#14336) fix typo in lcel example for rerank in doc * Exclude `max_tokens` from request if it's None (#14334)  We found a request with `max_tokens=None` results in the following error in Anthropic: ``` HTTPError: 400 Client Error: Bad Request for url: https://oregon.staging.cloud.databricks.com/serving-endpoints/corey-anthropic/invocations. Response text: {"error_code":"INVALID_PARAMETER_VALUE","message":"INVALID_PARAMETER_VALUE: max_tokens was not of type Integer: null"} ``` This PR excludes `max_tokens` if it's None. * feat(add): LLM integration of Cloudflare Workers AI (#14322) Add [Text Generation by Cloudflare Workers AI](https://developers.cloudflare.com/workers-ai/models/text-generation/). It's a new LLM integration. - Dependencies: N/A * Mask API key for baidu qianfan (#14281) Description: This PR masked baidu qianfan - Chat_Models API Key and added unit tests. Issue: the issue langchain-ai#12165. Tag maintainer: @eyurtsev --------- Co-authored-by: xiayi <xiayi@bytedance.com> * feat: mask api key for cerebriumai llm (#14272) - **Description:** Masking API key for CerebriumAI LLM to protect user secrets. - **Issue:** #12165 - **Dependencies:** None - **Tag maintainer:** @eyurtsev --------- Signed-off-by: Yuchen Liang <yuchenl3@andrew.cmu.edu> Co-authored-by: Harrison Chase <hw.chase.17@gmail.com> * Qdrant metadata payload keys (#13001) - **Description:** In Qdrant allows to input list of keys as the content_payload_key to retrieve multiple fields (the generated document will contain the dictionary {field: value} in a string), - **Issue:** Previously we were able to retrieve only one field from the vector database when making a search - **Dependencies:** - **Tag maintainer:** - **Twitter handle:** @jb_dlb --------- Co-authored-by: Jean Baptiste De La Broise <jeanbaptiste.delabroise@mdpi.com> * docs[patch]: fix ipynb links (#14325) Keeping it simple for now. Still iterating on our docs build in pursuit of making everything mdxv2 compatible for docusaurus 3, and the fewer custom scripts we're reliant on through that, the less likely the docs will break again. Other things to consider in future: Quarto rewriting in ipynbs: https://quarto.org/docs/extensions/nbfilter.html (but this won't do md/mdx files) Docusaurus plugins for rewriting these paths * Update doc-string in RunnableWithMessageHistory (#14262) Update doc-string in RunnableWithMessageHistory * docs[patch]: Fix broken link 'tip' in docs (#14349) * core[patch], langchain[patch]: ByteStore (#14312) * Fix multi vector retriever subclassing (#14350) Fixes #14342 @eyurtsev @baskaryan --------- Co-authored-by: Erick Friis <erick@langchain.dev> * infra: ci matrix (#14306) * langchain[patch]: import nits (#14354) import from core instead of langchain.schema * Include run_id (#14331) in the test run outputs * core[patch]: message history error typo (#14361) * [core/minor] Runnables: Implement a context api (#14046)  --------- Co-authored-by: Brace Sproul <braceasproul@gmail.com> * core[patch]: Release 0.0.11 (#14367) * langchain[patch]: Release 0.0.347 (#14368) * langchain[patch]: fix ChatVertexAI streaming (#14369) * langchain[patch]: Rollback multiple keys in Qdrant (#14390) This reverts commit 38813d7090294c0c96d4963a2a230db4fef5e37e. This is a temporary fix, as I don't see a clear way on how to use multiple keys with `Qdrant.from_texts`. Context: #14378 * API Reference building script update (#13587) The namespaces like `langchain.agents.format_scratchpad` clogging the API Reference sidebar. This change removes those 3-level namespaces from sidebar (this issue was discussed with @efriis ) --------- Co-authored-by: Erick Friis <erick@langchain.dev> * core[patch], langchain[patch]: fix required deps (#14373) * core[patch]: Release 0.0.12 (#14415) * langchain[patch]: Release 0.0.348 (#14417) * experimental[patch]: Release 0.0.45 (#14418) * docs: notebook linting (#14366) Many jupyter notebooks didn't pass linting. List of these files are presented in the [tool.ruff.lint.per-file-ignores] section of the pyproject.toml . Addressed these bugs: - fixed bugs; added missed imports; updated pyproject.toml Only the `document_loaders/tensorflow_datasets.ipyn`, `cookbook/gymnasium_agent_simulation.ipynb` are not completely fixed. I'm not sure about imports. --------- Co-authored-by: Erick Friis <erick@langchain.dev> * docs[patch]: `promptlayer` pages update (#14416) Updated provider page by adding LLM and ChatLLM references; removed a content that is duplicate text from the LLM referenced page. Updated the collback page * fix imports from core (#14430) * langchain[patch]: Fix scheduled testing (#14428) - integration tests in pyproject - integration test fixes * langchain[patch]: fix scheduled testing ci variables (#14459) * revoke serialization (#14456) * langchain[patch]: fix scheduled testing ci dep install (#14460) * langchain[patch]: xfail unstable vertex test (#14462) * Use deepcopy in RunLogPatch (#14244) This PR adds deepcopy usage in RunLogPatch. I included a unit-test that shows an issue that was caused in LangServe in the RemoteClient. ```python import jsonpatch s1 = {} s2 = {'value': []} s3 = {'value': ['a']} ops0 = list(jsonpatch.JsonPatch.from_diff(None, s1)) ops1 = list(jsonpatch.JsonPatch.from_diff(s1, s2)) ops2 = list(jsonpatch.JsonPatch.from_diff(s2, s3)) ops = ops0 + ops1 + ops2 jsonpatch.apply_patch(None, ops) {'value': ['a']} jsonpatch.apply_patch(None, ops) {'value': ['a', 'a']} jsonpatch.apply_patch(None, ops) {'value': ['a', 'a', 'a']} ``` * docs `Dependents` updated statistics (#14461) Updated statistics for the dependents (packages dependent on `langchain` package). Only packages with 100+ starts * core[patch], langchain[patch], experimental[patch]: import CI (#14414) * Update mongodb_atlas docs for GA (#14425) Updated the MongoDB Atlas Vector Search docs to indicate the service is Generally Available, updated the example to use the new index definition, and added an example that uses metadata pre-filtering for semantic search --------- Co-authored-by: Harrison Chase <hw.chase.17@gmail.com> * infra: add langchain-community release workflow (#14469) * docs `networkx`update (#14426) Added setting up instruction, package description and link * experimental[patch]: SmartLLMChain Output Key Customization (#14466) **Description** The `SmartLLMChain` was was fixed to output key "resolution". Unfortunately, this prevents the ability to use multiple `SmartLLMChain` in a `SequentialChain` because of colliding output keys. This change simply gives the option the customize the output key to allow for sequential chaining. The default behavior is the same as the current behavior. Now, it's possible to do the following: ``` from langchain.chat_models import ChatOpenAI from langchain.prompts import PromptTemplate from langchain_experimental.smart_llm import SmartLLMChain from langchain.chains import SequentialChain joke_prompt = PromptTemplate( input_variables=["content"], template="Tell me a joke about {content}.", ) review_prompt = PromptTemplate( input_variables=["scale", "joke"], template="Rate the following joke from 1 to {scale}: {joke}" ) llm = ChatOpenAI(temperature=0.9, model_name="gpt-4-32k") joke_chain = SmartLLMChain(llm=llm, prompt=joke_prompt, output_key="joke") review_chain = SmartLLMChain(llm=llm, prompt=review_prompt, output_key="review") chain = SequentialChain( chains=[joke_chain, review_chain], input_variables=["content", "scale"], output_variables=["review"], verbose=True ) response = chain.run({"content": "chickens", "scale": "10"}) print(response) ``` --------- Co-authored-by: Erick Friis <erick@langchain.dev> * Update README and vectorstore path for multi-modal template (#14473) * docs[patch]: link and description cleanup (#14471) Fixed inconsistencies; added links and descriptions --------- Co-authored-by: Erick Friis <erickfriis@gmail.com> * langchain[patch], docs[patch]: use byte store in multivectorretriever (#14474) * manual mapping (#14422) * docs[patch]: add missing imports for local_llms (#14453)  Keeping it consistent with everywhere else in the docs and adding the missing imports to be able to copy paste and run the code example. --------- Co-authored-by: Erick Friis <erick@langchain.dev> * docs[patch]: `microsoft` platform page update (#14476) Added `presidio` and `OneNote` references to `microsoft.mdx`; added link and description to the `presidio` notebook --------- Co-authored-by: Erick Friis <erickfriis@gmail.com> * docs[patch]: `google` platform page update (#14475) Added missed tools --------- Co-authored-by: Erick Friis <erickfriis@gmail.com> * RunnableWithMessageHistory: Fix input schema (#14516) Input schema should not have history key * community[major], core[patch], langchain[patch], experimental[patch]: Create langchain-community (#14463) Moved the following modules to new package langchain-community in a backwards compatible fashion: ``` mv langchain/langchain/adapters community/langchain_community mv langchain/langchain/callbacks community/langchain_community/callbacks mv langchain/langchain/chat_loaders community/langchain_community mv langchain/langchain/chat_models community/langchain_community mv langchain/langchain/document_loaders community/langchain_community mv langchain/langchain/docstore community/langchain_community mv langchain/langchain/document_transformers community/langchain_community mv langchain/langchain/embeddings community/langchain_community mv langchain/langchain/graphs community/langchain_community mv langchain/langchain/llms community/langchain_community mv langchain/langchain/memory/chat_message_histories community/langchain_community mv langchain/langchain/retrievers community/langchain_community mv langchain/langchain/storage community/langchain_community mv langchain/langchain/tools community/langchain_community mv langchain/langchain/utilities community/langchain_community mv langchain/langchain/vectorstores community/langchain_community mv langchain/langchain/agents/agent_toolkits community/langchain_community mv langchain/langchain/cache.py community/langchain_community mv langchain/langchain/adapters community/langchain_community mv langchain/langchain/callbacks community/langchain_community/callbacks mv langchain/langchain/chat_loaders community/langchain_community mv langchain/langchain/chat_models community/langchain_community mv langchain/langchain/document_loaders community/langchain_community mv langchain/langchain/docstore community/langchain_community mv langchain/langchain/document_transformers community/langchain_community mv langchain/langchain/embeddings community/langchain_community mv langchain/langchain/graphs community/langchain_community mv langchain/langchain/llms community/langchain_community mv langchain/langchain/memory/chat_message_histories community/langchain_community mv langchain/langchain/retrievers community/langchain_community mv langchain/langchain/storage community/langchain_community mv langchain/langchain/tools community/langchain_community mv langchain/langchain/utilities community/langchain_community mv langchain/langchain/vectorstores community/langchain_community mv langchain/langchain/agents/agent_toolkits community/langchain_community mv langchain/langchain/cache.py community/langchain_community ``` Moved the following to core ``` mv langchain/langchain/utils/json_schema.py core/langchain_core/utils mv langchain/langchain/utils/html.py core/langchain_core/utils mv langchain/langchain/utils/strings.py core/langchain_core/utils cat langchain/langchain/utils/env.py >> core/langchain_core/utils/env.py rm langchain/langchain/utils/env.py ``` See .scripts/community_split/script_integrations.sh for all changes * Move runnable context to beta (#14507)  * community[patch]: Fix agenttoolkits imports (#14559) * core[patch]: Release 0.0.13 (#14558) * infra: Turn release branch check back on (#14563) * infra: import CI fix (#14562) TIL `**` globstar doesn't work in make Makefile changes fix that. `__getattr__` changes allow import of all files, but raise error when accessing anything from the module. file deletions were corresponding libs change from #14559 * community[patch]: Release 0.0.1 (#14565) * infra: import CI speed (#14566) Was taking 10 mins. Now a few seconds. * langchain[patch]: Release 0.0.349 (#14570) * experimental[patch]: Release 0.0.46 (#14572) * infra: import checking bugfix (#14569) * docs[patch], templates[patch]: Import from core (#14575) Update imports to use core for the low-hanging fruit changes. Ran following ```bash git grep -l 'langchain.schema.runnable' {docs,templates,cookbook} | xargs sed -i '' 's/langchain\.schema\.runnable/langchain_core.runnables/g' git grep -l 'langchain.schema.output_parser' {docs,templates,cookbook} | xargs sed -i '' 's/langchain\.schema\.output_parser/langchain_core.output_parsers/g' git grep -l 'langchain.schema.messages' {docs,templates,cookbook} | xargs sed -i '' 's/langchain\.schema\.messages/langchain_core.messages/g' git grep -l 'langchain.schema.chat_histry' {docs,templates,cookbook} | xargs sed -i '' 's/langchain\.schema\.chat_history/langchain_core.chat_history/g' git grep -l 'langchain.schema.prompt_template' {docs,templates,cookbook} | xargs sed -i '' 's/langchain\.schema\.prompt_template/langchain_core.prompts/g' git grep -l 'from langchain.pydantic_v1' {docs,templates,cookbook} | xargs sed -i '' 's/from langchain\.pydantic_v1/from langchain_core.pydantic_v1/g' git grep -l 'from langchain.tools.base' {docs,templates,cookbook} | xargs sed -i '' 's/from langchain\.tools\.base/from langchain_core.tools/g' git grep -l 'from langchain.chat_models.base' {docs,templates,cookbook} | xargs sed -i '' 's/from langchain\.chat_models.base/from langchain_core.language_models.chat_models/g' git grep -l 'from langchain.llms.base' {docs,templates,cookbook} | xargs sed -i '' 's/from langchain\.llms\.base\ /from langchain_core.language_models.llms\ /g' git grep -l 'from langchain.embeddings.base' {docs,templates,cookbook} | xargs sed -i '' 's/from langchain\.embeddings\.base/from langchain_core.embeddings/g' git grep -l 'from langchain.vectorstores.base' {docs,templates,cookbook} | xargs sed -i '' 's/from langchain\.vectorstores\.base/from langchain_core.vectorstores/g' git grep -l 'from langchain.agents.tools' {docs,templates,cookbook} | xargs sed -i '' 's/from langchain\.agents\.tools/from langchain_core.tools/g' git grep -l 'from langchain.schema.output' {docs,templates,cookbook} | xargs sed -i '' 's/from langchain\.schema\.output\ /from langchain_core.outputs\ /g' git grep -l 'from langchain.schema.embeddings' {docs,templates,cookbook} | xargs sed -i '' 's/from langchain\.schema\.embeddings/from langchain_core.embeddings/g' git grep -l 'from langchain.schema.document' {docs,templates,cookbook} | xargs sed -i '' 's/from langchain\.schema\.document/from langchain_core.documents/g' git grep -l 'from langchain.schema.agent' {docs,templates,cookbook} | xargs sed -i '' 's/from langchain\.schema\.agent/from langchain_core.agents/g' git grep -l 'from langchain.schema.prompt ' {docs,templates,cookbook} | xargs sed -i '' 's/from langchain\.schema\.prompt\ /from langchain_core.prompt_values /g' git grep -l 'from langchain.schema.language_model' {docs,templates,cookbook} | xargs sed -i '' 's/from langchain\.schema\.language_model/from langchain_core.language_models/g' ``` * docs[patch]: Fix embeddings example for Databricks (#14576)  Fix `from langchain.llms import DatabricksEmbeddings` to `from langchain.embeddings import DatabricksEmbeddings`. Signed-off-by: harupy <17039389+harupy@users.noreply.github.com> * docs[patch]: update installation with core and community (#14577) * Update RunnableWithMessageHistory (#14351) This PR updates RunnableWithMessage history to support user specific configuration for the factory. It extends support to passing multiple named arguments into the factory if the factory takes more than a single argument. * Add Gmail Agent Example (#14567) Co-authored-by: Harrison Chase <hw.chase.17@gmail.com> * allow other namespaces (#14606) * core[minor]: Release 0.1.0 (#14607) * community[patch]: Release 0.0.2 (#14610) * langchain[patch]: Release 0.0.350 (#14612) * Add image (#14611) * experimental[patch]: Release 0.0.47 (#14617) * docs: core and community readme (#14623) * docs: fix links in readme (#14624) * Minor update to ensemble retriever to handle a mix of Documents or str (#14552) * Update Docugami Cookbook (#14626) **Description:** Update the information in the Docugami cookbook. Fix broken links and add information on our kg-rag template. Co-authored-by: Kenzie Mihardja <kenzie@docugami.com> * docs: update multi_modal_RAG_chroma.ipynb (#14602) seperate -> separate  * templates[patch]: fix pydantic imports (#14632) * fix a bug in RedisNum filter againt value 0 (#14587) - **Description:** There is a bug in RedisNum filter that filter towards value 0 will be parsed as "*". This is a fix to it. - **Issue:** NA - **Dependencies:** NA - **Tag maintainer:** NA - **Twitter handle:** NA * create mypy cache dir if it doesn't exist (#14579) ### Description When running `make lint` multiple times, i can see the error `mkdir: .mypy_cache: File exists`. Use `mkdir -p` to solve this problem. <img width="1512" alt="Screenshot 2023-12-12 at 11 22 01 AM" src="https://github.com/langchain-ai/langchain/assets/10000925/1429383d-3283-4e22-8882-5693bc50b502"> * fix: to rag-semi-structured template (#14568) **Description:** Fixes to rag-semi-structured template. - Added required libraries - pdfminer was causing issues when installing with pip. pdfminer.six works best - Changed the pdf name for demo from llama2 to llava  * docs `ollama` pages (#14561) added provider page; fixed broken links. * infra: rm community split scripts (#14633) * docs: update langchain diagram (#14619) * Update cohere provider docs (#14528) Preview since github wont preview .mdx : <img width="1401" alt="image" src="https://github.com/langchain-ai/langchain/assets/144115527/9e8ba3d9-24ff-4584-9da3-2c9b60e7e624"> * Added notebook tutorial on using Yellowbrick as a vector store with LangChain (#14509) - **Description:** a notebook documenting Yellowbrick as a vector store usage --------- Co-authored-by: markcusack <markcusack@markcusacksmac.lan> Co-authored-by: markcusack <markcusack@Mark-Cusack-sMac.local> * docs[patch] Fix some typos in merger_retriever.ipynb (#14502) This patch fixes some typos.  Signed-off-by: Masanari Iida <standby24x7@gmail.com> * feat: Yaml output parser (#14496) ## Description New YAML output parser as a drop-in replacement for the Pydantic output parser. Yaml is a much more token-efficient format than JSON, proving to be **~35% faster and using the same percentage fewer completion tokens**. ☑️ Formatted ☑️ Linted ☑️ Tested (analogous to the existing`test_pydantic_parser.py`) The YAML parser excels in situations where a list of objects is required, where the root object needs no key: ```python class Products(BaseModel): __root__: list[Product] ``` I ran the prompt `Generate 10 healthy, organic products` 10 times on one chain using the `PydanticOutputParser`, the other one using the`YamlOutputParser` with `Products` (see below) being the targeted model to be created. LLMs used were Fireworks' `lama-v2-34b-code-instruct` and OpenAI `gpt-3.5-turbo`. All runs succeeded without validation errors. ```python class Nutrition(BaseModel): sugar: int = Field(description="Sugar in grams") fat: float = Field(description="% of daily fat intake") class Product(BaseModel): name: str = Field(description="Product name") stats: Nutrition class Products(BaseModel): """A list of products""" products: list[Product] # Used `__root__` for the yaml chain ``` Stats after 10 runs reach were as follows: ### JSON ø time: 7.75s ø tokens: 380.8 ### YAML ø time: 5.12s ø tokens: 242.2 Looking forward to feedback, tips and contributions! * [docs]: add missing tiktoken dependency (#14497) Description: I was following the docs and got an error about missing tiktoken dependency. Adding it to the comment where the langchain and docarray libs are. * fix(embeddings): huggingface hub embeddings and TEI (#14489) **Description:** This PR fixes `HuggingFaceHubEmbeddings` by making the API token optional (as in the client beneath). Most models don't require one. I also updated the notebook for TEI (text-embeddings-inference) accordingly as requested here #14288. In addition, I fixed a mistake in the POST call parameters. **Tag maintainers:** @baskaryan * Fix token_usage None issue in ChatOpenAI with local Chatglm2-6B (#14493) When using local Chatglm2-6B by changing OPENAI_BASE_URL to localhost, the token_usage in ChatOpenAI becomes None. This leads to an AttributeError when trying to access token_usage.items(). This commit adds a check to ensure token_usage is not None before accessing its items. This change prevents the AttributeError and allows ChatOpenAI to work seamlessly with a local Chatglm2-6B model, aligning with the way it operates with the OpenAI API.  Co-authored-by: Harrison Chase <hw.chase.17@gmail.com> * DOC: model update in 'Using OpenAI Functions' docs (#14486) - **Description:** : I just update the openai functions docs to use the latest model (ex. gpt-3.5-turbo-1106) https://python.langchain.com/docs/modules/chains/how_to/openai_functions The reason is as follow: After reviewing the OpenAI Function Calling official guide at https://platform.openai.com/docs/guides/function-calling, the following information was noted: > "The latest models (gpt-3.5-turbo-1106 and gpt-4-1106-preview) have been trained to both detect when a function should be called (depending on the input) and to respond with JSON that adheres to the function signature more closely than previous models. With this capability also comes potential risks. We strongly recommend building in user confirmation flows before taking actions that impact the world on behalf of users (sending an email, posting something online, making a purchase, etc)." CC: @efriis * docs: Add Databricks Vector Search example notebook (#14158) This PR adds an example notebook for the Databricks Vector Search vector store. It also adds an introduction to the Databricks Vector Search product on the Databricks's provider page. --------- Co-authored-by: Bagatur <baskaryan@gmail.com> * Fixed `DeprecationWarning` for `PromptTemplate.from_file` module-level calls (#14468) Resolves https://github.com/langchain-ai/langchain/issues/14467 * Add Gemini Notebook (#14661) * cli[patch]: integration template (#14571) * Fix RRF and lucene escape characters for neo4j vector store (#14646) * Remove Lucene special characters (fixes https://github.com/langchain-ai/langchain/issues/14232) * Fixes RRF normalization for hybrid search * [Nit] Add newline in notebook (#14665) For bullet list formatting * infra: skip extended testing for partner packages (#14630) Tested by merging into #14627 * cli[patch]: rc (#14667) * Update Vertex AI to include Gemini (#14670) h/t to @lkuligin - **Description:** added new models on VertexAI - **Twitter handle:** @lkuligin --------- Co-authored-by: Leonid Kuligin <lkuligin@yandex.ru> Co-authored-by: Harrison Chase <hw.chase.17@gmail.com> * cli[patch]: unicode issue (#14672) Some operating systems compile template, resulting in unicode decode errors * communty[patch]: Release 0.0.3 (#14673) * [Partner] Add langchain-google-genai package (gemini) (#14621) Add a new ChatGoogleGenerativeAI class in a `langchain-google-genai` package. Still todo: add a deprecation warning in PALM --------- Co-authored-by: Erick Friis <erick@langchain.dev> Co-authored-by: Leonid Kuligin <lkuligin@yandex.ru> Co-authored-by: Bagatur <baskaryan@gmail.com> * \Fix tool_calls message merge (#14613)  * google-genai[patch]: Release 0.0.2 (#14677) * Wfh/google docs update (#14676) - Add gemini references - Fix the notebook (ultra isn't generally available; also gemini will randomly filter out responses, so added a fallback) --------- Co-authored-by: Leonid Kuligin <lkuligin@yandex.ru> * docs: build partner api refs (#14675) * docs: fix api ref link (#14679) Don't point to stable, let api docs choose default version * docs: per-package version in api docs (#14683) * core[patch]: Fix runnable with message history (#14629) Fix bug shown in #14458. Namely, that saving inputs to history fails when the input to base runnable is a list of messages * templates[patch]: Add cohere librarian template (#14601) Adding the example I build for the Cohere hackathon. It can: use a vector database to reccommend books <img width="840" alt="image" src="https://github.com/langchain-ai/langchain/assets/144115527/96543a18-217b-4445-ab4b-950c7cced915"> Use a prompt template to provide information about the library <img width="834" alt="image" src="https://github.com/langchain-ai/langchain/assets/144115527/996c8e0f-cab0-4213-bcc9-9baf84f1494b"> Use Cohere RAG to provide grounded results <img width="822" alt="image" src="https://github.com/langchain-ai/langchain/assets/144115527/7bb4a883-5316-41a9-9d2e-19fd49a43dcb"> --------- Co-authored-by: Erick Friis <erick@langchain.dev> * docs[patch]: fix bullet points (#14684) - docs fixes - escape - bullets * community[patch]: Fixed issue with importing Row from sqlalchemy (#14488) - **Description:** Fixed import of Row in cache.py, - **Issue:** the issue # #13464 https://creditone.us.to/langchain-ai/langchain/issues/13464, - **Dependencies:** None, - **Twitter handle:** @frankybridman Co-authored-by: Harrison Chase <hw.chase.17@gmail.com> * community[patch]: Correct type annotation for azure_ad_token_provider Closed: #14402 (#14432) Description Fix https://github.com/langchain-ai/langchain/issues/14402, Similar changes: https://github.com/langchain-ai/langchain/pull/14166 Twitter handle [lin_bob57617](https://twitter.com/lin_bob57617) * community[patch]: fix dashvector endpoint params error (#14484)  Co-authored-by: fangkeke <3339698829@qq.com> Co-authored-by: Harrison Chase <hw.chase.17@gmail.com> * docs: platform pages update (#14637) Updated examples and platform pages. - added missed tools - added links and descriptions * docs: api ref nav Python Docs -> Docs (#14686) * Template for multi-modal w/ multi-vector (#14618) Results - ![image](https://github.com/langchain-ai/langchain/assets/122662504/16bac14d-74d7-47b1-aed0-72ae25a81f39) * Gemini multi-modal RAG template (#14678) ![Screenshot 2023-12-13 at 12 53 39 PM](https://github.com/langchain-ai/langchain/assets/122662504/a6bc3b0b-f177-4367-b9c8-b8862c847026) * [Partner] Gemini Embeddings (#14690) Add support for Gemini embeddings in the langchain-google-genai package * [Integration] NVIDIA AI Playground (#14648) Description: Added NVIDIA AI Playground Initial support for a selection of models (Llama models, Mistral, etc.) Dependencies: These models do depend on the AI Playground services in NVIDIA NGC. API keys with a significant amount of trial compute are available (10K queries as of the time of writing). H/t to @VKudlay * [Workflows] Add nvidia-aiplay to _release.yml (#14722) As the title says. In the future will want to have a script to automate this * Add dense proposals (#14719) Indexing strategy based on decomposing candidate propositions while indexing. * [Hub|tracing] Tag hub prompts (#14720) If you're using the hub, you'll likely be interested in tracking the commit/object when tracing. This PR adds it to the config * Update `google_generative_ai.ipynb` (#14704) * docs[patch]: fix databricks metadata (#14727) * docs: updated branding for Google AI (#14728) Replace this entire comment with: - **Description:** a small fix in branding * infra: add integration test workflow (#14688) * infra: Pre-release integration tests for partner pkgs (#14687) * google-genai[patch]: add google-genai integration deps and extras (#14731) * community[patch]: fix pgvector sqlalchemy (#14726) Fixes #14699 * infra: add action checkout to pre-release-checks (#14732) * Revert "[Hub|tracing] Tag hub prompts" (#14735) Reverts langchain-ai/langchain#14720 * core[patch]: Release 0.1.1 (#14738) * infra: docs build install community editable (#14739) * infra: fix pre-release integration test and add unit test (#14742) * docs: Remove trailing "`" in pip install command (#14730) hi! just a simple typo fix in the local LLM python docs - **Description:** removing a trailing "\`" character in a `!pip install ...` command - **Issue:** n/a - **Dependencies:** n/a - **Tag maintainer:** n/a - **Twitter handle:** n/a * google-genai[patch], community[patch]: Added support for new Google GenerativeAI models (#14530) Replace this entire comment with: - **Description:** added support for new Google GenerativeAI models - **Twitter handle:** lkuligin --------- Co-authored-by: Erick Friis <erick@langchain.dev> * [Tracing] String Stacktrace (#14131) Add full stacktrace * [Evals] End project (#14324) Also does some cleanup. Now that we support updating/ending projects, do this automatically. Then you can edit the name of the project in the app. * Fix OAI Tool Message (#14746) See format here: https://platform.openai.com/docs/guides/function-calling/parallel-function-calling It expects a "name" argument, which we aren't providing by default. ![image](https://github.com/langchain-ai/langchain/assets/13333726/7cd82978-337c-40a1-b099-3bb25cd57eb4) Alternative is to add the 'name' field directly to the message if people prefer. * Update propositional-retrieval template (#14766) More descriptive name. Add parser in ingest. Update image link * [Documentation] Updates to NVIDIA Playground/Foundation Model naming.… (#14770) … (#14723) - **Description:** Minor updates per marketing requests. Namely, name decisions (AI Foundation Models / AI Playground) - **Tag maintainer:** @hinthornw Do want to pass around the PR for a bit and ask a few more marketing questions before merge, but just want to make sure I'm not working in a vacuum. No major changes to code functionality intended; the PR should be for documentation and only minor tweaks. Note: QA model is a bit borked across staging/prod right now. Relevant teams have been informed and are looking into it, and I'm placeholdered the response to that of a working version in the notebook. Co-authored-by: Vadim Kudlay <32310964+VKudlay@users.noreply.github.com> * community[minor]: Add SurrealDB vectorstore (#13331) **Description:** Vectorstore implementation around [SurrealDB](https://www.surrealdb.com) --------- Co-authored-by: Bagatur <baskaryan@gmail.com> * langchain[patch]: remove unused imports (#14680) Co-authored-by: Bagatur <baskaryan@gmail.com> * docs: `Steam` update (#14778) Updated the page title. It was inconsistent. Updated page with links; description and setting details. * docs: `cloudflare` update (#14779) Added provider page. Added links, descriptions * Add image support for Ollama (#14713) Support [LLaVA](https://ollama.ai/library/llava): * Upgrade Ollama * `ollama pull llava` Ensure compatibility with [image prompt template](https://github.com/langchain-ai/langchain/pull/14263) --------- Co-authored-by: jacoblee93 <jacoblee93@gmail.com> * docs: `google drive` update (#14781) The [Google Drive toolkit](https://python.langchain.com/docs/integrations/toolkits/google_drive) page is a duplicate of the [Google Drive tool](https://python.langchain.com/docs/integrations/tools/google_drive) page. - Removed the `Google Drive toolkit` page (it shouldn't be a toolkit but tool) - Removed the correspondent reference in the Google platform page - Redirected the removed page to the tool page. * community[patch]: Update YandexGPT API (#14773) Update LLMand Chat model to use new api version --------- Co-authored-by: Dmitry Tyumentsev <dmitry.tyumentsev@raftds.com> * community[patch]: Implement similarity_score_threshold for MongoDB Vector Store (#14740) Adds the option for `similarity_score_threshold` when using `MongoDBAtlasVectorSearch` as a vector store retriever. Example use: ``` vector_search = MongoDBAtlasVectorSearch.from_documents(...) qa_retriever = vector_search.as_retriever( search_type="similarity_score_threshold", search_kwargs={ "score_threshold": 0.5, } ) qa = RetrievalQA.from_chain_type( llm=OpenAI(), chain_type="stuff", retriever=qa_retriever, ) docs = qa({"query": "..."}) ``` I've tested this feature locally, using a MongoDB Atlas Cluster with a vector search index. * docs[patch]: fix zoom (#14786) not sure why quarto is removing divs * Permit updates in indexing (#14482) * docs: developer docs (#14776) Builds out a developer documentation section in the docs - Links it from contributing.md - Adds an initial guide on how to contribute an integration --------- Co-authored-by: Bagatur <baskaryan@gmail.com> * infra: cut down on integration steps (#14785)  --------- Co-authored-by: Bagatur <baskaryan@gmail.com> * community[patch]: support for Sybase SQL anywhere added. (#14821) - **Description:** support for Sybase SQL anywhere added in sql_database.py file at path langchain\libs\community\langchain_community\utilities - **Issue:** It will resolve default schema setting for Sybase SQL anywhere - **Dependencies:** No, - **Tag maintainer:** @baskaryan, @eyurtsev, @hwchase17, - **Twitter handle:** NA --------- Co-authored-by: learn360sujeet <121271779+learn360sujeet@users.noreply.github.com> Co-authored-by: Bagatur <baskaryan@gmail.com> * community[patch]: fix agenerate return value (#14815) Fixed: - `_agenerate` return value in the YandexGPT Chat Model - duplicate line in the documentation Co-authored-by: Dmitry Tyumentsev <dmitry.tyumentsev@raftds.com> * docs: ensure consistency in declaring LANGCHAIN_API_KEY... (#14823) ... variable, accompanied by a quote Co-authored-by: Yacine Bouakkaz <Yacine.Bouakkaz@evokegroup.com> * docs redundant pages (#14774) [ScaNN](https://python.langchain.com/docs/integrations/providers/scann) and [DynamoDB](https://python.langchain.com/docs/integrations/platforms/aws#aws-dynamodb) pages in `providers` are redundant because we have those references in the Google and AWS platform pages. It is confusing. - I removed unnecessary pages, redirected files to new nams; * docs: Typo in Templates README.md (#14812) Corrected path reference from package/pirate-speak to packages/pirate-speak * community: Add logprobs in gen output (#14826) Now that it's supported again for OAI chat models . Shame this wouldn't include it in the `.invoke()` output though (it's not included in the message itself). Would need to do a follow-up for that to be the case * docs: Fix link typo to `/docs/integrations/text_embedding/nvidia_ai_endpoints` (#14827) This page doesn't exist: - https://python.langchain.com/docs/integrations/text_embeddings/nvidia_ai_endpoints but this one does: - https://python.langchain.com/docs/integrations/text_embedding/nvidia_ai_endpoints * docs: typo in rag use case (#14800) Description: Fixes minor typo to documentation * docs: Fix the broken link to Extraction page (#14806) **Description:** fixing a broken link to the extraction doc page * community[minor]: New model parameters and dynamic batching for VertexAIEmbeddings (#13999) - **Description:** VertexAIEmbeddings performance improvements - **Twitter handle:** @vladkol ## Improvements - Dynamic batch size, starting from 250, lowering down to 5. Batch size varies across regions. Some regions support larger batches, and it significantly improves performance. When running large batches of texts in `us-central1`, performance gain can be up to 3.5x. The dynamic batching also makes sure every batch is below 20K token limit. - New model parameter `embeddings_type` that translates to `task_type` parameter of the API. Newer model versions support [different embeddings task types](https://cloud.google.com/vertex-ai/docs/generative-ai/embeddings/get-text-embeddings#api_changes_to_models_released_on_or_after_august_2023). * Update parser (#14831) Gpt-3.5 sometimes calls with empty string arguments instead of `{}` I'd assume it's because the typescript representation on their backend makes it a bit ambiguous. * [Bugfix] Ensure tool output is a str, for OAI Assistant (#14830) Tool outputs have to be strings apparently. Ensure they are formatted correctly before passing as intermediate steps. ``` BadRequestError: Error code: 400 - {'error': {'message': '1 validation error for Request\nbody -> tool_outputs -> 0 -> output\n str type expected (type=type_error.str)', 'type': 'invalid_request_error', 'param': None, 'code': None}} ``` * community[patch]: Update Tongyi default model_name (#14844) <img width="1305" alt="Screenshot 2023-12-18 at 9 54 01 PM" src="https://github.com/langchain-ai/langchain/assets/10000925/c943fd81-cd48-46eb-8dff-4680424d9ba9"> The current model is no longer available. * docs: update NVIDIA integration (#14780) - **Description:** Modification of descriptions for marketing purposes and transitioning towards `platforms` directory if possible. - **Issue:** Some marketing opportunities, lodging PR and awaiting later discussions. - This PR is intended to be merged when decisions settle/hopefully after further considerations. Submitting as Draft for now. Nobody @'d yet. --------- Co-authored-by: Bagatur <baskaryan@gmail.com> * docs[patch]: gemini keywords (#14856) * docs[patch]: more keywords (#14858) * community[patch]: Release 0.0.4 (#14864) * langchain[patch]: Release 0.0.351 (#14867) * add methods to deserialize prompts that were old (#14857) * community: replace deprecated davinci models (#14860) This is technically a breaking change because it'll switch out default models from `text-davinci-003` to `gpt-3.5-turbo-instruct`, but OpenAI is shutting off those endpoints on 1/4 anyways. Feels less disruptive to switch out the default instead. * WIP: sql research assistant (#14240) * docstrings `core` update (#14871) Added missed docstrings * Fix token text splitter duplicates (#14848) - **Description:** - Add a break case to `text_splitter.py::split_text_on_tokens()` to avoid unwanted item at the end of result. - Add a testcase to enforce the behavior. - **Issue:** - #14649 - #5897 - **Dependencies:** n/a, --- **Quick illustration of change:** ``` text = "foo bar baz 123" tokenizer = Tokenizer( chunk_overlap=3, tokens_per_chunk=7 ) output = split_text_on_tokens(text=text, tokenizer=tokenizer) ``` output before change: `["foo bar", "bar baz", "baz 123", "123"]` output after change: `["foo bar", "bar baz", "baz 123"]` * docstrings `langchain` update (#14870) Added missed docstrings * docs: fixed tiktoken link error (#14840)  - **Description:** fixed tiktoken link error, - **Issue:** no, - **Dependencies:** no, - **Tag maintainer:** @baskaryan, - **Twitter handle:** SignetCode! * docs: fix typo in contributing re installing integration test deps (#14861) **Description** The contributing docs lists a poetry command to install community for dev work that includes a poetry group called `integration_tests`. This is a mistake: the poetry group for integration tests is called `test_integration`, not `integration_tests`. See here: https://github.com/langchain-ai/langchain/blob/master/libs/community/pyproject.toml#L119 * Update kendra.py to avoid Kendra query ValidationException (#14866) Fixing issue - https://github.com/langchain-ai/langchain/issues/14494 to avoid Kendra query ValidationException  --------- Co-authored-by: Harrison Chase <hw.chase.17@gmail.com> * Harrison/agent docs custom (#14877) * Improve prompt injection detection (#14842) - **Description:** This is addition to [my previous PR](https://github.com/langchain-ai/langchain/pull/13930) with improvements to flexibility allowing different models and notebook to use ONNX runtime for faster speed. Since the last PR, [our model](https://huggingface.co/laiyer/deberta-v3-base-prompt-injection) got more than 660k downloads, and with the [public benchmark](https://huggingface.co/spaces/laiyer/prompt-injection-benchmark) showed much fewer false-positives than the previous one from deepset. Additionally, on the ONNX runtime, it can be running 3x faster on the CPU, which might be handy for builders using Langchain. **Issue:** N/A - **Dependencies:** N/A - **Tag maintainer:** N/A - **Twitter handle:** `@laiyer_ai` * OPENAI_PROXY not working (#14833) Replace this entire comment with: - **Description:** OPENAI_PROXY is not working for openai==1.3.9, The `proxies` argument is deprecated. The `http_client` argument should be passed instead, - **Issue:** OPENAI_PROXY is not working, - **Dependencies:** None, - **Tag maintainer:** @hwchase17 , - **Twitter handle:** timothy66666 * Docs `tencent` pages update (#14879) - updated `Tencent` provider page: added a chat model and document loader references; company description - updated Chat model and Document loader pages with descriptions, links - renamed files to consistent formats; redirected file names Note: I was getting this linting error on code that **was not changed in my PR**! > Error: docs/docs/guides/safety/hugging_face_prompt_injection.ipynb:1:1: I001 Import block is un-sorted or un-formatted > make: *** [Makefile:47: lint_package] Error 1 I've fixed this error in the notebook * added history and support for system_message as param (#14824) - **Description:** added support for chat_history for Google GenerativeAI (to actually use the `chat` API) plus since Gemini currently doesn't have a support for SystemMessage, added support for it only if a user provides additional `convert_system_message_to_human` flag during model initialization (in this case, SystemMessage would be prepanded to the first HumanMessage) - **Issue:** #14710 - **Dependencies:** any dependencies required for this change, - **Tag maintainer:** for a quicker response, tag the relevant maintainer (see below), - **Twitter handle:** lkuligin --------- Co-authored-by: William FH <13333726+hinthornw@users.noreply.github.com> * [Partner] Google GenAi new release (#14882) to support the system message merging Also fix integration tests that weren't passing * [Partner] Update google integration test (#14883) Gemini has decided that pickle rick is unsafe: https://github.com/langchain-ai/langchain/actions/runs/7256642294/job/19769249444#step:8:189 ![image](https://github.com/langchain-ai/langchain/assets/13333726/cfbf4312-53b6-4290-84ee-6ce0742e739e) * [Partner] NVIDIA TRT Package (#14733) Simplify #13976 and add as a separate package. - [] Add README - [X] Add doc notebook - [X] Add simple LLM integration --------- Co-authored-by: Jeremy Dyer <jdye64@gmail.com> * docs: docstrings `langchain_community` update (#14889) Addded missed docstrings. Fixed inconsistency in docstrings. **Note** CC @efriis There were PR errors on `langchain_experimental/prompt_injection_identifier/hugging_face_identifier.py` But, I didn't touch this file in this PR! Can it be some cache problems? I fixed this error. * docs: add reference for XataVectorStore constructor (#14903) Adds doc reference to the XataVectorStore constructor for use with existing Xata table contents. @tsg @philkra * langchain[patch]: export sagemaker LLMContentHandler (#14906) Resolves #14904 …

langchain[patch]: fix ChatVertexAI streaming

2670e6e

dosubot bot added size:XS This PR changes 0-9 lines, ignoring generated files. Ɑ: models Related to LLMs or chat model modules 🤖:bug Related to a bug, vulnerability, unexpected error with an existing feature labels Dec 7, 2023

documentation update

d96e08a

dosubot bot added size:L This PR changes 100-499 lines, ignoring generated files. and removed size:XS This PR changes 0-9 lines, ignoring generated files. labels Dec 7, 2023

remove pip output

b2de08f

vercel bot deployed to Preview December 7, 2023 00:43 View deployment

scheduled stream test

4a43b09

vercel bot deployed to Preview December 7, 2023 01:57 View deployment

format

76ae737

vercel bot deployed to Preview December 7, 2023 02:05 View deployment

scheduled test

0f6b066

vercel bot deployed to Preview December 7, 2023 17:44 View deployment

efriis merged commit 54040b0 into master Dec 7, 2023
40 checks passed

efriis deleted the erick/langchain-patch---fix-chatvertexai-streaming- branch December 7, 2023 17:46

aymeric-roucher pushed a commit to andrewrreed/langchain that referenced this pull request Dec 11, 2023

langchain[patch]: fix ChatVertexAI streaming (langchain-ai#14369)

8536011

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

langchain[patch]: fix ChatVertexAI streaming #14369

langchain[patch]: fix ChatVertexAI streaming #14369

efriis commented Dec 7, 2023

vercel bot commented Dec 7, 2023 •

edited

langchain[patch]: fix ChatVertexAI streaming #14369

langchain[patch]: fix ChatVertexAI streaming #14369

Conversation

efriis commented Dec 7, 2023

vercel bot commented Dec 7, 2023 • edited

vercel bot commented Dec 7, 2023 •

edited