LangChainAI's LangGraph, a tool for modifying agent runtimes, has been released, enabling robust agent architectures. LangGraph uses Pregel graph structure, channels, topics, and nodes. LangChainAI has revamped its docs to highlight tool usage, making it easy for LLMs to call tools. The LangGraph JS has been released, providing examples for modifying agent runtimes in JavaScript. The article discusses the teamwork between the application layer and infrastructure layer in LLMs. Open source examples for improving LLM accuracy and verifiability have been shared, integrating with Pinecone, LangChainAI, and TogetherCompute.
LangGraph is to agents what LCEL is to chains. It removes mystery and introduces infinite composability, deep inspection, and clear execution into your flow. It's the legos I've been craving to build some really complex agent projects https://t.co/vVBzlBMvbf
Some open source examples for how to make your own LLM be more accurate, up-to-date and verifiable. Awesome to see these integrations with @pinecone @LangChainAI @togethercompute Link in the follow-up tweet. https://t.co/djUHEUeNeA
Ever wondered how LLMs go from lines of code to mind-blowing conversations? It's all about teamwork between two key layers: Application Layer and infrastructure layer. Learn more about LLMs: https://t.co/2L7TlzJOaL https://t.co/Eb8iCdYsEx
π¦πΈοΈLangGraph JS Released LangGraph makes it easy to modify agent runtimes with @LangChainAI. So far in JS we have a couple examples to help you get started: π€Example of replicating the LangChain AgentExecutor π·A chat agent executor (aimed specifically at chat models) Superβ¦ https://t.co/LOZQK40hsN
New awesome article on minimising LLM hallucinations with @pydantic by @jxnlco. https://t.co/cnl5q3fbB9 The coolest thing about LLMs and validation is that all the same principles, know-how and best practices from any other application development apply!
Revamped tool usage docs are in @LangChainAI JS/TS too πͺ! Little known fact: you can use tools with chains as well as agents, depending on how much autonomy you want to give to the LLM. https://t.co/S2tP7yQL0e https://t.co/9HMJoJlOqL
π§Tool Usage w/ LangChain A big use case for LangChain is making it easy for LLMs to call tools. This can range from putting a simple natural-language-interface on top of a function, to agents We've revamped our docs to highlight this use case more. New guides include:β¦ https://t.co/tAMFfrX3s3
Testing out LLM apps fast is key to know what works well. Haven't fully figured all the parts... ...but to me it seems like the πππ¦π§ πππ π£π₯π’π§π’π§π¬π£ππ‘π process for teams simply leverages the #langchain ecosystem: π§΅
Such a great mix of theory & practice on building #RAG applications for production by @activeloopai, @towards_AI, @intel. Hands-on projects with @langchainai & @llama_index. Worth a try! https://t.co/3gXGNEY2WJ π
LangGraph πΈοΈ might just be the next shift to enable much more robust agent architectures through the use of a - Pregel graph structure - channels - topics - nodes - more! Check out our deep dive teaser video to understand LangGraph from the code up - longer video coming soon https://t.co/a1fJnttRnV https://t.co/U8Y9pTn8jw