An open-source framework for LLM applications, connected to Databricks through databricks-langchain. Its prompt templates write variables in single braces, {variable}, whereas the registry uses double braces.
Read more: Databricks docs
In the Ultra Transcenders books
Each book explains LangChain in context, with comparison tables and the common traps.
Terms in this definition
- LLM
Large language model, usually a transformer network with billions of parameters that learnt to predict the next token from vast amounts of text. Compared with a small language model it is more capable across tasks, but slower and costlier.
- Databricks
Analytics platform built on Apache Spark where data is processed in notebooks; Unity Catalog is the governance model it recommends.
- DOUBLE
An 8-byte floating-point type with double precision. A materialized view that sums or averages one may need a full refresh, which casting to
DECIMALavoids.
Related terms
- AzureAIOpenTelemetryTracer
Callback tracer for LangChain and LangGraph that sends OpenTelemetry spans to Foundry and Azure Monitor, but only if supplied as
config={'callbacks':[tracer]}; theenable_content_recordingsetting decides whether message content is captured. - ChatModel
An older MLflow agent interface, now legacy, that takes a chat completions
messageslist as ChatAgent does, whereas ResponsesAgent uses aninputfield. LangChain'sChatDatabrickschat model class is something different. - Format-specific chunking
An approach that cuts documents along structure the format already has, like headers in Markdown or HTML; LangChain's
MarkdownHeaderTextSplitteris one example. - OpenAI Agents SDK
OpenAI's framework for multi-agent work, with an agent runner, sessions and tools, and no reliance on LangChain. Projects for it can be scaffolded with the Agent Bricks CLI.