LangChain For Generative AI: Using OpenAI LLMs in Python

Learn to work with large language models in LangChain using OpenAI LLMs and Hugging Face, manage tokens and memory, test with T5, and compare base models to GPT.

Embed text with OpenAI text embedding A2002 via LangChain, obtain 1536-dimensional vectors, and compare positive and negative movie reviews using dot-product similarity to reveal sentiment.

Explore prompting best practices to improve model outputs, including clear, specific prompts; delimiter formatting to prevent injections; system messages; and few-shot and chain-of-thought techniques for robust reasoning.

Explore using LangChain's built-in prompt templates to build prompts, automate input formatting, and parse outputs into json, with a chat prompt template extracting a company's name, sector, and valuation.

Explore output parsers in LangChain to automate format instructions and parse json outputs using the response schema module and structured output parser within prompt templates.

Learn chaining in LangChain: a sequence of prompt–LM pairs where each link passes output to the next, forming simple single-input, single-output chains and more complex multiple-input, multiple-output chains.

Build an Arxiv summarizer with LangChain agents, install Arxiv, load a chat model, and fetch the latest 2023 paper to extract its key insight on bias toward African languages.

Revisit vector stores with OpenAI embeddings and Chroma, loading and splitting documents to index chunks for similarity search. Retrieve relevant text as context for answering questions with a transformer-based encoder–decoder.

Apply maximal marginal relevance to retrieve diverse document chunks from a vector db after loading and splitting sources. Use metadata filtering and self query retrieval to focus on encoder content.

Load and split documents, store them in a vector store, and query chunks with LangChain and a GPT-3.5 turbo chat model to answer and return source documents.

Scale arxiv research bot by indexing ten PDFs from arxiv rss feeds into a chrome db using Lang Chain and embeddings, then use a conversational retrieval chain with Open Flamingo.

Hook your chatbot to Wikipedia using LangChain and OpenAI by loading Wikipedia documents, querying related topics, and building a retrieval chain with a vector store.

Learn function calling in LangChain with OpenAI LLMs by converting a math function into a tool, describing json arguments, and using a model with tools via the expression language.

Build agents with tools using LangChain to connect arithmetic tools to a GPT model, enabling automated problem solving with context via prompts and an agent scratchpad.

Build an economics chatbot with LangChain by connecting an LM to Bureau of Labor Statistics data, using tools to fetch CPI, PPI, and unemployment rate series.

Discover a free weekly AI research newsletter that curates arXiv papers on multimodal and text LLMs, embodied agents, robotics, and quantization, with summaries of research questions, methods, results, and takeaways.