langchain-nexa-ai 0.1.3

An integration package connecting NexaAI and LangChain

langchain-nexa-ai

This package contains the LangChain integration with Nexa AI.

Installation

pip install -U langchain-core langchain-nexa-ai

And you should get an api key from Nexa Hub and set it as an environment variable (NEXA_API_KEY)

LLMs

NexaAILLM class exposes Octopus LLMs from NexaAI. We currently support four catogories: shopping, conference, streaming, and travel. See our langchain-nexa-ai tutorial and Nexa AI documentation for more details!

from langchain_nexa_ai import NexaAILLM

octopus_llm = NexaAILLM()
result = octopus_llm.invoke("Show recommended products for electronics.", category="shopping")
print(result)

If NEXA_API_KEY is not set in env, you can also pass api_key as an argument when initializing:

octopus_llm = NexaAILLM(api_key=api_key)

You can also pass a list of catogories (corresponding to each of your prompts) when using generate method of NexaAILLM.

result = octopus_llm.generate(
    prompts=[
        "Show recommended products for electronics.",
        "Find a Hotel to stay in Toyko from June 1 to June 10."
    ],
    categories=[
        "shopping",
        "travel"
    ]
)

Exploit LCEL

from langchain_nexa_ai import NexaAILLM
from langchain_core.output_parsers import JsonOutputParser

octopus_llm = NexaAILLM()
parser = JsonOutputParser()

chain = octopus_llm | parser

result = chain.invoke(
    input="Show me recommended electronics products",
    config={
        "llms": {
            "NexaAILLM": {
                "categories": ["shopping"]
            }
        }
    }
)
print(result)

Chat Models

Coming soon.

TODO

• streaming
• docs