Oracle Cloud Infrastructure (OCI) Generative AI is a fully managed service that provides a set of state-of-the-art, customizable large language models (LLMs) that cover a wide range of use cases, and which is available through a single API. Using the OCI Generative AI service you can access ready-to-use pretrained models, or create and host your own fine-tuned custom models based on your own data on dedicated AI clusters. Detailed documentation of the service and API is available here and here. This notebook explains how to use OCI’s Generative AI complete models with LangChain.

​
Setup

Ensure that the oci sdk and the langchain-community package are installed 
!pip install -U oci langchain-community

​
Usage

from langchain_community.llms.oci_generative_ai import OCIGenAI

llm = OCIGenAI(
    model_id="cohere.command",
    service_endpoint="https://inference.generativeai.us-chicago-1.oci.oraclecloud.com",
    compartment_id="MY_OCID",
    model_kwargs={"temperature": 0, "max_tokens": 500},
)

response = llm.invoke("Tell me one fact about earth", temperature=0.7)
print(response)

​
Chaining with prompt templates

from langchain_core.prompts import PromptTemplate

llm = OCIGenAI(
    model_id="cohere.command",
    service_endpoint="https://inference.generativeai.us-chicago-1.oci.oraclecloud.com",
    compartment_id="MY_OCID",
    model_kwargs={"temperature": 0, "max_tokens": 500},
)

prompt = PromptTemplate(input_variables=["query"], template="{query}")
llm_chain = prompt | llm

response = llm_chain.invoke("what is the capital of france?")
print(response)

​
Streaming

llm = OCIGenAI(
    model_id="cohere.command",
    service_endpoint="https://inference.generativeai.us-chicago-1.oci.oraclecloud.com",
    compartment_id="MY_OCID",
    model_kwargs={"temperature": 0, "max_tokens": 500},
)

for chunk in llm.stream("Write me a song about sparkling water."):
    print(chunk, end="", flush=True)

​
Authentication

The authentication methods supported for LlamaIndex are equivalent to those used with other OCI services and follow the standard SDK authentication methods, specifically API Key, session token, instance principal, and resource principal. API key is the default authentication method used in the examples above. The following example demonstrates how to use a different authentication method (session token) 
llm = OCIGenAI(
    model_id="cohere.command",
    service_endpoint="https://inference.generativeai.us-chicago-1.oci.oraclecloud.com",
    compartment_id="MY_OCID",
    auth_type="SECURITY_TOKEN",
    auth_profile="MY_PROFILE",  # replace with your profile name
    auth_file_location="MY_CONFIG_FILE_LOCATION",  # replace with file location where profile name configs present
)

​
Dedicated AI Cluster

To access models hosted in a dedicated AI cluster create an endpoint whose assigned OCID (currently prefixed by ‘ocid1.generativeaiendpoint.oc1.us-chicago-1’) is used as your model ID. When accessing models hosted in a dedicated AI cluster you will need to initialize the OCIGenAI interface with two extra required params (“provider” and “context_size”). 
llm = OCIGenAI(
    model_id="ocid1.generativeaiendpoint.oc1.us-chicago-1....",
    service_endpoint="https://inference.generativeai.us-chicago-1.oci.oraclecloud.com",
    compartment_id="DEDICATED_COMPARTMENT_OCID",
    auth_profile="MY_PROFILE",  # replace with your profile name,
    auth_file_location="MY_CONFIG_FILE_LOCATION",  # replace with file location where profile name configs present
    provider="MODEL_PROVIDER",  # e.g., "cohere" or "meta"
    context_size="MODEL_CONTEXT_SIZE",  # e.g., 128000
)