LangChain¶
If you have worked with LLMs the you must have definitely come across LangChain. LangChain is a framework that provides off-the-shelf modules for building “Chains” (Pipelines) using AI & Data systems.
You can think of it as a LEGO blocks where you combine different colored or type of blocks to build the final product.
We are introducing a new LangChain compatible LLM module for Horde-Client. With this LLMClient you can easily integrate Horde-Client with your new or existing LangChain pipeline.
Example¶
from horde_client import HordeClientLLM, TextGenParams
from langchain import LLMChain
from langchain.prompts import PromptTemplate
# Prepare HordeClientLLM
params = TextGenParams(
max_context_length=256,
max_length=64,
temperature=0.8
)
llm = HordeClientLLM(
# To access public Horde Client
insecure=True,
# TextGen Parameters for HordeAI Client
params=params
)
# Prompt Template
template = """### Instruction:
Create a fancy company name for a company that makes {product}.
### Response:
"""
prompt= PromptTemplate(input_variables=["product"], template=template)
# Chain Prompt with LLM
chain = LLMChain(
llm = llm,
prompt = prompt
)
print(chain.run("colorful socks"))
# "Socktopia"
print(chain.run("mobiles"))
# "Elysium Technologies Inc."