DEV Community

Cover image for Use any LLM with Just 8 Lines of Code πŸš€
Durvesh Danve
Durvesh Danve

Posted on β€’ Edited on

16 2 2 3 3

Use any LLM with Just 8 Lines of Code πŸš€

Ever wondered how easy it could be to harness the power of cutting-edge AI models in your projects?

With just 8 lines of Python code, you can start using a powerful Large Language Model (LLM) without diving into the complexities of training one from scratch.

Let’s see how!

Tools we'll be using:



1. Huggingface pretrained model (in this case, falcon)
2. Python
3. Langchain
4. Google Colab


Enter fullscreen mode Exit fullscreen mode

First, open Google Colab and create a new notebook.

Let's start coding:

Step 1:
Install the necessary libraries:



!!pip install langchain huggingface_hub langchain_community


Enter fullscreen mode Exit fullscreen mode

Step 2:
Set up your Hugging Face API token as an environment variable:



import os
os.environ["HUGGINGFACEHUB_API_TOKEN"] = "YOUR_TOKEN"


Enter fullscreen mode Exit fullscreen mode

To get your token:

  1. Visit Hugging Face and sign in or create an account.
  2. Navigate to the settings page and select the Access Token tab.
  3. Create a token and replace "YOUR_TOKEN" with your actual token.

Huggingface Access Token section

Step 3:
Import HuggingFaceHub from langchain :



from langchain import HuggingFaceHub


Enter fullscreen mode Exit fullscreen mode

Initialize your Large Language Model (LLM):



llm = HuggingFaceHub(repo_id="tiiuae/falcon-7b-instruct", model_kwargs={"temperature":0.6})


Enter fullscreen mode Exit fullscreen mode

I’m using the tiiuae/falcon-7b-instruct model here, but there are plenty of models available. You can explore them here.

Let’s test the model:



prompt = 'Generate a Python function to print the Fibonacci series. Ensure the code is optimized for efficiency and has minimal time complexity'
response = llm(prompt)
print(response)


Enter fullscreen mode Exit fullscreen mode

and this results into :



def fibonacci(n):
    if n == 0:
        return 0
    elif n == 1:
        return 1
    else:
        return fibonacci(n - 1)+fibonacci(n - 2)


Enter fullscreen mode Exit fullscreen mode

And just like that, with only 8 lines of code, we’ve set up our own version of ChatGPT! πŸŽ‰πŸ’»

Complete Code



# Install necessary libraries
!pip install langchain huggingface_hub langchain_community

import os
os.environ["HUGGINGFACEHUB_API_TOKEN"] = "YOUR_TOKEN"

from langchain import HuggingFaceHub

# Initialize the model
llm = HuggingFaceHub(repo_id="tiiuae/falcon-7b-instruct", model_kwargs={"temperature":0.6})

# Use the model to generate a response
prompt = 'Generate a Python function to print the Fibonacci series. Ensure the code is optimized for efficiency and has minimal time complexity'
response = llm(prompt)
print(response)


Enter fullscreen mode Exit fullscreen mode

Top comments (9)

Collapse
 
syeo66 profile image
Red Ochsenbein (he/him) β€’

"Use a LLM..." would be more appropriate. You're not building a model

Collapse
 
durvesh_danve profile image
Durvesh Danve β€’

Yes, you're right. Thanks for the suggestion!

Collapse
 
alanarmrfme profile image
Alana E β€’

Building?

You're using an API??

Misleading or just a mistake of wording???

Collapse
 
durvesh_danve profile image
Durvesh Danve β€’

Thanks for sharing your thoughts, The intention behind the title was to highlight how easy it can be to use LLM with minimal code.

Collapse
 
anna_lapushner profile image
anna lapushner β€’

I love that you are coding! I love that you are publishing your process! Thank you for inviting me into the world of this powerful sequence, the awesome Fibonacci...

Collapse
 
durvesh_danve profile image
Durvesh Danve β€’

Hehe Thanks!

Collapse
 
anna_lapushner profile image
anna lapushner β€’

Take care my friend!

Collapse
 
tanmay_borde_f9ab7ebbaa14 profile image
Tanmay Borde β€’

Awesome!

Collapse
 
durvesh_danve profile image
Durvesh Danve β€’

Thanks!