Anthropic tutorial: how to create a 100 000 token context AI chatbot

Is Claude good for chatbots?

Of course it is! Anthropic’s Claude has been designed to be a chatbot. First thing is the fact that as it was being built, safety was the priority which resulted in good reviews from users! The Second one is a fact, that its big, 100 000 tokens, context window helps with long conversations and its ability to generate long texts is perfect for chatbots. It will also allow us to provide additional context to model. Thanks to that it will be able to extract more information and be more accurate in its responses. Let's check how to build a simple chatbot with Claude! And with this knowledge you can build many more Anthropic application!

I described getting early access to Anthropic Claude in a previous tutorial.

Coding

Dependencies

We can proceed to create the project! Let's start by creating the main folder.

mkdir chatbot-claude

cd chatbot-claude

Now we can create a virtual environment and install the necessary libraries.

python3 -m venv venv

# Linux/Mac OS
source venv/bin/activate

# Windows
venv\Scripts\activate.bat

pip install anthropic python-dotenv

The last step will be to create a main.py file and import the libraries.

import os

import anthropic
from dotenv import load_dotenv

load_dotenv()

Chatbot

We will follow the template of another project that we once created. I would like, as there, to display the cost of generating a response after each message. This will be important, because I don’t want to overshoot the amount of money used to generate the response.

First I will initialize the Anthropic client, the context, and prepare constants with prices for generating one token (in USD).

anthropic_client = anthropic.Client(api_key=os.getenv("ANTHROPIC_API_KEY"))

# you can put some context here, to specialize the Model
context = ""

# Anthropic Claude pricing: https://cdn2.assets-servd.host/anthropic-website/production/images/model_pricing_may2023.pdf
PRICE_PROMPT = 1.102e-5
PRICE_COMPLETION = 3.268e-5

Let the next step be preparation of a function to count the number of tokens in the prompt and response. Then the return text with information about the means used.

def count_used_tokens(prompt, completion):
    prompt_token_count = anthropic.count_tokens(prompt)
    completion_token_count = anthropic.count_tokens(completion)

    prompt_cost = prompt_token_count * PRICE_PROMPT
    completion_cost = completion_token_count * PRICE_COMPLETION

    total_cost = prompt_cost + completion_cost

    return (
        "🟡 Used tokens this round: "
        + f"Prompt: {prompt_token_count} tokens, "
        + f"Completion: {completion_token_count} tokens - "
        + f"{format(total_cost, '.5f')} USD)"
    )

The final step is to run a chat loop, retrieve text from the user, collect context and generate a response.

while True:
    user_inp = input("You: ")

    current_inp = anthropic.HUMAN_PROMPT + user_inp + anthropic.AI_PROMPT
    context += current_inp

    prompt = context

    completion = anthropic_client.completion(
        prompt=prompt, model="claude-v1.3-100k", max_tokens_to_sample=1000
    )["completion"]

    context += completion

    print("Anthropic: " + completion)
    print(count_used_tokens(prompt, completion))

Let’s test it!

There you go, our app is ready. Let's test it out!

python main.py

Quick and easy to set up, right?

As you can see, our chatbot responds correctly! It's worth testing it with more examples, and most importantly, check the context! Here the field is quite large, as it amounts to 100k tokens!

I encourage you to follow our Artificial Intelligence Hackathon, especially the upcoming Anthropic Hackathon!

Why shouldn’t you build your own customized chatbot or any other Anthropic apps? And where better to do so, then with other like-minded people in a short period of time and with assistance of mentors?

Thank you for your time! - Jakub Misiło @newnative