Harnessing the Power of Llama 2: A Guide to Text Generation with Meta's Cognitive Actions

23 Apr 2025
Harnessing the Power of Llama 2: A Guide to Text Generation with Meta's Cognitive Actions

The meta/llama-2-70b API provides developers with access to the Llama 2 language model, a powerful 70 billion parameter model designed for a broad range of text generation tasks. By integrating Cognitive Actions offered by this API, you can leverage advanced text generation capabilities in your applications, enhancing user experiences through creative and context-aware outputs. This article will guide you through the primary action available in this spec: generating text with Llama 2.

Prerequisites

Before diving into the integration of Cognitive Actions, ensure that you have:

  • An API key for accessing the Cognitive Actions platform.
  • Familiarity with JSON data structures for constructing requests and handling responses.

Authentication typically involves passing your API key in the HTTP headers of your requests, allowing your application to securely communicate with the Cognitive Actions service.

Cognitive Actions Overview

Generate Text with Llama 2

Purpose:
The "Generate Text with Llama 2" action utilizes the base version of the Llama 2 model to generate or predict text based on a given prompt. This model is versatile and allows for control over output randomness and variability through various parameters.

Category:
Text Generation

Input

The input for this action requires a JSON object structured according to the following schema:

{
  "prompt": "string (required)",
  "seed": "integer (optional)",
  "topK": "integer (optional, default: 50)",
  "topP": "number (optional, default: 0.9)",
  "debug": "boolean (optional, default: false)",
  "temperature": "number (optional, default: 0.75)",
  "modelWeightsPath": "string (optional)",
  "terminationSequences": "string (optional)",
  "maximumGeneratedTokens": "integer (optional, default: 128)",
  "minimumGeneratedTokens": "integer (optional, default: -1)"
}

Example Input:

Here's an example of a JSON payload you might send to the action:

{
  "topP": 1,
  "prompt": "original prompt: garden with flowers and dna strands\nimproved prompt: psychedelic 3d vector art illustration of garden full of colorful double helix dna strands and exotic flowers by lisa frank, beeple and tim hildebrandt, hyper realism, art deco, intricate, elegant, highly detailed, unreal engine, octane render, smooth\n\noriginal prompt: humanoid plant monster\nimproved prompt: ",
  "temperature": 0.75
}

Output

The output of the action will typically return a string containing the generated text based on the input prompt.

Example Output:

3d vector art illustration of a humanoid plant monster in a psychedelic garden, by lisa frank, beeple and tim hildebrandt, hyper realism, art deco, intricate, elegant, highly detailed, unreal engine, octane render, smooth...

Conceptual Usage Example (Python)

Below is a conceptual Python code snippet demonstrating how you might call the "Generate Text with Llama 2" action using a hypothetical Cognitive Actions API endpoint:

import requests
import json

# Replace with your Cognitive Actions API key and endpoint
COGNITIVE_ACTIONS_API_KEY = "YOUR_COGNITIVE_ACTIONS_API_KEY"
COGNITIVE_ACTIONS_EXECUTE_URL = "https://api.cognitiveactions.com/actions/execute"  # Hypothetical endpoint

action_id = "2484e6e7-184b-41a3-a55c-8a9dd0c1e4eb"  # Action ID for Generate Text with Llama 2

# Construct the input payload based on the action's requirements
payload = {
    "topP": 1,
    "prompt": "original prompt: garden with flowers and dna strands\nimproved prompt: psychedelic 3d vector art illustration of garden full of colorful double helix dna strands and exotic flowers by lisa frank, beeple and tim hildebrandt, hyper realism, art deco, intricate, elegant, highly detailed, unreal engine, octane render, smooth\n\noriginal prompt: humanoid plant monster\nimproved prompt: ",
    "temperature": 0.75
}

headers = {
    "Authorization": f"Bearer {COGNITIVE_ACTIONS_API_KEY}",
    "Content-Type": "application/json"
}

try:
    response = requests.post(
        COGNITIVE_ACTIONS_EXECUTE_URL,
        headers=headers,
        json={"action_id": action_id, "inputs": payload}  # Hypothetical structure
    )
    response.raise_for_status()  # Raise an exception for bad status codes (4xx or 5xx)

    result = response.json()
    print("Action executed successfully:")
    print(json.dumps(result, indent=2))

except requests.exceptions.RequestException as e:
    print(f"Error executing action {action_id}: {e}")
    if e.response is not None:
        print(f"Response status: {e.response.status_code}")
        try:
            print(f"Response body: {e.response.json()}")
        except json.JSONDecodeError:
            print(f"Response body: {e.response.text}")

In this code snippet:

  • Replace YOUR_COGNITIVE_ACTIONS_API_KEY with your actual API key.
  • The action_id corresponds to the specific action being called.
  • The payload variable constructs the JSON input based on the action's requirements.

Conclusion

Integrating the "Generate Text with Llama 2" action into your applications provides a powerful way to enhance text generation capabilities. With a variety of input parameters, you can customize the output to fit your needs, making it suitable for creative writing, content generation, and much more. Explore additional use cases to fully leverage the potential of Llama 2 in your projects!

← Back to All Articles