Unlocking Code Generation with Deepseek Coder 33B Cognitive Actions

25 Apr 2025
Unlocking Code Generation with Deepseek Coder 33B Cognitive Actions

In the world of software development, efficiency is key. The Deepseek Coder 33B Instruct GGUF Cognitive Actions empower developers to leverage advanced AI capabilities for code generation, specifically designed for repository-level code completion tasks. By utilizing a quantized 33B parameter language model, these actions provide a seamless and intelligent solution for generating high-quality code snippets, ultimately saving time and enhancing productivity.

Prerequisites

Before you dive into integrating the Deepseek Coder 33B Cognitive Actions, ensure you meet the following prerequisites:

  • An API key for the Cognitive Actions platform. This key is essential for authenticating your requests.
  • Familiarity with JSON payload structure and Python programming to facilitate integration.

Authentication typically involves passing your API key in the request headers, allowing you to securely access the cognitive actions.

Cognitive Actions Overview

Generate Code Completion with Deepseek Coder 33B

Description:
This action utilizes a state-of-the-art 33B parameter language model from Deepseek for advanced code completion tasks. It is designed to understand context effectively with a 16k context window, enabling accurate and relevant code generation.

Category: Code Generation

Input

The input schema for this action is structured as follows:

  • prompt (required): The main instruction or query for the AI model.
  • temperature (optional): Controls randomness in output (default: 0.8).
  • maxNewTokens (optional): Maximum tokens to generate (-1 indicates no limit).
  • systemPrompt (optional): A guiding message for the AI's behavior.
  • repeatPenalty (optional): Discourages repetitive phrases (default: 1.1).
  • promptTemplate (optional): Defines the structure of the interaction.

Example Input:

{
  "prompt": "please create a rust enum called prediction status, with three variants starting, in progress and completed. Please only include valid rust code, do not include any commentary or explanations.",
  "temperature": 0.8,
  "maxNewTokens": -1,
  "systemPrompt": "You are an AI programming assistant, utilizing the Deepseek Code model, developed by Deepseek Company, and you only answer questions related to computer science. For politically sensitive questions, security and privacy issues, and other non-computer science questions, you will refuse to answer.",
  "repeatPenalty": 1.1,
  "promptTemplate": "{system_prompt}/n### Instruction: {prompt}/n### Response: "
}

Output

The output of this action typically consists of the generated code based on the provided prompt. Here's an example of what you might receive:

[
  "\n",
  "```",
  "rust",
  "\n",
  "#",
  "[",
  "der",
  "ive",
  "(",
  "Debug",
  ")]",
  "\n",
  "enum",
  " Pred",
  "iction",
  "Status",
  " {",
  "\n",
  "   ",
  " Starting",
  ",",
  "\n",
  "   ",
  " In",
  "Progress",
  ",",
  "\n",
  "   ",
  " Comple",
  "ted",
  ",",
  "\n",
  "}",
  "\n",
  "```",
  "\n",
  ""
]

Conceptual Usage Example (Python)

Below is a conceptual Python snippet that demonstrates how to call the Deepseek Coder 33B action:

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 = "717cd8d0-f6cf-4c78-b695-ef8f082fc30d"  # Action ID for Generate Code Completion with Deepseek Coder 33B

# Construct the input payload based on the action's requirements
payload = {
    "prompt": "please create a rust enum called prediction status, with three variants starting, in progress and completed. Please only include valid rust code, do not include any commentary or explanations.",
    "temperature": 0.8,
    "maxNewTokens": -1,
    "systemPrompt": "You are an AI programming assistant, utilizing the Deepseek Code model, developed by Deepseek Company, and you only answer questions related to computer science. For politically sensitive questions, security and privacy issues, and other non-computer science questions, you will refuse to answer.",
    "repeatPenalty": 1.1,
    "promptTemplate": "{system_prompt}/n### Instruction: {prompt}/n### Response: "
}

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 snippet, make sure to replace the placeholder API key and endpoint with your actual values. The action_id variable is set to the identifier for the code completion action, and the payload is structured according to the input schema provided.

Conclusion

The Deepseek Coder 33B Cognitive Actions provide an excellent opportunity for developers to automate code generation tasks effectively. By integrating this powerful tool into your applications, you can enhance coding efficiency and reduce manual coding time. Next steps could include experimenting with different prompts and configuration settings to see how they affect the output, or combining this action with others for more complex workflows. Happy coding!

← Back to All Articles