Marco333/qwen2.5-0.5b-game-commands-stt

TEXT GENERATIONConcurrency Cost:1Model Size:0.5BQuant:BF16Ctx Length:32kPublished:May 6, 2026Architecture:Transformer Cold

Marco333/qwen2.5-0.5b-game-commands-stt is a 0.5 billion parameter Qwen2.5-Instruct model fine-tuned by Marco333, specifically designed for interpreting noisy speech-to-text inputs into predefined game commands. With a context length of 32768 tokens, this model excels at robustly identifying game actions like "Attack" or "OpenDoor" from imperfect voice inputs, making it ideal for integrating voice control in gaming environments. It outputs commands in a structured JSON format, simplifying integration into game logic.

Loading preview...

Overview

This model, Marco333/qwen2.5-0.5b-game-commands-stt, is a specialized fine-tuned version of the Qwen2.5-0.5B-Instruct model. Developed by Marco333, it is specifically optimized for interpreting noisy speech-to-text (STT) inputs from gaming environments and converting them into predefined game commands. It leverages a system prompt that defines a set of valid game commands and expects a JSON output, making it highly suitable for robust voice control integration.

Key Capabilities

  • Robust Command Recognition: Accurately identifies game commands from imperfect, noisy speech-to-text inputs.
  • Structured Output: Responds with a JSON object {"command": "<CommandName>"} or {"command": "Unknown"} for easy parsing.
  • Predefined Command Set: Trained to recognize a specific list of common game actions including Attack, CastSpell, OpenDoor, Jump, Reload, SaveGame, and more.
  • Efficient: Based on a 0.5 billion parameter model, offering a balance of performance and computational efficiency for real-time applications.

Good for

  • Voice Control in Games: Implementing reliable voice commands for in-game actions, even with noisy microphone input.
  • Accessibility Features: Providing alternative input methods for players.
  • Rapid Prototyping: Quickly integrating command recognition into game development workflows due to its focused nature and structured output.
  • Edge Deployment: Its smaller size (0.5B parameters) makes it suitable for deployment in environments with limited resources.

Finetunes
1 models