PrimeIntellect/Qwen3-1.7B-Wordle-RL
TEXT GENERATIONConcurrency Cost:1Model Size:2BQuant:BF16Ctx Length:32kPublished:Sep 24, 2025License:apache-2.0Architecture:Transformer0.0K Open Weights Warm
PrimeIntellect/Qwen3-1.7B-Wordle-RL is a 2 billion parameter model, fine-tuned using Reinforcement Learning (RL) specifically for the game Wordle. This model is based on the Qwen3 architecture and is an RL fine-tune of PrimeIntellect/Qwen3-1.7B-Wordle-SFT, designed to excel at Wordle gameplay. It leverages a 40960 token context length to process game states and make optimal Wordle guesses.
Loading preview...
Model Overview
PrimeIntellect/Qwen3-1.7B-Wordle-RL is a specialized language model with 2 billion parameters, developed by PrimeIntellect. It is an advanced iteration of the Qwen3-1.7B-Wordle-SFT model, having undergone further fine-tuning through Reinforcement Learning (RL).
Key Capabilities
- Wordle Optimization: This model is explicitly trained and fine-tuned to play the game Wordle, aiming for optimal performance in guessing the secret word.
- Reinforcement Learning: Utilizes RL techniques to improve its Wordle strategy and decision-making over time.
- Qwen3 Architecture: Built upon the Qwen3 model family, providing a robust foundation for its language understanding and generation capabilities.
- Extended Context Window: Features a substantial context length of 40960 tokens, allowing it to process complex game states and historical information during Wordle gameplay.
Good For
- Wordle Game Agents: Ideal for developing AI agents or bots specifically designed to play and solve Wordle puzzles efficiently.
- RL Research: Can serve as a practical example or baseline for research into applying Reinforcement Learning to language-based games and tasks.
Further details on the RL fine-tuning process are available here.