qgallouedec/gemma-3-12b-it-codeforces-SFT
The qgallouedec/gemma-3-12b-it-codeforces-SFT model is a 12 billion parameter instruction-tuned variant of Google's Gemma-3-12b-it architecture. It has been specifically fine-tuned on the open-r1/codeforces-cots dataset using the TRL framework, optimizing its performance for competitive programming and code-related problem-solving tasks. This model is designed to excel at understanding and generating solutions for algorithmic challenges, leveraging its 32768 token context length.
Loading preview...
Model Overview
The qgallouedec/gemma-3-12b-it-codeforces-SFT is a specialized large language model built upon the Google Gemma-3-12b-it architecture. This 12 billion parameter model has undergone Supervised Fine-Tuning (SFT) using the TRL library, specifically leveraging the open-r1/codeforces-cots dataset.
Key Capabilities
- Code-centric Problem Solving: Optimized for tasks related to competitive programming and algorithmic challenges, drawing from its training on Codeforces data.
- Instruction Following: Benefits from its base as an instruction-tuned model, enabling it to understand and execute complex prompts.
- Contextual Understanding: Features a 32768 token context length, allowing it to process and generate longer, more intricate code snippets or problem descriptions.
Training Details
The model was fine-tuned using the SFT method, with specific framework versions including TRL 0.16.0.dev0, Transformers 4.50.0.dev0, and PyTorch 2.6.0. This targeted training approach aims to enhance its proficiency in generating relevant and accurate responses for coding-related queries.