ytu-ce-cosmos/Turkish-Gemma-9b-T1

Overview

Turkish-Gemma-9b-T1, developed by ytu-ce-cosmos, is an advanced Turkish language model built upon the Turkish-Gemma-9b-v0.1 base. Its primary focus is enhancing multi-step reasoning capabilities, making it particularly adept at complex problem-solving in Turkish.

Key Capabilities

• Multi-step Reasoning: Significantly improved intermediate inference for tasks involving multiple clues or conditions.
• Mathematical & Logical Accuracy: Enhanced performance in arithmetic, probability, sequences, rational reasoning, and logic puzzles.
• Instruction Following: Demonstrates better adherence to user prompts, leading to more precise and relevant outputs.
• Reduced Hallucinations: The model is designed to produce more grounded answers and indicate uncertainty when appropriate, minimizing factual errors.

Performance Highlights

Evaluated against a custom dataset of 1,450 questions, Turkish-Gemma-9b-T1 achieved a 68.65% win rate, outperforming several larger models including Qwen3-32B and Google's Gemma-3-27b-it. On the Turkish Gsm8k benchmark, it scored 77.41, placing it competitively among high-performing models.

Good for

• Applications requiring complex logical deduction and step-by-step problem-solving in Turkish.
• Educational tools for mathematics and logic in Turkish.
• Scenarios where accurate instruction following and reduced hallucination are critical for Turkish language tasks.
• Developers seeking a robust Turkish LLM optimized for reasoning tasks, especially when Temperature=0.6, TopP=0.95, TopK=20, and MinP=0 are used for generation.