Polygl0t/Tucano2-qwen-1.5B-Think
Polygl0t/Tucano2-qwen-1.5B-Think is a 1.51 billion parameter instruction-tuned Portuguese language model based on the Qwen3 Transformer architecture, developed by Polygl0t. It is specifically fine-tuned for reasoning tasks, generating Chain-of-Thought (CoT) style traces encapsulated within and tokens. This model excels at Portuguese language understanding and complex problem-solving, making it suitable for research and development in Portuguese NLP.
Loading preview...
Model Overview
Tucano2-qwen-1.5B-Think is a 1.51 billion parameter instruction-tuned Portuguese language model developed by Polygl0t, built upon the Qwen3 Transformer architecture. It is specifically designed for reasoning tasks, generating Chain-of-Thought (CoT) style traces within <think> and </think> special tokens. The model was trained using one round of supervised fine-tuning (SFT) and one round of Anchored Preference Optimization (APO) on Portuguese datasets like Polygl0t/gigaverbo-v2-sft and Polygl0t/gigaverbo-v2-preferences, with a context length of 4,096 tokens.
Key Capabilities
- Portuguese Reasoning: Optimized to produce detailed, step-by-step reasoning processes for complex queries in Portuguese.
- Instruction Following: Demonstrates capabilities in following instructions, as evaluated by IFEval-PT.
- Open and Reproducible: All datasets, source code, and training recipes for the Tucano2 series are fully open and reproducible.
Good For
- Research and Development: Serves as a foundation for research and development in Portuguese language modeling.
- Fine-tuning: Can be fine-tuned and adapted for specific deployment scenarios, provided users conduct their own risk and bias assessments.
- Educational Applications: Useful for tasks requiring explicit reasoning steps, such as explaining mathematical problems or historical events in Portuguese.
Limitations
Like many LLMs, Tucano2-qwen-1.5B-Think is subject to hallucinations, biases, and toxicity inherited from its training data. It is primarily designed for Portuguese, and other languages may challenge its comprehension. The model may also exhibit repetition or verbosity.