fluently/FluentlyQwen3-1.7B
FluentlyQwen3-1.7B is a 1.7 billion parameter causal language model developed by Fluently, built upon the Qwen3 architecture. This model features a 32,768 token context length and is enhanced through Supervised Fine-Tuning (SFT) and GRPO training, focusing on diverse datasets. It aims to improve general capabilities across tasks like communication, translation, mathematics, coding, and agent functions. A key differentiator is its integrated 'thinking' capability, allowing for enhanced reasoning in responses.
Loading preview...
FluentlyQwen3-1.7B Overview
FluentlyQwen3-1.7B is a 1.7 billion parameter causal language model developed by Fluently, leveraging the Qwen3 architecture. It boasts a substantial context length of 32,768 tokens. The model's development involved Supervised Fine-Tuning (SFT) and GRPO training, utilizing diverse datasets to enhance its foundational capabilities.
Key Capabilities and Enhancements
This model demonstrates improved performance across a range of general tasks, including:
- Basic Communication: Enhanced conversational abilities.
- Translation: Better language translation accuracy.
- Mathematics & Sciences: Improved understanding and problem-solving in mathematics, physics, biology, and medicine.
- Coding: Stronger code generation and comprehension.
- Agent Functions: Enhanced capabilities for agent-based applications.
Unique Feature: Thinking Mode
A notable feature of FluentlyQwen3-1.7B is its integrated 'thinking' capability, enabled by default. This allows the model to engage in a reasoning process, generating internal 'think' content before producing its final response, similar to QwQ-32B. This mode is designed to improve the quality of generated outputs. Users can also explicitly disable this thinking mode for scenarios prioritizing efficiency over complex reasoning, aligning its behavior with previous Qwen2.5-Instruct models. Recommended generation parameters are provided for both thinking and non-thinking modes to optimize performance.