SeongryongJung/Qwen3-8B-Biology-RLSD-TR
SeongryongJung/Qwen3-8B-Biology-RLSD-TR is an 8 billion parameter Qwen3-based language model fine-tuned using the RLSD_TR method specifically for biological knowledge. It achieves a 55.00% mean@16 score on the Biology / SciKnowEval biology dataset, demonstrating specialized performance in this domain. The model is optimized for tasks requiring deep understanding and generation of biological information, leveraging a 32768 token context length.
Loading preview...
Overview
SeongryongJung/Qwen3-8B-Biology-RLSD-TR is an 8 billion parameter language model built upon the Qwen3 architecture. This model has been specifically fine-tuned using the RLSD_TR (Reinforcement Learning with Self-Distillation and Trust-Region) method, targeting the domain of biology. Its training focused on the Biology / SciKnowEval biology dataset, achieving a validation mean@16 score of 55.00% after 100 training steps.
Key Capabilities
- Specialized Biological Knowledge: Fine-tuned on a dedicated biology dataset, making it proficient in biological contexts.
- Performance on SciKnowEval: Achieved a 55.00%
mean@16score on the Biology / SciKnowEval biology validation set. - RLSD_TR Fine-tuning: Utilizes a specific reinforcement learning method for optimization, indicating a focus on generating high-quality, relevant responses within its domain.
- Qwen3 Base: Benefits from the foundational capabilities of the Qwen3-8B model.
Training Details
The model was trained with a batch size of 32 for 100 total steps, using a learning rate of 1e-6. It processed 3200 training samples and utilized a maximum prompt length of 2048 tokens and a maximum response length of 8192 tokens, with a total model length of 10240 tokens.
Usage
This model is suitable for applications requiring a strong understanding and generation of biological text, such as scientific literature analysis, biological question answering, or educational tools in biology.