Hyeongwon/P2-split1_only_answer_Qwen3-4B-Base_0502-bs64-epoch6-lr5e6
Hyeongwon/P2-split1_only_answer_Qwen3-4B-Base_0502-bs64-epoch6-lr5e6 is a 4 billion parameter language model developed by Hyeongwon, fine-tuned from the Qwen3-4B-Base architecture. This model is specifically trained using Supervised Fine-Tuning (SFT) with the TRL framework, focusing on generating direct answers. It is designed for tasks requiring concise and relevant responses, leveraging its 32768 token context length for processing detailed prompts.
Loading preview...
Model Overview
This model, developed by Hyeongwon, is a 4 billion parameter language model fine-tuned from the Qwen3-4B-Base architecture. It has been specifically trained using Supervised Fine-Tuning (SFT) with the TRL framework, indicating an optimization for generating direct and relevant answers to given prompts.
Key Capabilities
- Answer Generation: Optimized for producing concise and focused answers.
- Fine-tuned Performance: Leverages SFT to enhance its ability to respond accurately to questions.
- Context Handling: Benefits from a 32768 token context length, allowing it to process and understand longer, more complex queries before generating a response.
Training Details
The model was trained using the TRL (Transformer Reinforcement Learning) framework, specifically employing SFT. The training utilized various framework versions including TRL 0.25.1, Transformers 4.57.3, Pytorch 2.9.1, Datasets 3.6.0, and Tokenizers 0.22.2. This fine-tuning process aims to specialize the model for direct answer generation tasks.