Xiao-Youth/LECTOR-4B
LECTOR-4B is a 4 billion parameter causal language model developed by Xiao-Youth, based on the Qwen3-4B-Instruct architecture. It is specifically fine-tuned for Content-Conditional Introduction Generation (CCIG) in scientific papers, extracting reasoning graphs and generating logic-aware introductions. This model excels at tasks involving scientific reasoning graph extraction and guided introduction generation, leveraging its 32768 token context length.
Loading preview...
LECTOR-4B: Scientific Introduction Generation Model
LECTOR-4B is a 4 billion parameter model developed by Xiao-Youth, specifically designed for Content-Conditional Introduction Generation (CCIG) in scientific contexts. It is built upon the Qwen/Qwen3-4B-Instruct base model and is part of the larger LECTOR project, which focuses on the joint optimization of scientific reasoning graphs and introduction generation.
Key Capabilities
- Scientific Reasoning Graph Extraction: Processes the main body of a scientific paper to extract its underlying reasoning logic.
- Logic-Aware Introduction Generation: Generates scientific introductions guided by the extracted reasoning graph and relevant citation context.
- Content-Conditional Generation: Tailors introductions based on the specific content of the paper's main body.
Intended Use Cases
This model is primarily intended for research applications in:
- Scientific reasoning graph extraction.
- Logic-aware scientific writing assistance.
- Automated content-conditional introduction generation for academic papers.
Limitations
Users should be aware that LECTOR-4B may produce incorrect reasoning graphs, generate unsupported claims, or include citation errors. All generated text requires manual verification against the original source paper content and references to ensure accuracy and reliability. For full details on prompting, rollout, and evaluation, refer to the official project repository.