Xtra-Computing/XtraGPT-7B

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:7.6BQuant:FP8Ctx Length:32kPublished:Feb 4, 2025License:mg0-2.0Architecture:Transformer0.0K Warm

XtraGPT-7B, developed by Xtra-Computing, is a 7 billion parameter language model based on Qwen/Qwen2.5-7B-Instruct, specifically fine-tuned for human-AI collaborative academic paper revision. It excels at understanding full paper context and executing criteria-guided revision instructions, trained on 140,000 instruction-revision pairs from top-tier conference papers. Its primary use case is to provide context-aware and controllable revisions for academic writing, supporting an iterative human-AI workflow.

Loading preview...

XtraGPT: Context-Aware Academic Paper Revision

XtraGPT is a family of open-source Large Language Models (LLMs) from Xtra-Computing, specifically engineered for human-AI collaborative academic paper revision. Unlike general-purpose models, XtraGPT is fine-tuned to deeply understand the full context of a research paper and perform precise, criteria-guided revisions.

This model (XtraGPT-7B) is built upon Qwen/Qwen2.5-7B-Instruct and was trained on a specialized dataset of 140,000 high-quality instruction-revision pairs derived from top-tier ICLR conference papers.

Key Capabilities

  • Context-Aware: Processes the entire paper to ensure revisions maintain consistency with the global narrative and context.
  • Controllable: Follows specific user instructions, adhering to 20 academic writing criteria across 6 distinct paper sections (e.g., Abstract, Introduction).
  • Iterative Workflow: Designed to integrate seamlessly into a "Human-AI Collaborative" (HAC) lifecycle, allowing authors to retain creative control while leveraging AI assistance.

Good for

  • Academic researchers and authors seeking AI assistance for paper revision.
  • Improving conciseness, clarity, and adherence to academic standards in research papers.
  • Generating revisions that are consistent with the overall paper content and specific user instructions.
  • Integrating into tools that support human-in-the-loop editing workflows for scientific writing.