Skip to content

Add CAJAL — Offline Scientific Paper Generator (4B-9B GGUF, No API Keys) #5

Description

@Agnuxo1

Proposal: Add CAJAL

Repository: https://github.com/Agnuxo1/CAJAL
HuggingFace: https://huggingface.co/Agnuxo/CAJAL-9B-P2PCLAW
Paper: https://arxiv.org/pdf/2604.19792

What it does

CAJAL generates publication-ready scientific papers — completely offline:

  • Runs on 4B-9B GGUF models via Ollama/llama.cpp (tested on RTX 3090, works on CPU too)
  • Zero API calls — no OpenAI, no Anthropic, no internet required after model download
  • Full IMRaD structure (Introduction, Methods, Results, Discussion) with real arXiv citations
  • AI Tribunal — 3 reviewers critique the paper; weak sections get rewritten automatically
  • LaTeX export ready for journal submission

Why it fits here

Most offline AI tools focus on chat, coding, or general productivity. CAJAL is the first tool that brings scientific writing to the offline world — critical for researchers in:

  • Air-gapped environments (government labs, classified facilities)
  • Developing regions with unreliable internet
  • Anyone who wants verifiable, reproducible paper generation without cloud dependency

Key stats

  • 4B (mobile-friendly) to 9B (desktop-quality) parameter options
  • Apache 2.0 licensed
  • Part of P2PCLAW decentralized research network

License

Apache 2.0

Suggested placement

Under "Notable Open-Source Projects" or a new "Scientific / Academic Tools" subsection.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions