AtomicFission-7B-v1-GGUF

AtomicFission-7B-v1-GGUF

GGUF quantized versions of AtomicFission-7B-v1 for efficient CPU/GPU inference.

☢️ Model Description

These are GGUF format quantizations of AtomicFission-7B-v1, a Mistral-7B model fine-tuned on publicly available nuclear documentation. The model specializes in nuclear engineering topics including reactor physics, safety systems, and operational procedures.

Available Quantizations

Quick Start

Using llama.cpp

# Download a model
wget https://huggingface.co/dougeeai/AtomicFission-7B-v1-GGUF/resolve/main/AtomicFission-7B-v1-Q5_K_M.gguf

# Run inference
./main -m AtomicFission-7B-v1-Q5_K_M.gguf \
  -p "Question: What is reactor criticality?\nAnswer:" \
  -n 200 --gpu-layers 35

Using LM Studio

1. Download LM Studio
2. Search for "AtomicFission"
3. Download preferred quantization
4. Load and chat

Using Ollama

Create a Modelfile:

FROM ./AtomicFission-7B-v1-Q5_K_M.gguf
TEMPLATE """Question: {{ .Prompt }}
Answer: """

Then: ollama create atomicfission -f Modelfile

Model Characteristics

Strengths:

• Nuclear reactor technical specifications
• Safety system descriptions
• Reactor physics concepts
• SMR and advanced reactor designs
• Operational procedures

Optimal Prompt Format:

Question: [Your nuclear engineering question]
Answer:

Performance Notes

• Q5_K_M: Recommended - excellent quality with reasonable size
• Q4_K_M: Solid - good for consumer GPUs
• Q6_K/Q8_0: For maximum quality when RAM isn't constrained
• Q3_K_M/Q2_K: For mobile or embedded deployment

Inference speed scales inversely with model size. Q4_K variants offer the best speed/quality trade-off for most users.

⚠️☢️⚠️ Important Disclaimer

This model is for educational and research purposes only. It should NOT be used for:

• Operational nuclear facility decisions
• Safety-critical applications
• Regulatory compliance
• Professional engineering work
• Any production use cases

The model may hallucinate technical details. Always verify information with authoritative sources and qualified nuclear professionals.

This is an independent personal project not affiliated with any government agency, company, or employer.

License

Apache-2.0 (inherited from Mistral-7B base model)

Citation

If you use this model in research, please cite:

@misc{atomicfission2024,
  title={AtomicFission-7B-v1: A Nuclear Reactor Operations Language Model},
  author={dougeeai},
  year={2024},
  publisher={HuggingFace}
}

Base Model

Full model card and transformers format: dougeeai/AtomicFission-7B-v1