SentinelAI

SentinelAI is an experimental, transformer-based language model project for cybersecurity-oriented text and security-log analysis.

Model repository: Uchiha80971/SentinelAI

Status: Experimental research prototype. SentinelAI has not been validated as a production intrusion-detection system. Do not use its outputs as the sole basis for security decisions.

Overview

SentinelAI is intended for experimentation with language-model-assisted cybersecurity analysis. The repository includes model weights, configuration, tokenizer data, and custom model implementation files.

Potential research uses include:

• Exploring how a language model responds to security-log descriptions.
• Producing draft explanations of supplied security events.
• Experimenting with model-based security classification and decision-fusion pipelines.

These are intended use cases, not claims of validated performance.

Repository contents

• config.json — model configuration.
• pytorch_model.bin — exported model weights.
• sentinel_tokenizer.json — tokenizer definition.
• model/ — custom model implementation files.

The model uses a custom architecture implementation, so loading the weights may require the matching Python source code. Uploading the files to Hugging Face does not by itself guarantee compatibility with the hosted Inference API or an online chat widget.

Architecture

The current configuration describes a decoder-style, transformer-based language model with approximately 90–96 million parameters, a hidden size of 768, 12 transformer layers, 12 attention heads, a feed-forward dimension of 3072, and a context length of 768 tokens.

Configuration and implementation files in this repository should be treated as the source of truth for the exact architecture and compatibility requirements.

Intended use

SentinelAI is intended for:

• Research and educational experimentation.
• Local testing of language-model-based security analysis.
• Prototyping workflows that combine model-generated explanations with deterministic security rules and human review.

Limitations and safety

SentinelAI may generate inaccurate, irrelevant, or fabricated details, including users, baselines, event sequences, and explanations that are not present in the input. It may also misclassify security events or fail to recognize suspicious activity.

In preliminary manual tests, the model outputs were sometimes inconsistent, and a port-scan example was not reliably identified by the model-based analysis. These observations are not a formal benchmark and should not be interpreted as measured accuracy.

Do not use SentinelAI as a standalone security control. For any practical evaluation:

• Validate predictions against a labeled test set.
• Report precision, recall, F1 score, and false-negative rate.
• Use deterministic rules and established detection tools for critical indicators.
• Keep a human analyst in the decision loop.
• Avoid submitting secrets, personal data, or sensitive production logs to public demos.

Training details

Training dataset composition, licensing, exact training compute, and a reproducible training recipe are not documented in this model card. Add those details when they are available; do not infer them from the checkpoint alone.

Local use

This repository currently provides the model artifacts and custom implementation. It does not yet promise a one-line transformers.pipeline() workflow.

To run the model locally, use a compatible Python/PyTorch environment, the matching files in model/, config.json, and sentinel_tokenizer.json, and the project's inference code. Exact commands depend on the matching implementation and checkpoint-loading code.

License

No license is declared in this model card. Before enabling reuse or redistribution, add a license that you have the rights to use and that matches the training data and code terms.

Acknowledgements

Created as part of the SentinelAI / SentientAI cybersecurity AI project.

Citation

If you reference this project, cite the Hugging Face repository:

Uchiha80971. SentinelAI. Hugging Face Model Hub.
https://huggingface.co/Uchiha80971/SentinelAI