---
license: other
library_name: llama.cpp
tags:
- gguf
- gpt2
- code
- text-generation
- local-inference
- nsfw
- adult-content
- withinusai
language:
- en
datasets:
- jjmachan/NSFW-reddit
model_type: gguf
inference: false
base_model:
- openai-community/gpt2-medium
---

# GPT-2.5.2-NSFW-Codex-0.4B-GGUF

**GPT-2.5.2-NSFW-Codex-0.4B-GGUF** is a compact GGUF-format language model release from **WithIn Us AI** for local text generation workflows. This repository is marked as containing **sensitive content** and is intended only for appropriate, lawful, age-appropriate, and policy-compliant use.

## Model Summary

This model is a small-footprint local inference release intended for experimentation with:

- adult-content-aware text generation
- niche prompt completion workflows
- compact offline inference
- GGUF-compatible runtimes such as llama.cpp-based stacks

Because this is a **0.4B-class** model, it is best suited to lightweight experimentation and short generations rather than deep reasoning, long-context reliability, or high-accuracy instruction following.

## Safety / Content Notice

This repository is marked **Not-For-All-Audiences** on Hugging Face.

The model may generate or continue prompts involving adult, explicit, or otherwise sensitive themes. Use of this model should be restricted to:

- adults only
- lawful contexts
- consensual and non-exploitative fictional use
- settings where sensitive outputs are acceptable and expected

This model must **not** be used to generate exploitative, abusive, non-consensual, or otherwise harmful content.

## Training Data

The repository page visibly references the following dataset:

- `jjmachan/NSFW-reddit`

If additional datasets, merges, or fine-tuning steps were used, they should be added here explicitly.

## Intended Use

Recommended use cases include:

- local offline generation experiments
- adult-content classification or prompt-behavior testing
- research on small-model behavior in sensitive domains
- GGUF deployment in personal sandboxed environments

## Out-of-Scope Use

This model is not appropriate for:

- minors
- workplace or classroom environments
- production systems without filtering
- harassment, coercion, abuse, or exploitative roleplay
- illegal sexual content
- non-consensual or violent sexual content
- sexualized content involving minors or age ambiguity
- real-person sexual deepfake or image-related misuse
- safety-critical advice of any kind

## Format / Runtime

This model is distributed in **GGUF** format for use with local inference runtimes that support GGUF, such as **llama.cpp** and compatible frontends.

## Performance Expectations

As a compact **0.4B** model, this release is optimized more for portability and low-resource inference than for high reliability. Users should expect possible weaknesses in:

- long-form coherence
- factual accuracy
- instruction retention
- nuanced safety boundaries
- complex reasoning

Human review is required for any meaningful downstream use.

## Prompting Notes

Best results usually come from prompts that are:

- explicit about tone and format
- short to medium length
- clear about constraints
- used in controlled local settings

For safety, downstream applications should add:

- input filtering
- output moderation
- age gating
- logging and review controls where appropriate

## Limitations

Like other small language models, this model may:

- hallucinate
- produce repetitive outputs
- ignore instructions
- drift into undesired content
- generate unsafe or low-quality text without strong prompting and controls

## License

This draft uses:

- `license: other`

Replace this section with the exact **WithIn Us AI** license terms you want for redistribution and use. If this release is derived from upstream models, merged checkpoints, or third-party datasets, include clear attribution and any applicable upstream restrictions.

## Acknowledgments

Thanks to:

- the GGUF / llama.cpp ecosystem
- Hugging Face hosting infrastructure
- the original creators of any upstream models or datasets used in this release

## Disclaimer
This model may generate -explicit, inaccurate, biased, offensive, or otherwise unsuitable content. It should be used only in lawful, age-appropriate, policy-compliant contexts with appropriate safeguards.