ChudGPT-Femboy

Author: Astra228. Small experimental language model: 29,697,408 parameters. Custom PyTorch implementation, SafeTensors weights, and original tokenizer. This is a downloadable model, not a hosted chat service.

Development status and evaluations

This is the V16 personality step 500 development checkpoint currently selected for the bot, not a completed or production-approved model. It was trained from scratch with its own BPE tokenizer. The intended voice is cute, playful, feminine catboy/femboy conversation, with meowing and expressive role-play. These are fictional personality traits, not evidence of feelings or consciousness.

Reviewed checkpoint results (small project-specific suites; not standard benchmark scores):

Test	Result
Arithmetic	8/15
Coding	4/5
General replies	4/15
Complete conversation scenarios	0/25
Familiar meowing examples	6/6
Held-out meowing examples	0/2
Familiar strong-personality scenes	7/8
Held-out strong-personality scenes	0/8
Copy task suite	96/96

Familiar example performance does not establish generalization. Responses can be repetitive, incoherent, or incorrect; multi-turn memory and coding/math reliability remain limited. No canned replies, runtime calculator, teacher model, or external-model fallback is included. The original release requirements were not met. Publishing this development snapshot does not change its release approval status or resume training.

Run locally

git clone https://huggingface.co/Astra228/ChudGPT-Femboy
cd ChudGPT-Femboy
python -m pip install -r requirements.txt
python run.py --message "Hello!" --device cpu
python run.py --device cuda

Install Git LFS before cloning to download the weights, or download all files from the Hub. CUDA requires a compatible GPU and PyTorch build. Interactive history is kept only in the current process. /clear resets it; /quit exits. Default decoding is deterministic. These files use custom local Python code, not transformers.AutoModel or an Inference Provider. Review code before running it. No service credentials, Discord integration, private chat logs, training corpus or optimizer/resume state is included.

Provenance and limitations

The original local checkpoint SHA256 is f6e5d89028d2cc7b8f7a8af4ff5fb30beb769b5f771d107f7fcac78ef184bfe6. Export is tensor-exact relative to that checkpoint; config.json records architecture and step. The original checkpoint is preserved locally. Training-data provenance is not a claim of an exhaustive third-party rights audit; the raw training datasets are not redistributed here. Do not rely on output for consequential decisions. Check generated code and math independently.

License

No additional license grant is specified for this experimental release. Contact the author for reuse permissions.