nextShakespeare

nextShakespeare is a decoder-only, character-level Transformer language model trained on Tiny Shakespeare style text. This repo uses custom PyTorch code rather than transformers native model classes.

Model assets

• Weights: checkpoints/best_model.pt
• Config: configs/model.yaml
• Vocabulary: data/vocab/char_vocab.json
• Runtime code: llm/

Quickstart

git clone https://huggingface.co/LiManshu/nextShakespeare
cd nextShakespeare
pip install -r requirements.txt
python inference.py --prompt "First Citizen:\n" --max_length 200 --temperature 0.8

Notes

• This is a custom-code checkpoint. It is not directly loadable via AutoModel.from_pretrained() yet.
• For a web demo, see the companion Space: https://huggingface.co/spaces/LiManshu/manshu-init