Views
No views yet
generator.py and local src/ runtime files.checkpoints/model.pt: slim inference checkpoint containing model weights and config.tokenizer/tokenizer.json: Tokenizers BPE tokenizer JSON.generator.py: CLI inference entry point for JSON/XML structured output.src/: minimal custom PyTorch model and guided-decoding runtime.requirements.txt: minimal inference dependencies.config.json: exported metadata for this bundle.full,sliding,csa,hca,sliding,csa,hca,fullpip install -r requirements.txtpython generator.py --checkpoint checkpoints/model.pt --tokenizer tokenizer/tokenizer.json --format json --instruction "Return a JSON object with keys answer and confidence. Answer whether 2+2=4." --json_keys answer,confidence --json_key_types answer:boolean,confidence:number --temperature 0python generator.py --checkpoint checkpoints/model.pt --tokenizer tokenizer/tokenizer.json --format json --instruction "Answer yes or no: is 2+2=4?" --regex "\s*([Yy]es|[Nn]o)" --temperature 0