Transformer

Models lacking required metadata fields like `chat_template`, `eos_token`, `pipeline_tag`, and `library_name` may fail to work with Hugging Face's inference interface, missing the 'Use this model' button and auto-detection filters, or produce malformed outputs like infinite generation.

Models on Hugging Face that are fine-tuned, quantized, or adapted versions of base models require explicit metadata (base_model, base_model_relation) to maintain proper lineage. Without this, model relationships are misclassified or lost, making it difficult to understand a model's provenance and dependencies.