Correct chat template example for Transformers

Add Colab quickstart, chat template, and release metadata

Handle left-padded batches without NaN logits

Set decoder-friendly tokenizer padding defaults

Complete standard tokenizer and generation metadata

Use the standard Transformers initialization hook

Make tied embeddings compatible with Transformers loading

Fix relative import for HF dynamic modules

Add standard Transformers and AutoTokenizer support

Show the supplied Vortex banner in the model card

Publish Vortex Alpha research preview

initial commit