Card: base model is Qwen/Qwen3.5-2B-Base

Fix base model: point adapter at Qwen/Qwen3.5-2B-Base (the true training base; -Base, not the finetune)

Card: drop release-process meta ('private, public on release' / 'on release')

Card: point install to the public code repo (EdisonScientific/glyph)

Card: add 'Installing the code' section (link + install for the glyph package)

Add MarkushGlyph LoRA adapter weights

Add model card

Add processor config

Add chat template

Add fixed adapter_config (base -> Qwen/Qwen3.5-2B)

initial commit