Upload folder using huggingface_hub

Update README.md

Metadata: base_model points to canonical zai-org/GLM-5.3 for tree discoverability; Inferact credited in card body

Card: technical rewrite — comparison table, W4A4/W4A16 split, storage figures

Upload weights: NVFP4 target + MTP (weight-only)

Lineage: base_model_relation: quantized (force Quantized edge; not a merge/finetune)

Lineage: list both base (BF16) and quantized-from (Inferact) in base_model

Card: drop dead library field (vllm not a valid library_name; covered by tags)

Add model card + LICENSE

Update README.md

initial commit