Upload README.md with huggingface_hub

Update training time

Expand training section with full hyperparams and data breakdown

Redesign model card: professional layout, remove duplicates, add badges

Add benchmark charts to model card

Add chart: bfcl_breakdown.png

Add chart: v1_vs_v2.png

Add chart: comparison_models.png

Add vLLM serving instructions

Fix: re-merge with all weights (k_norm) for vLLM compatibility

Add official BFCL leaderboard scores

Fix double GPU typo

Fix GPU name

Add capabilities note, anonymize GPU

Fix: wrap special tokens in code formatting

Fix: remove CLI warning artifact, clean up GGUF section

Add GGUF section to model card

Add evaluation script

Add example inference script

Add 1000-query evaluation dataset

Add detailed model card

Gemma 4 E4B tool-calling v02 (SFT merged, 94.4% accuracy on 1000 queries)

initial commit