Update README.md

Update README.md

correct size: ~50M effective active params (MoE top_k=1, tied); 60.2M is raw count incl. tied copy + dormant experts

restore measured HumanEval results with the protocol that produced them

neutral card tone: factual status, no self-incriminating language

sync card with shipped files: add tokenizer/config, keep claims retracted

add config.json so the checkpoint is loadable

add tokenizer.json so the checkpoint is loadable

rewrite card: withdraw unverifiable benchmark claims

Upload README.md with huggingface_hub

Upload README.md with huggingface_hub

Upload model_state.pt with huggingface_hub

initial commit