linda1476/Gemma4-E4B-RP-RLAIF_Pubilc

Upload README.md with huggingface_hub

r5_v2 adopted: GGUF+adapter replaced, card updated with full honest arc (22.8%->46.3%, hard_fail halved, human 65%)

model card: add pairwise preference result (base preferred; honest verdict)

Gemma4 E4B RP RLAIF: Q4_K_M GGUF + LoRA adapter + model card

initial commit