mondk

granite-4.2-3b-humanizer

README

License: apache-2.0

granite-4.2-3b-humanizer (LoRA)

QLoRA SFT of ibm-granite/granite-4.2-3b on mondk/Open-Rewrite-Sft (32 pairs original->rewrite), prompt style copied from jialinyyzz/humanizer (completion prompt + ### Rewritten:). Use with PEFT on top of the base model. Training: 28 train / 4 eval, 10 epochs, train_loss ~0.04.

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Model Details

Model Provider

mondk

Model Tree

Base

ibm-granite/granite-4.2-3b

Adapter

this model