Collection of lightweights (between 2b and 3b parameters) instruction-tuned prompt enhancing models. Feel free to compare them !