Excited to share our new work led by Sahana, on improving the quality of rationales generated by small LMs!