Community questions fine-tuning on summarized CoT traces

Reddit discussion critiques fine-tuning on distilled chain-of-thought traces, arguing it cannot surpass base model capabilities. Users highlight that such fine-tunes miss the original reasoning diversity and may amplify biases.