s1K probe training data

<p dir="ltr">Data required to replicate the training of thought calibration probes. Files provided include:</p><ul><li>s1K per-step embeddings by DeepSeek-R1 distilled Qwen 2.5 32B and Llama 3.3 70B, and QwQ 32B</li><li>Labels for supervised correctness, consistency, novelty, and leafness</li></ul><p dir="ltr">These files are used in notebook <b>2-probe.ipynb</b>.</p>