Probe-guided early exit reduces tokens by up to 30% on GPQA-Diamond with similar accuracy on DeepSeek-R1 671B and GPT-OSS 120B

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• On GPQA-Diamond multihop questions, activation probes show genuine belief shifts during CoT generation rather than early stabilization, contrasting with MMLUfinding0.785
  Empirical finding contrasting difficult questions with easy ones, supporting genuine reasoning on hard tasks
• DeepSeek-R1 Llama 8b gains 0.16% accuracy on GSM8k with positive intervention (more reflections) at cost of ~2000 additional tokensfinding0.773
  Only model showing marginal benefit from increased reflection, at substantial token cost
• Probe-Guided Early Exitconcept0.766
  Using activation probes to terminate CoT generation early when the model's belief is already stable, saving compute
• For simple factual tasks F0-F3, probe directions show a sharp geometric transition in middle layers, with late-layer probes converging to high cosine similarity; A3 and F4-F5 show no clear transition.finding0.749
  Geometric evidence for convergence to stable truth directions only for simpler tasks.
• Model final answer is decodable from activations far earlier in CoT than CoT monitor detects on MMLU recall-based questions for both DeepSeek-R1 671B and GPT-OSS 120Bfinding0.743
  Core empirical result demonstrating early belief formation in easy tasks
• QwQ-32B on MATH-500: 21.0% reasoning token reduction at intervention strength -0.96 with only 0.34% accuracy lossfinding0.739
  Demonstrates reflection redundancy in stronger model on harder math benchmark
• Probes trained on A1 degrade significantly when evaluated on A2 and more on A3; training on A2 achieves only AUROC ~0.65 on A3.finding0.735
  Shows rapid generalization decay for arithmetic truth directions with each additional operation.
• MM probe trained on likely dataset achieves NIE of 0.70 (false→true) on LLaMA-2-13B, surprisingly strong but weaker than truth probesfinding0.733
  Likely-trained MM probe is a surprisingly effective causal baseline due to correlation between truth and probability on sp_en_trans