finding

active

finding:scope-generalization-cot-boosts-2-digit-in-distribution-but-worsens-3-4-digit-ood

Scope generalization: CoT boosts 2-digit in-distribution but worsens 3-4 digit OOD

CoT increases dr for OOD operands.

Source paper

extracted_from

The Guanyin Protocol: A Framework for Immediately Establishing an Understanding of Both Causality and Compassion in LLM Systems Using Semantic Anchoring

(2025) · Edward Yi Chang · Kaya, Zeyneb N. · Ethan Chang

Neighborhood — ranked by edge-count

Communities (2)

community

Chain-of-Thought reasoning robustness & safety
members_of
CoT effects on generalization, multimodal QA accuracy, and AI safety alignment training.
Chain-of-thought generalization trade-offs
members_of
Empirical studies showing CoT reasoning improves ID performance while harming OOD generalization, with probability calibration as a mitigation strategy.

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

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

CoT improves in-distribution but may harm out-of-distribution generalizationclaim0.855
Interpretation of scope generalization results
CoT boosts 2-digit ID accuracy but often worsens 3-4 digit OODfinding0.843
Scope generalization results after LoRA+CoT fine-tuning
Out-of-Distribution (OOD) Generalizationconcept0.773
Machine learning generalization when training and test distributions differ; linked to causal invariance.
Short rationales (LoRA+CoT) sometimes improve in-distribution performance but do not reliably reduce cross-base harmfinding0.756
E2 finding showing CoT's limited benefit for OOD transfer, consistent with larger dr out of scope
scope generalizationconcept0.738
Generalization from 2-digit to 3-4 digit arithmetic; limited by mismatch dr.
Multimodal-CoT trained with InstructBLIP/ChatGPT-generated rationales achieves 87.76% accuracy on ScienceQA, comparable to human-annotated rationale performance of 90.45%finding0.729
Evidence that Multimodal-CoT can operate without human-annotated reasoning chains by using large models to generate pseudo-rationales.
Multimodal-CoT with vision features achieves higher validation accuracy at early training epochs (epoch 1-3) compared to one-stage and two-stage language-only baselines on ScienceQAfinding0.727
Evidence that multimodal information accelerates convergence speed during training.
Larger S_max correlates with smaller θ50 across backbones in E3 (negative association consistent across pooling and metric choices)finding0.724
Key geometry-to-behavior bridge finding in E3; robust to pooling choice, cosine vs. L2, and frozen external encoder