REVIEW 2 cited by
Conditions for Length Generalization in Learning Reasoning Skills
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Reasoning is a fundamental capability of AI agents. Recently, large language models (LLMs) have shown remarkable abilities to perform reasoning tasks. However, numerous evaluations of the reasoning capabilities of LLMs have also showed some limitations. An outstanding limitation is length generalization, meaning that when trained on reasoning problems of smaller lengths or sizes, the resulting models struggle with problems of larger sizes or lengths. This potentially indicates some theoretical limitations of generalization in learning reasoning skills. These evaluations and their observations motivated us to perform a theoretical study of the length generalization problem. This work focuses on reasoning tasks that can be formulated as Markov dynamic processes (MDPs) and/or directed acyclic graphs (DAGs). It identifies and proves conditions that decide whether the length generalization problem can be solved or not for a reasoning task in a particular representation. Experiments are also conducted to verify the theoretical results.
Forward citations
Cited by 2 Pith papers
-
Enhancing Auto-regressive Chain-of-Thought through Loop-Aligned Reasoning
Loop-aligned supervision lets a looped Transformer generate CoT chains beyond training length, and those chains improve an auto-regressive CoT model's length generalization.
-
Ignore the KL Penalty! Boosting Exploration on Critical Tokens to Enhance RL Fine-Tuning
A certainty-weighted KL penalty that down-weights the penalty on low-confidence tokens improves RL fine-tuning exploration for arithmetic in GPT-2.
Discussion (0). Continue with ORCID to comment.