A recurrent-depth architecture enables language models to improve reasoning performance by iterating computation in latent space, achieving gains equivalent to much larger models on benchmarks.
arXiv preprint arXiv:2406.04229 , year=
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
fields
cs.LG 3representative citing papers
LLMs display high variance and major accuracy drops on GSM-Symbolic variants of grade-school math problems, indicating they replicate training patterns rather than execute logical reasoning.
Small language models adapted for closed-loop graph algorithm execution achieve reliable policies for traversal and coloring but show high sensitivity to error accumulation in weighted procedures, demonstrating that step accuracy does not ensure rollout reliability.
citing papers explorer
-
Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach
A recurrent-depth architecture enables language models to improve reasoning performance by iterating computation in latent space, achieving gains equivalent to much larger models on benchmarks.
-
GSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models
LLMs display high variance and major accuracy drops on GSM-Symbolic variants of grade-school math problems, indicating they replicate training patterns rather than execute logical reasoning.
-
Closed-Loop Graph Algorithm Execution with Small Language Models: Step Accuracy and Rollout Reliability
Small language models adapted for closed-loop graph algorithm execution achieve reliable policies for traversal and coloring but show high sensitivity to error accumulation in weighted procedures, demonstrating that step accuracy does not ensure rollout reliability.