LaTER reduces LLM token usage 16-33% on reasoning benchmarks by exploring in latent space then switching to explicit CoT verification, with gains like 70% to 73.3% on AIME 2025 in the training-free version.
SeLaR: Selective Latent Reasoning in Large Language Models
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
abstract
Chain-of-Thought (CoT) has become a cornerstone of reasoning in large language models, yet its effectiveness is constrained by the limited expressiveness of discrete token sampling. Recent latent reasoning approaches attempt to alleviate this limitation by replacing discrete tokens with soft embeddings (probability-weighted mixtures of token embeddings) or hidden states, but they commonly suffer from two issues: (1) global activation injects perturbations into high-confidence steps, impairing reasoning stability; and (2) soft embeddings quickly collapse toward the highest-probability token, limiting exploration of alternative trajectories. To address these challenges, we propose SeLaR (Selective Latent Reasoning), a lightweight and training-free framework. SeLaR introduces an entropy-gated mechanism that activates soft embeddings only at low-confidence steps, while preserving discrete decoding at high-confidence steps. Additionally, we propose an entropy-aware contrastive regularization that pushes soft embeddings away from the dominant (highest-probability) token's direction, encouraging sustained exploration of multiple latent reasoning paths. Experiments on five reasoning benchmarks demonstrate that SeLaR consistently outperforms standard CoT and state-of-the-art training-free methods.
years
2026 4representative citing papers
A looped padded Transformer with parallel gold-CoT cross-entropy supervision matches explicit CoT accuracy at 3B scale and is 2.5–6.9× faster in the thought phase.
InfoKV is an entropy-aware KV cache compression framework that integrates token-level predictive uncertainty and layer-wise representation evolution with attention scores to improve performance on long-context reasoning benchmarks.
TARPO is a pure RL framework using a token-wise action router to switch between discrete token generation and latent reasoning in LLMs, with joint optimization showing outperformance on benchmarks.
citing papers explorer
-
LaTER: Efficient Test-Time Reasoning via Latent Exploration and Explicit Verification
LaTER reduces LLM token usage 16-33% on reasoning benchmarks by exploring in latent space then switching to explicit CoT verification, with gains like 70% to 73.3% on AIME 2025 in the training-free version.
-
Bridging the Gap Between Latent and Explicit Reasoning with Looped Transformers
A looped padded Transformer with parallel gold-CoT cross-entropy supervision matches explicit CoT accuracy at 3B scale and is 2.5–6.9× faster in the thought phase.
-
Information-Aware KV Cache Compression for Long Reasoning
InfoKV is an entropy-aware KV cache compression framework that integrates token-level predictive uncertainty and layer-wise representation evolution with attention scores to improve performance on long-context reasoning benchmarks.
-
TARPO: Token-Wise Latent-Explicit Reasoning via Action-Routing Policy Optimization
TARPO is a pure RL framework using a token-wise action router to switch between discrete token generation and latent reasoning in LLMs, with joint optimization showing outperformance on benchmarks.