REVIEW 22 cited by
Distilling System 2 into System 1
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
Large language models (LLMs) can spend extra compute during inference to generate intermediate thoughts, which helps to produce better final responses. Since Chain-of-Thought (Wei et al., 2022), many such System 2 techniques have been proposed such as Rephrase and Respond (Deng et al., 2023a), System 2 Attention (Weston and Sukhbaatar, 2023) and Branch-Solve-Merge (Saha et al., 2023). In this work we investigate self-supervised methods to ``compile'' (distill) higher quality outputs from System 2 techniques back into LLM generations without intermediate reasoning token sequences, as this reasoning has been distilled into System 1. We show that several such techniques can be successfully distilled, resulting in improved results compared to the original System 1 performance, and with less inference cost than System 2. We posit that such System 2 distillation will be an important feature of future continually learning AI systems, enabling them to focus System 2 capabilities on the reasoning tasks that they cannot yet do well.
Forward citations
Cited by 22 Pith papers
-
CritiqueDriveVLM: From Verifier-Guided Reinforcement Learning to Latent Thought Distillation for Autonomous Driving
Verifier-guided multi-turn RL plus latent-state distillation raises DriveLMM-o1 MCQ to 76.54% for a tool-free teacher and yields a 28-token student at 416 ms (88% faster).
-
Physics of Language Models: Part 4.1, Architecture Design and the Magic of Canon Layers
Canon layers—residual 1-d causal convolutions over adjacent tokens—boost synthetic reasoning depth 2-4x, lift NoPE to RoPE level, and bring GLA up to Mamba2/GDN, with qualitative real-world confirmation.
-
Are Large Reasoning Models Interruptible?
Interrupting large reasoning models or changing the problem mid-thought sharply degrades accuracy, with up to 60% drops and distinct failure modes: reasoning leakage, panic, and self-doubt.
-
KRETA: A Benchmark for Korean Reading and Reasoning in Text-Rich VQA Attuned to Diverse Visual Contexts
KRETA, a 2,577-item Korean text-rich VQA benchmark, shows vision-language models recognize Korean text well but lag in multi-step reasoning, especially in open-source models.
-
NaturalThoughts: Selecting and Distilling Reasoning Traces for General Reasoning Tasks
Selecting difficult, strategy-diverse reasoning traces from DeepSeek-R1 on NaturalReasoning questions improves supervised distillation of reasoning into 7B-70B student models.
-
Do Thinking Tokens Help or Trap? Towards More Efficient Large Reasoning Model
Suppressing 'thinking tokens' in a 1.5B reasoning model preserves accuracy while cutting tokens, and the proposed DuP-PO RL method improves both accuracy and efficiency over GRPO.
-
ETA: Efficiency through Thinking Ahead, A Dual Approach to Self-Driving with Large Models
An asynchronous dual-system architecture forecasts large-model features into the current frame and adds a small-model update to drive in near real time, scoring 69.53 on Bench2Drive at 50 ms.
-
How Far Are We from Optimal Reasoning Efficiency?
The authors define a reasoning efficiency frontier and a gap metric (REG), then train models with REO-RL to shrink the gap by at least 50% with only small accuracy losses.
-
What makes Reasoning Models Different? Follow the Reasoning Leader for Efficient Decoding
FoReaL-Decoding lets a strong reasoning model generate the first few tokens of each sentence and a weaker model complete the sentence, cutting theoretical FLOPs by 30-55% while retaining 86-100% of accuracy on four ma...
-
Integrating Neural and Symbolic Components in a Model of Pragmatic Question-Answering
A neuro-symbolic Rational Speech Act model with LLM proposers and evaluators predicts human question-answer patterns about as well as the fully hand-specified probabilistic model.
-
Self-Route: Automatic Mode Switching via Capability Estimation for Efficient Reasoning
Self-Route uses hidden-layer representations from a brief pre-inference plan to route each question to either short or long chain-of-thought, cutting tokens by 30-55% with under 2% accuracy loss.
-
System-1.5 Reasoning: Traversal in Language and Latent Spaces with Dynamic Shortcuts
System-1.5 Reasoning lets LLMs reason in latent space with early exits and step-skipping, matching chain-of-thought accuracy at over 20x speedup on GSM8K and StrategyQA.
-
TAG-INSTRUCT: Controlled Instruction Complexity Enhancement through Structure-based Augmentation
Operating on compressed semantic tags instead of raw text gives more controllable and effective instruction complexity augmentation for LLM fine-tuning, with reported gains over Evol-Instruct, Tree-Instruct, Auto-Inst...
-
Don't "Overthink" Passage Reranking: Is Reasoning Truly Necessary?
Standard pointwise rerankers beat reasoning-based pointwise rerankers under identical training, and removing the reasoning process at inference partially recovers the gap.
-
TrimR: Verifier-based Training-Free Thinking Compression for Efficient Test-Time Scaling
TrimR uses a lightweight verifier to detect redundant thoughts in reasoning models and truncates them, cutting runtime by up to 70% without retraining.
-
Reasoning on a Budget: A Survey of Adaptive and Controllable Test-Time Compute in LLMs
The survey's L1/L2 taxonomy and benchmark show that current reasoning models waste compute on easy problems and underthink hard ones, motivating more adaptive inference.
-
Activation Control for Efficiently Eliciting Long Chain-of-thought Ability of Language Models
Training-free amplification of selected last-layer activations, combined with 'wait' token insertion, elicits long chain-of-thought reasoning in base LLMs and improves accuracy on math and science benchmarks.
-
From Long to Short: LLMs Excel at Trimming Own Reasoning Chains
EDIT searches across step-count prompts to find the shortest reasoning path the model consistently answers correctly.
-
Wait, We Don't Need to "Wait"! Removing Thinking Tokens Improves Reasoning Efficiency
Suppressing "Wait"-like reflection tokens at decode time reduces reasoning token counts by 27-51% across five R1-style model families, with mixed accuracy effects.
-
Pangu Embedded: An Efficient Dual-system LLM Reasoner with Metacognition
Pangu Embedded, a 7B reasoner trained with iterative distillation, RL, and an adaptive fast/slow thinking scheme, reports superior benchmark scores to similarly sized Qwen3-8B and GLM-4-9B.
-
Towards Concise and Adaptive Thinking in Large Reasoning Models: A Survey
A comprehensive review that categorizes methods for shortening and adaptively triggering chain-of-thought reasoning in large language models.
- Optimizing Length Compression in Large Reasoning Models
Discussion (0). Sign in to comment.