pith. sign in

arxiv: 2201.11903 · v6 · pith:S4TFVHNWnew · submitted 2022-01-28 · 💻 cs.CL · cs.AI

Chain-of-Thought Prompting Elicits Reasoning in Large Language Models

Jason Wei , Xuezhi Wang , Dale Schuurmans , Maarten Bosma , Brian Ichter , Fei Xia , Ed Chi , Quoc Le
show 1 more author
Denny Zhou
This is my paper

Pith reviewed 2026-05-10 12:49 UTC · model grok-4.3

classification 💻 cs.CL cs.AI
keywords chain of thoughtpromptingreasoninglarge language modelsGSM8Kfew-shot promptingarithmetic reasoning
0
0 comments X

The pith

Chain of thought prompting lets large language models reach state-of-the-art accuracy on math word problems using only eight examples.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that large language models gain substantial reasoning ability when their prompts include a few examples that demonstrate a chain of intermediate steps rather than direct answers. This simple change improves results on arithmetic, commonsense, and symbolic tasks without any model training or parameter updates. The gains are especially large for the biggest models tested, where a 540 billion parameter system using eight such examples surpasses the prior best result on the GSM8K math benchmark, including systems that were fine-tuned and equipped with separate verifiers. A sympathetic reader would care because the method shows how to unlock capabilities already present in scaled models through prompt design alone.

Core claim

Generating a chain of thought, a series of intermediate reasoning steps, significantly improves the ability of large language models to perform complex reasoning. Such reasoning abilities emerge naturally in sufficiently large language models via chain of thought prompting, where a few chain of thought demonstrations are provided as exemplars in the prompt.

What carries the argument

Chain of thought prompting: the inclusion of a small number of input examples that each show a sequence of explicit reasoning steps before the final answer.

If this is right

  • A 540B model with eight chain-of-thought exemplars reaches state-of-the-art accuracy on GSM8K, beating fine-tuned GPT-3 with a verifier.
  • The same prompting method improves results across arithmetic, commonsense, and symbolic reasoning benchmarks.
  • Reasoning performance scales with model size once chain-of-thought exemplars are supplied.
  • Complex tasks become solvable without retraining or additional fine-tuning data.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The method could extend to problems that require much longer reasoning chains if the prompt examples are lengthened accordingly.
  • It opens a route to more interpretable model outputs by making the generated steps visible to users.
  • Standard few-shot prompting may systematically underestimate what current models can do on reasoning benchmarks.
  • Varying the structure of the reasoning steps in the examples would test whether models follow the logic or merely copy surface patterns.

Load-bearing premise

The performance gains are caused by the explicit reasoning steps rather than by simply supplying longer or more detailed prompts in general.

What would settle it

A controlled test in which prompts of matched length contain no reasoning steps but still produce the same accuracy gains on GSM8K would falsify the claim.

read the original abstract

We explore how generating a chain of thought -- a series of intermediate reasoning steps -- significantly improves the ability of large language models to perform complex reasoning. In particular, we show how such reasoning abilities emerge naturally in sufficiently large language models via a simple method called chain of thought prompting, where a few chain of thought demonstrations are provided as exemplars in prompting. Experiments on three large language models show that chain of thought prompting improves performance on a range of arithmetic, commonsense, and symbolic reasoning tasks. The empirical gains can be striking. For instance, prompting a 540B-parameter language model with just eight chain of thought exemplars achieves state of the art accuracy on the GSM8K benchmark of math word problems, surpassing even finetuned GPT-3 with a verifier.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 4 minor

Summary. The paper introduces chain-of-thought (CoT) prompting, in which few-shot exemplars are augmented with explicit sequences of intermediate reasoning steps. Experiments across three large language models (including PaLM-540B) demonstrate that this format yields substantial accuracy gains on arithmetic, commonsense, and symbolic reasoning benchmarks relative to standard few-shot prompting. The headline result is that eight fixed CoT exemplars enable the 540B model to reach state-of-the-art accuracy on GSM8K, surpassing a fine-tuned GPT-3 model equipped with a verifier. Gains are shown to emerge only at large scale, with supporting ablations and multiple runs on most tasks.

Significance. If the empirical results hold, the work is significant because it shows that complex reasoning capabilities can be elicited from LLMs via a simple, training-free prompting change. The consistent improvements across task families, the clear scaling threshold, the use of held-out test sets, and the reporting of multiple runs and error bars provide solid grounding. The approach has immediate practical value for deploying LLMs on reasoning problems and raises interesting questions about how reasoning emerges in large models.

minor comments (4)
  1. [§3.1] The answer-extraction procedure for GSM8K (and similar math tasks) should be described more explicitly, including how cases where the model fails to produce a boxed final answer are handled and whether any post-processing rules were tuned on the test set.
  2. [Figure 2] Figure 2 (scaling curves) would benefit from error bars on every point rather than only on selected runs; this would make the emergence threshold at large scale easier to assess visually.
  3. [§4.3] The paper compares CoT prompting against standard few-shot baselines but does not include a control that matches prompt length while removing the reasoning structure (e.g., repeated filler sentences). While the existing ablations make a length-only explanation unlikely, this additional control would further isolate the contribution of the reasoning format.
  4. [Appendix B] A brief discussion of how exemplar selection was performed (random vs. curated) and whether results are sensitive to the particular eight GSM8K exemplars would strengthen reproducibility claims.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their thorough summary of our work, for highlighting its significance, and for recommending acceptance. No major comments or criticisms were raised in the report.

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper presents purely empirical results from prompting experiments on large language models, with all accuracy metrics measured on held-out standard benchmarks such as GSM8K. No equations, derivations, or fitted parameters appear that could reduce claimed gains to quantities defined by the same inputs. The method is a straightforward prompting technique whose effects are directly observed via comparisons to few-shot baselines, and no self-citation chain or uniqueness theorem is invoked to justify the central claims. The derivation chain is therefore self-contained and consists only of experimental observations.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim is supported by empirical measurement rather than new theoretical axioms or fitted parameters; the work assumes only that large language models can follow in-context patterns, a standard premise in prompting research.

axioms (1)
  • domain assumption Large language models can learn to imitate reasoning patterns shown in a small number of in-context examples
    Invoked throughout the prompting experiments; this is the background assumption shared by all few-shot prompting methods.

pith-pipeline@v0.9.0 · 5447 in / 1297 out tokens · 31141 ms · 2026-05-10T12:49:15.009832+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 60 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Can LLMs Write Correct TLA+ Specifications? Evaluating Natural-Language-to-TLA+ Generation

    cs.AI 2026-06 accept novelty 8.0

    Across 30 LLMs and 205 TLA+ tasks, syntactic correctness reaches at most 26.6% and semantic correctness 8.6%, with all successes limited to progressive prompting and no advantage from larger models.

  2. REALISTA: Realistic Latent Adversarial Attacks that Elicit LLM Hallucinations

    cs.CL 2026-05 unverdicted novelty 8.0

    REALISTA optimizes continuous combinations of valid editing directions in latent space to produce realistic adversarial prompts that elicit hallucinations more effectively than prior methods, including on large reason...

  3. Slot Machines: How LLMs Keep Track of Multiple Entities

    cs.CL 2026-04 unverdicted novelty 8.0

    LLM activations encode current and prior entities in orthogonal slots, but models only use the current slot for explicit factual retrieval despite prior-slot information being linearly decodable.

  4. Stability and Generalization in Looped Transformers

    cs.LG 2026-04 unverdicted novelty 8.0

    Looped transformers with recall and outer normalization produce reachable, input-dependent fixed points with stable gradients, enabling generalization, while those without recall cannot; a new internal recall variant ...

  5. GIANTS: Generative Insight Anticipation from Scientific Literature

    cs.CL 2026-04 unverdicted novelty 8.0

    GIANTS-4B, trained with RL on a new 17k-example benchmark of parent-to-child paper insights, achieves 34% relative improvement over gemini-3-pro in LM-judge similarity and is rated higher-impact by a citation predictor.

  6. On the Mirage of Long-Range Dependency, with an Application to Integer Multiplication

    cs.LG 2026-03 unverdicted novelty 8.0

    Long-range dependency in integer multiplication is a mirage from 1D representation; a 2D grid reduces it to local 3x3 operations, letting a 321-parameter neural cellular automaton generalize perfectly to inputs 683 ti...

  7. Evaluating Large Language Models in Scientific Discovery

    cs.AI 2025-12 unverdicted novelty 8.0

    The SDE benchmark shows LLMs lag on scientific discovery tasks relative to general science tests, with diminishing scaling returns and shared weaknesses across models.

  8. MobiBench: Multi-Branch, Modular Benchmark for Mobile GUI Agents

    cs.AI 2025-12 accept novelty 8.0

    MobiBench is the first modular multi-path offline benchmark for mobile GUI agents, achieving 94.72% agreement with human evaluators while allowing component-level analysis.

  9. Agentic AI for Multi-Stage Physics Experiments at a Large-Scale User Facility Particle Accelerator

    physics.acc-ph 2025-09 unverdicted novelty 8.0

    A language-model-driven agentic AI system autonomously executes multi-stage physics experiments at a production synchrotron light source, reducing preparation time by two orders of magnitude while upholding safety con...

  10. DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines

    cs.CL 2023-10 conditional novelty 8.0

    DSPy compiles short declarative programs into LM pipelines that self-optimize and outperform both standard few-shot prompting and expert-written chains on math, retrieval, and QA tasks.

  11. Tree of Thoughts: Deliberate Problem Solving with Large Language Models

    cs.CL 2023-05 accept novelty 8.0

    Tree of Thoughts enables language models to solve complex planning tasks by generating, evaluating, and searching over coherent intermediate thoughts in a tree, raising Game of 24 success from 4% to 74% with GPT-4.

  12. Generative Agents: Interactive Simulacra of Human Behavior

    cs.HC 2023-04 accept novelty 8.0

    Generative agents with memory streams, reflection, and planning using LLMs exhibit believable individual and emergent social behaviors in a simulated town.

  13. Instruction Tuning with GPT-4

    cs.CL 2023-04 unverdicted novelty 8.0

    GPT-4-generated instruction data produces superior zero-shot performance in finetuned LLaMA models versus prior state-of-the-art data.

  14. Progress measures for grokking via mechanistic interpretability

    cs.LG 2023-01 accept novelty 8.0

    Grokking arises from gradual amplification of a Fourier-based circuit in the weights followed by removal of memorizing components.

  15. PAL: Program-aided Language Models

    cs.CL 2022-11 conditional novelty 8.0

    PAL improves few-shot reasoning accuracy by having LLMs generate executable programs rather than text-based chains of thought, outperforming much larger models on math and logic benchmarks.

  16. Code as Policies: Language Model Programs for Embodied Control

    cs.RO 2022-09 accept novelty 8.0

    Language models generate robot policy code from natural language commands via few-shot prompting, enabling spatial-geometric reasoning, generalization, and precise control on real robots.

  17. Rethinking the Role of Demonstrations: What Makes In-Context Learning Work?

    cs.CL 2022-02 accept novelty 8.0

    Randomly replacing labels in in-context demonstrations barely hurts performance, showing that label space, input distribution, and sequence format drive in-context learning more than ground-truth labels.

  18. TO-Master: an LLM-agent framework for automated topology optimization

    cs.CE 2026-07 unverdicted novelty 7.0

    TO-Master is an LLM agent framework that orchestrates finite-element topology optimization from conversational inputs, supporting 2D/3D compliance, thermal, stress-constrained, and multi-load cases while reproducing b...

  19. DART: Difficulty-Adaptive Routing for Zero-Shot Video Temporal Grounding

    cs.CV 2026-07 unverdicted novelty 7.0

    DART routes zero-shot video temporal grounding queries by difficulty using DPP entropy, achieving up to 3.5 mIoU gains with 7x fewer frames on Charades-STA and ActivityNet Captions.

  20. MultiUAV-Plat: An LLM-Oriented Platform, Benchmark and Framework for Multi-UAV Collaborative Task Planning

    cs.AI 2026-06 unverdicted novelty 7.0

    MultiUAV-Plat supplies a new RESTful simulation platform and 1500-task benchmark where Agent4Drone reaches 57.9% task pass rate versus 30.6% for ReAct baseline across 75 multi-UAV missions.

  21. What Drives Interactive Improvement from Feedback?

    cs.AI 2026-06 unverdicted novelty 7.0

    Controlled student-teacher experiments across four benchmarks show interactive gains are driven more by the student's ability to use feedback than by teacher quality, with self-feedback adding little beyond unguided retries.

  22. Grounded Iterative Language Planning: How Parameterized World Models Reduce Hallucination Propagation in LLM Agents

    cs.AI 2026-06 unverdicted novelty 7.0

    GILP trains a parameterized backbone for valid actions and state predictions, then uses a consistency gate with LLM drafts to reduce hallucinated-state rate from 0.176 to 0.035 on GPT-4o-mini while raising success fro...

  23. Structure Before Collapse: Transient semantic geometry in next-token prediction

    cs.LG 2026-06 unverdicted novelty 7.0

    Semantic geometry emerges transiently early in next-token prediction training before collapsing to Neural Collapse symmetry in synthetic settings with latent semantic factors.

  24. C3-Bench: A Context-Aware Change Captioning Benchmark

    cs.CV 2026-06 unverdicted novelty 7.0

    C3-Bench supplies a multi-domain dataset and LLM-based evaluation protocol that exposes systematic failures in existing change captioning models outside their training regimes.

  25. Lost in Aggregation: A Multi-Scale Diagnostic Benchmark for LLM Spatial Navigation

    physics.soc-ph 2026-06 unverdicted novelty 7.0

    A new diagnostic benchmark decomposes LLM spatial navigation into three cognitive scales and shows that cross-scale aggregation, not single-level deficits, causes failure beyond small mazes.

  26. A Verifiable Search Is Not a Learnable Chain-of-Thought

    cs.LG 2026-06 unverdicted novelty 7.0

    Verifiable search procedures cannot be learned as forward chain-of-thought by language models; they instead learn memorization, verification, or require precomputed catalogs.

  27. Deep Research in Physical Sciences: A Multi-Agent Framework and Comprehensive Benchmark

    physics.comp-ph 2026-06 unverdicted novelty 7.0

    PhySciBench benchmark shows current AI models achieve at most 33.5% accuracy on physical science tasks; DelveAgent framework improves accuracy by up to 7.5 points and cuts costs to one-third.

  28. Beyond Prediction: Tail-Aware Scheduling for LLM Inference

    cs.LG 2026-06 unverdicted novelty 7.0

    Presents a distribution-aware scheduling framework for LLM inference that reduces P99 TTLT by 35-50% and TTFT by 34-47% versus SRPT with perfect length knowledge using statistical signals instead of predictions.

  29. Structure from Reasoning, Numbers from Search: On-Premise Open LLMs as Structural Priors for Coupled MIMO Controller Tuning

    cs.AI 2026-06 conditional novelty 7.0

    Open LLMs function as structural priors for MIMO controller tuning by proposing asymmetric structures on coupled plants, reaching better penalized cost with fewer evaluations than pure optimization or classical methods.

  30. Optical Reasoning: Rethinking Images as an Expressive Reasoning Medium Beyond Text

    cs.AI 2026-06 unverdicted novelty 7.0

    Optical reasoning encodes rationales in images rather than text, matching or exceeding text-based performance on math, science, and multimodal benchmarks while cutting tokens by 28.57% on language tasks and 16% on mul...

  31. Self-Harness: Harnesses That Improve Themselves

    cs.CL 2026-06 unverdicted novelty 7.0

    Self-Harness lets LLM agents autonomously refine their interaction harnesses through weakness mining, proposal generation, and validation, raising held-out pass rates on Terminal-Bench-2.0 from 40.5% to 61.9%, 23.8% t...

  32. Continuous Language Diffusion as a Decoder-Interface Problem

    cs.CL 2026-06 unverdicted novelty 7.0

    Continuous language diffusion works by entering high-margin decoder basins where frozen T5 embeddings recover 93-96% of native decisions and linear readouts reach 97.9% agreement, implying models should be evaluated a...

  33. SciTrace: Trajectory-Aware Safety Reasoning for Scientific Discovery Agents

    cs.AI 2026-06 unverdicted novelty 7.0

    SciTrace embeds cumulative safety deliberation and trajectory-aware verification into scientific agent pipelines, claiming SOTA safety gains and detection of 78.8% of compositional risks missed by single-step checks.

  34. Toward Calibrated, Fair, and accurate Deepfake Detection

    cs.LG 2026-06 unverdicted novelty 7.0

    Face-Feature Tuning is a label-free logit remapping method that reduces FPR/TPR gaps across groups in deepfake detection while preserving overall accuracy.

  35. Testing LLM Arithmetic Reasoning Generalization with Automatic Numeric-Remapping Attacks

    cs.CR 2026-06 unverdicted novelty 7.0

    An automatic numeric-remapping attack generator reveals 12-26 point accuracy drops on GSM8K for three LLMs while MAWPS and MultiArith stay near 98%.

  36. Fixing FOLIO and MALLS: Verified Annotations and an LLM-assisted Framework to Focus Human Relabeling

    cs.CL 2026-06 unverdicted novelty 7.0

    Audit finds 36-39% incorrect FOL labels in FOLIO and MALLS; corrections raise LLM accuracy 9-22 points and an LLM-guided review framework achieves 90% dataset quality after checking fewer than 24% of examples.

  37. The Regularizing Power of Language-Training Deepfake Detectors

    cs.CV 2026-05 unverdicted novelty 7.0

    A dual-encoder deepfake detector pairs a frozen specialist with a LoRA-tuned MLLM, trained first via binary alignment then via RL to reward explain-then-classify behavior, yielding improved cross-dataset performance a...

  38. Social Reasoning in Machines: Investigating Collective Truth-Seeking Dynamics in Large Language Model Debate

    cs.MA 2026-05 unverdicted novelty 7.0

    LLM multi-agent debate with epistemically diverse models improves truth-seeking performance on questionnaires and is claimed to be mechanistically grounded in the Argumentative Theory of Reasoning.

  39. Memory Architectures for Multi-Turn Text-to-SQL: A Benchmark and Empirical Study

    cs.CL 2026-05 unverdicted novelty 7.0

    EnterpriseMem-Bench shows stateless multi-turn Text-to-SQL accuracy drops to zero by turn 3, working memory is the main driver of gains, and additional memory components yield model- and dataset-dependent effects from...

  40. ARBITER: Reasoning Trajectory Basins and Majority Vote Failures in Test-Time Sampling

    cs.LG 2026-05 unverdicted novelty 7.0

    ARBITER models reasoning trajectory basins in test-time sampling and uses model-internal signals to correct majority-vote failures, recovering part of the oracle gap on math benchmarks.

  41. Spectral Retrieval: Multi-Scale Sinc Convolution over Token Embeddings for Localized Retrieval in LLM Multi-Agent Systems

    cs.IR 2026-05 unverdicted novelty 7.0

    Spectral Retrieval uses multi-scale sinc convolutions on token embeddings to interpolate between per-token MaxSim and mean-pooling, achieving large gains on synthetic and LIMIT-small benchmarks for localized retrieval.

  42. Inductive Deductive Synthesis: Enabling AI to Generate Formally Verified Systems

    cs.AI 2026-05 unverdicted novelty 7.0

    IDS is an agentic LLM system that incrementally synthesizes both implementation and proof for distributed key-value stores, succeeding on all 7 specs where prior agents succeeded on only 2.

  43. TO-Agents: A Multi-Agent AI Pipeline for Preference-Guided Topology Optimization

    cs.AI 2026-05 unverdicted novelty 7.0

    A multi-agent pipeline iteratively refines topology optimization outputs to match natural language preferences for branched structures, achieving 60% success rate across replicates in cantilever and phone-stand tasks.

  44. CopT: Contrastive On-Policy Thinking with Continuous Spaces for General and Agentic Reasoning

    cs.CL 2026-05 unverdicted novelty 7.0

    CopT reverses CoT by eliciting a draft answer first then using continuous-embedding contrastive verification and on-policy thinking to reflect and correct, yielding up to 23% higher accuracy and 57% fewer tokens witho...

  45. DecisionBench: A Benchmark for Emergent Delegation in Long-Horizon Agentic Workflows

    cs.AI 2026-05 unverdicted novelty 7.0

    DecisionBench supplies a fixed task suite, model pool, delegation interface, and multi-axis metrics to evaluate emergent delegation, showing similar quality across awareness conditions but 15-31 point headroom under p...

  46. Counterfactual Likelihood Tests for Indirect Influence in Private Reasoning Channels

    cs.LG 2026-05 unverdicted novelty 7.0

    Counterfactual likelihood tests detect indirect influence through public channels in private reasoning models, validated on a 7B role-channel model showing asymmetric A-to-B influence and complete pathway identificati...

  47. Test-Time Hinting for Black-Box Vision-Language Models

    cs.CV 2026-05 unverdicted novelty 7.0

    Test-Time Hinting trains a hint generator to prepend contextual guidance to VLM prompts, improving accuracy on natural-image VQA benchmarks with generalization to unseen tasks and models.

  48. Learning, Fast and Slow: Towards LLMs That Adapt Continually

    cs.LG 2026-05 unverdicted novelty 7.0

    Fast-Slow Training uses context optimization as fast weights alongside parameter updates as slow weights to achieve up to 3x better sample efficiency, higher performance, and less catastrophic forgetting than standard...

  49. Reflection Anchors for Propagation-Aware Visual Retention in Long-Chain Multimodal Reasoning

    cs.CV 2026-05 unverdicted novelty 7.0

    RAPO uses an information-theoretic lower bound on visual gain to select high-entropy reflection anchors and optimizes a chain-masked KL surrogate, delivering gains over baselines on reasoning benchmarks across LVLM backbones.

  50. Beyond Accuracy: Evaluating Strategy Diversity in LLM Mathematical Reasoning

    cs.AI 2026-05 unverdicted novelty 7.0

    Frontier LLMs achieve 95-100% accuracy on AMC/AIME problems but recover far fewer distinct valid strategies than human references, while collectively generating 50 novel strategies.

  51. SearchSkill: Teaching LLMs to Use Search Tools with Evolving Skill Banks

    cs.AI 2026-05 unverdicted novelty 7.0

    SearchSkill improves exact match scores and retrieval efficiency on open-domain QA by conditioning LLM actions on skills from an evolving SkillBank updated from failure patterns via two-stage SFT.

  52. CA-SQL: Complexity-Aware Inference Time Reasoning for Text-to-SQL via Exploration and Compute Budget Allocation

    cs.CL 2026-05 unverdicted novelty 7.0

    CA-SQL achieves 51.72% execution accuracy on the challenging tier of the BIRD benchmark using GPT-4o-mini by scaling exploration breadth according to estimated task difficulty, evolutionary prompt seeding, and candida...

  53. KL for a KL: On-Policy Distillation with Control Variate Baseline

    cs.LG 2026-05 unverdicted novelty 7.0

    vOPD stabilizes on-policy distillation gradients by subtracting a closed-form per-token negative reverse KL baseline as a detached control variate, preserving unbiasedness while lowering variance and matching expensiv...

  54. Can RL Teach Long-Horizon Reasoning to LLMs? Expressiveness Is Key

    cs.AI 2026-05 unverdicted novelty 7.0

    RL training on more expressive logical tasks follows a steeper power-law scaling with reasoning depth and transfers more efficiently to math and reasoning benchmarks.

  55. Can RL Teach Long-Horizon Reasoning to LLMs? Expressiveness Is Key

    cs.AI 2026-05 unverdicted novelty 7.0

    RL training compute for logical reasoning follows a power law in proof depth whose exponent rises with logic expressiveness, and more expressive training yields larger gains on downstream benchmarks.

  56. TRIP-Evaluate: An Open Multimodal Benchmark for Evaluating Large Models in Transportation

    cs.CV 2026-04 accept novelty 7.0

    TRIP-Evaluate is a new open multimodal benchmark with 837 text, image, and point-cloud items organized by a role-task-knowledge taxonomy to evaluate large models on transportation workflows.

  57. XGRAG: A Graph-Native Framework for Explaining KG-based Retrieval-Augmented Generation

    cs.AI 2026-04 unverdicted novelty 7.0

    XGRAG uses graph perturbations to quantify component contributions in GraphRAG and achieves 14.81% better explanation quality than text-based baselines on QA datasets, with correlations to graph centrality.

  58. RAG-Reflect: Agentic Retrieval-Augmented Generation with Reflections for Comment-Driven Code Maintenance on Stack Overflow

    cs.SE 2026-04 unverdicted novelty 7.0

    RAG-Reflect achieves F1=0.78 on valid comment-edit prediction using retrieval-augmented reasoning and self-reflection, outperforming baselines and approaching fine-tuned models without retraining.

  59. R2IF: Aligning Reasoning with Decisions via Composite Rewards for Interpretable LLM Function Calling

    cs.LG 2026-04 unverdicted novelty 7.0

    R2IF improves LLM function-calling accuracy by up to 34.62% on BFCL using a composite reward system with CER and SMV components optimized via GRPO, while increasing interpretability through positive CoT effectiveness.

  60. HiPO: Hierarchical Preference Optimization for Adaptive Reasoning in LLMs

    cs.AI 2026-04 unverdicted novelty 7.0

    HiPO improves LLM reasoning performance by optimizing preferences separately on response segments rather than entire outputs.

Reference graph

Works this paper leans on

86 extracted references · 86 canonical work pages · cited by 350 Pith papers · 14 internal anchors

  1. [1]

    Michael Ahn, Anthony Brohan, Noah Brown, Yevgen Chebotar, Omar Cortes, Byron David, Chelsea Finn, Keerthana Gopalakrishnan, Karol Hausman, Alex Herzog, et al. 2022. https://arxiv.org/abs/2204.01691 Do as I can, not as I say: Grounding language in robotic affordances . arXiv preprint arXiv:2204.01691

    work page internal anchor Pith review arXiv 2022

  2. [2]

    Aida Amini, Saadia Gabriel, Shanchuan Lin, Rik Koncel-Kedziorski, Yejin Choi, and Hannaneh Hajishirzi. 2019. https://aclanthology.org/N19-1245 M ath QA : Towards interpretable math word problem solving with operation-based formalisms . In Proceedings of the 2019 Conference of the North A merican Chapter of the Association for Computational Linguistics: Hu...

    work page 2019

  3. [3]

    Daniel Andor, Luheng He, Kenton Lee, and Emily Pitler. 2019. https://doi.org/10.18653/v1/D19-1609 Giving BERT a calculator: Finding operations and arguments with reading comprehension . EMNLP

    work page doi:10.18653/v1/d19-1609 2019

  4. [4]

    Jacob Andreas, Dan Klein, and Sergey Levine. 2018. https://aclanthology.org/N18-1197 Learning with latent language . NAACL

    work page 2018

  5. [5]

    Jacob Austin, Augustus Odena, Maxwell Nye, Maarten Bosma, Henryk Michalewski, David Dohan, Ellen Jiang, Carrie Cai, Michael Terry, Quoc Le, et al. 2021. https://arxiv.org/abs/2108.07732 Program synthesis with large language models . arXiv preprint arXiv:2108.07732

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  6. [6]

    BIG-bench collaboration . 2021. https://github.com/google/BIG-bench/ Beyond the imitation game: Measuring and extrapolating the capabilities of language models . In preparation

    work page 2021

  7. [7]

    Kaj Bostrom, Xinyu Zhao, Swarat Chaudhuri, and Greg Durrett. 2021. https://doi.org/10.18653/v1/2021.emnlp-main.506 Flexible generation of natural language deductions . EMNLP

    work page doi:10.18653/v1/2021.emnlp-main.506 2021

  8. [8]

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel Ziegler, Jeffrey Wu, Clemens Winter, Chris Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott Gr...

    work page 2020

  9. [9]

    Jonathon Cai, Richard Shin, and Dawn Song. 2017. https://arxiv.org/abs/1704.06611 Making neural programming architectures generalize via recursion . ICLR

    work page Pith review arXiv 2017

  10. [10]

    Oana-Maria Camburu, Tim Rockt \"a schel, Thomas Lukasiewicz, and Phil Blunsom. 2018. https://arxiv.org/pdf/1812.01193.pdf e- SNLI : Natural language inference with natural language explanations . NeurIPS

    work page internal anchor Pith review arXiv 2018

  11. [11]

    Howard Chen, Jacqueline He, Karthik Narasimhan, and Danqi Chen. 2022. https://arxiv.org/abs/2204.11790 Can rationalization improve robustness? NAACL

    work page arXiv 2022

  12. [12]

    Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde de Oliveira Pinto, Jared Kaplan, Harri Edwards, Yuri Burda, Nicholas Joseph, Greg Brockman, et al. 2021. https://arxiv.org/abs/2107.03374 Evaluating large language models trained on code . arXiv preprint arXiv:2107.03374

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  13. [13]

    Xinyun Chen, Chen Liang, Adams Wei Yu, Denny Zhou, Dawn Song, and Quoc V. Le. 2019. https://openreview.net/forum?id=ryxjnREFwH Neural symbolic reader: Scalable integration of distributed and symbolic representations for reading comprehension . ICLR

    work page 2019

  14. [14]

    Ting-Rui Chiang and Yun-Nung Chen. 2019. https://doi.org/10.18653/v1/N19-1272 Semantically-aligned equation generation for solving and reasoning math word problems . In Proceedings of the 2019 Conference of the North A merican Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) , pages 26...

    work page doi:10.18653/v1/n19-1272 2019

  15. [15]

    Peter Clark, Oyvind Tafjord, and Kyle Richardson. 2020. https://www.ijcai.org/proceedings/2020/0537.pdf Transformers as soft reasoners over language . IJCAI

    work page 2020

  16. [16]

    Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. 2021. https://arxiv.org/abs/2110.14168 Training verifiers to solve math word problems . arXiv preprint arXiv:2110.14168

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  17. [17]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. https://aclanthology.org/N19-1423 BERT : Pre-training of deep bidirectional transformers for language understanding . NAACL

    work page 2019

  18. [18]

    Honghua Dong, Jiayuan Mao, Tian Lin, Chong Wang, Lihong Li, and Denny Zhou. 2019. https://arxiv.org/abs/1904.11694 Neural logic machines . ICLR

    work page Pith review arXiv 2019

  19. [19]

    Dheeru Dua, Sameer Singh, and Matt Gardner. 2020. https://aclanthology.org/2020.acl-main.497 Benefits of intermediate annotations in reading comprehension . ACL

    work page 2020

  20. [20]

    Mor Geva, Daniel Khashabi, Elad Segal, Tushar Khot, Dan Roth, and Jonathan Berant. 2021. https://doi.org/10.1162/tacl_a_00370 Did aristotle use a laptop? A question answering benchmark with implicit reasoning strategies . TACL

    work page doi:10.1162/tacl_a_00370 2021

  21. [21]

    Yuling Gu, Bhavana Dalvi Mishra, and Peter Clark. 2022. https://arxiv.org/pdf/2112.08656.pdf DREAM : Uncovering mental models behind language models . NAACL

    work page arXiv 2022

  22. [22]

    Braden Hancock, Paroma Varma, Stephanie Wang, Martin Bringmann, Percy Liang, and Christopher R \'e . 2018. https://doi.org/10.18653/v1/P18-1175 Training classifiers with natural language explanations . ACL

    work page doi:10.18653/v1/p18-1175 2018

  23. [23]

    Peter Hase and Mohit Bansal. 2022. https://arxiv.org/abs/2102.02201 When can models learn from explanations? a formal framework for understanding the roles of explanation data . ACL

    work page arXiv 2022

  24. [24]

    Dan Hendrycks, Collin Burns, Saurav Kadavath, Akul Arora, Steven Basart, Eric Tang, Dawn Song, and Jacob Steinhardt. 2021. https://arxiv.org/abs/2103.03874 Measuring mathematical problem solving with the math dataset . arXiv preprint arXiv:2103.03874

    work page internal anchor Pith review arXiv 2021

  25. [25]

    Mohammad Javad Hosseini, Hannaneh Hajishirzi, Oren Etzioni, and Nate Kushman. 2014. https://doi.org/10.3115/v1/D14-1058 Learning to solve arithmetic word problems with verb categorization . EMNLP

    work page doi:10.3115/v1/d14-1058 2014

  26. [26]

    Zhanming Jie, Jierui Li, and Wei Lu. 2022. https://arxiv.org/abs/2203.10316 Learning to reason deductively: Math word problem solving as complex relation extraction . arXiv preprint arXiv:2203.10316

    work page arXiv 2022

  27. [27]

    Jared Kaplan, Sam McCandlish, Tom Henighan, Tom B Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford, Jeffrey Wu, and Dario Amodei. 2020. https://arxiv.org/abs/2001.08361 Scaling laws for neural language models . arXiv preprint arXiv:2001.08361

    work page internal anchor Pith review Pith/arXiv arXiv 2020

  28. [28]

    Rik Koncel-Kedziorski, Subhro Roy, Aida Amini, Nate Kushman, and Hannaneh Hajishirzi. 2016. https://doi.org/10.18653/v1/N16-1136 MAWPS : A math word problem repository . NAACL

    work page doi:10.18653/v1/n16-1136 2016

  29. [29]

    2022 , month = may, number =

    Andrew K. Lampinen, Ishita Dasgupta, Stephanie C.Y. Chan, Kory Matthewson, Michael Henry Tessler, Antonia Creswell, James L. McClelland, Jane X. Wang, and Felix Hill. 2022. https://arxiv.org/abs/2204.02329 Can language models learn from explanations in context? arXiv preprint arXiv:2204.02329

    work page arXiv 2022

  30. [30]

    Yihuai Lan, Lei Wang, Qiyuan Zhang, Yunshi Lan, Bing Tian Dai, Yan Wang, Dongxiang Zhang, and Ee-Peng Lim. 2021. https://arxiv.org/abs/2109.00799 MWPT oolkit: An open-source framework for deep learning-based math word problem solvers . arXiv preprint arXiv:2109.00799

    work page arXiv 2021

  31. [31]

    Teven Le Scao and Alexander Rush. 2021. https://doi.org/10.18653/v1/2021.naacl-main.208 How many data points is a prompt worth? NAACL

    work page doi:10.18653/v1/2021.naacl-main.208 2021

  32. [32]

    Brian Lester, Rami Al-Rfou, and Noah Constant. 2021. https://doi.org/10.18653/v1/2021.emnlp-main.243 The power of scale for parameter-efficient prompt tuning . EMNLP

    work page doi:10.18653/v1/2021.emnlp-main.243 2021

  33. [33]

    Iddo Lev, Bill MacCartney, Christopher Manning, and Roger Levy. 2004. https://aclanthology.org/W04-0902 Solving logic puzzles: From robust processing to precise semantics . Proceedings of the 2nd Workshop on Text Meaning and Interpretation

    work page 2004

  34. [34]

    Xiang Lisa Li and Percy Liang. 2021. https://doi.org/10.18653/v1/2021.acl-long.353 Prefix-tuning: Optimizing continuous prompts for generation . ACL

    work page doi:10.18653/v1/2021.acl-long.353 2021

  35. [35]

    Zhengzhong Liang, Steven Bethard, and Mihai Surdeanu. 2021. https://doi.org/10.18653/v1/2021.naacl-main.97 Explainable multi-hop verbal reasoning through internal monologue . NAACL

    work page doi:10.18653/v1/2021.naacl-main.97 2021

  36. [36]

    Wang Ling, Dani Yogatama, Chris Dyer, and Phil Blunsom. 2017. https://doi.org/10.18653/v1/P17-1015 Program induction by rationale generation: Learning to solve and explain algebraic word problems . ACL

    work page doi:10.18653/v1/p17-1015 2017

  37. [37]

    Pengfei Liu, Weizhe Yuan, Jinlan Fu, Zhengbao Jiang, Hiroaki Hayashi, and Graham Neubig. 2021. https://arxiv.org/abs/2107.13586 Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing . arXiv preprint arXiv:2107.13586

    work page arXiv 2021

  38. [38]

    Bodhisattwa Prasad Majumder, Oana-Maria Camburu, Thomas Lukasiewicz, and Julian McAuley. 2021. https://arxiv.org/abs/2106.13876 Rationale-inspired natural language explanations with commonsense . arXiv preprint arXiv:2106.13876

    work page arXiv 2021

  39. [39]

    Ana Marasovi \'c , Iz Beltagy, Doug Downey, and Matthew E Peters. 2022. http://arxiv.org/abs/2111.08284 Few-shot self-rationalization with natural language prompts . NAACL Findings

    work page arXiv 2022

  40. [40]

    Joshua Maynez, Shashi Narayan, Bernd Bohnet, and Ryan McDonald. 2020. https://doi.org/10.18653/v1/2020.acl-main.173 On faithfulness and factuality in abstractive summarization . In ACL

    work page doi:10.18653/v1/2020.acl-main.173 2020

  41. [41]

    Shen Yun Miao, Chao Chun Liang, and Keh Yih Su. 2020. https://doi.org/10.18653/v1/2020.acl-main.92 A diverse corpus for evaluating and developing E nglish math word problem solvers . ACL

    work page doi:10.18653/v1/2020.acl-main.92 2020

  42. [42]

    Sewon Min, Xinxi Lyu, Ari Holtzman, Mikel Artetxe, Mike Lewis, Hannaneh Hajishirzi, and Luke Zettlemoyer. 2022. https://arxiv.org/abs/2202.12837 Rethinking the role of demonstrations: What makes in-context learning work? arXiv preprint arXiv:2202.12837

    work page internal anchor Pith review arXiv 2022

  43. [43]

    Sharan Narang, Colin Raffel, Katherine Lee, Adam Roberts, Noah Fiedel, and Karishma Malkan. 2020. https://arxiv.org/abs/2004.14546 WT5 ?! T raining text-to-text models to explain their predictions . arXiv preprint arXiv:2004.14546

    work page arXiv 2020

  44. [44]

    Maxwell Nye, Anders Johan Andreassen, Guy Gur-Ari, Henryk Michalewski, Jacob Austin, David Bieber, David Dohan, Aitor Lewkowycz, Maarten Bosma, David Luan, et al. 2021. http://arxiv.org/abs/2112.00114 Show your work: Scratchpads for intermediate computation with language models . arXiv preprint arXiv:2112.00114

    work page internal anchor Pith review arXiv 2021

  45. [45]

    Training language models to follow instructions with human feedback

    Long Ouyang, Jeff Wu, Xu Jiang, Diogo Almeida, Carroll L. Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. 2022. https://arxiv.org/abs/2203.02155 Training language models to follow instructions with human feedback . arXiv preprint arXiv:2203.02155

    work page internal anchor Pith review arXiv 2022

  46. [46]

    Arkil Patel, Satwik Bhattamishra, and Navin Goyal. 2021. https://aclanthology.org/2021.naacl-main.168.pdf Are NLP models really able to solve simple math word problems? NAACL

    work page 2021

  47. [47]

    Peters, Mark Neumann, Mohit Iyyer, Matt Gardner, Christopher Clark, Kenton Lee, and Luke Zettlemoyer

    Matthew E. Peters, Mark Neumann, Mohit Iyyer, Matt Gardner, Christopher Clark, Kenton Lee, and Luke Zettlemoyer. 2018. https://aclanthology.org/N18-1202 Deep contextualized word representations . NAACL

    work page 2018

  48. [48]

    Xinyu Pi, Qian Liu, Bei Chen, Morteza Ziyadi, Zeqi Lin, Yan Gao, Qiang Fu, Jian-Guang Lou, and Weizhu Chen. 2022. https://arxiv.org/abs/2201.11473 Reasoning like program executors . arXiv preprint arXiv:2201.11473

    work page arXiv 2022

  49. [49]

    Piotr Pi e kos, Mateusz Malinowski, and Henryk Michalewski. 2021. https://doi.org/10.18653/v1/2021.acl-short.49 Measuring and improving BERT ' s mathematical abilities by predicting the order of reasoning. ACL

    work page doi:10.18653/v1/2021.acl-short.49 2021

  50. [50]

    Scaling Language Models: Methods, Analysis & Insights from Training Gopher

    Jack W. Rae, Sebastian Borgeaud, Trevor Cai, Katie Millican, Jordan Hoffmann, Francis Song, John Aslanides, Sarah Henderson, Roman Ring, Susannah Young, et al. 2021. https://arxiv.org/abs/2112.11446 Scaling language models: Methods, analysis & insights from training G opher . arXiv preprint arXiv:2112.11446

    work page internal anchor Pith review arXiv 2021

  51. [51]

    Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J Liu. 2020. https://arxiv.org/abs/1910.10683 Exploring the limits of transfer learning with a unified text-to-text transformer . Journal of Machine Learning Research, 21:1--67

    work page internal anchor Pith review arXiv 2020

  52. [52]

    and Tsvetkov, Yulia

    Dheeraj Rajagopal, Vidhisha Balachandran, Eduard H. Hovy, and Yulia Tsvetkov. 2021. https://doi.org/10.18653/v1/2021.emnlp-main.64 SelfExplain : A self-explaining architecture for neural text classifiers . EMNLP

    work page doi:10.18653/v1/2021.emnlp-main.64 2021

  53. [53]

    Nazneen Fatema Rajani, Bryan McCann, Caiming Xiong, and Richard Socher. 2019. https://doi.org/10.18653/v1/P19-1487 Explain yourself! L everaging language models for commonsense reasoning . ACL

    work page doi:10.18653/v1/p19-1487 2019

  54. [54]

    Qiu Ran, Yankai Lin, Peng Li, Jie Zhou, and Zhiyuan Liu. 2019. https://doi.org/10.18653/v1/D19-1251 N um N et: Machine reading comprehension with numerical reasoning . EMNLP

    work page doi:10.18653/v1/d19-1251 2019

  55. [55]

    Hannah Rashkin, Vitaly Nikolaev, Matthew Lamm, Michael Collins, Dipanjan Das, Slav Petrov, Gaurav Singh Tomar, Iulia Turc, and David Reitter. 2021. https://arxiv.org/abs/2112.12870 Measuring attribution in natural language generation models . arXiv preprint arXiv:2112.12870

    work page arXiv 2021

  56. [56]

    Gabriel Recchia. 2021. https://arxiv.org/abs/2109.02102 Teaching autoregressive language models complex tasks by demonstration . arXiv preprint arXiv:2109.02102

    work page arXiv 2021

  57. [57]

    Emily Reif, Daphne Ippolito, Ann Yuan, Andy Coenen, Chris Callison-Burch, and Jason Wei. 2022. https://arxiv.org/abs/2109.03910 A recipe for arbitrary text style transfer with large language models . ACL

    work page arXiv 2022

  58. [58]

    Laria Reynolds and Kyle McDonell. 2021. https://arxiv.org/abs/2102.07350 Prompt programming for large language models: Beyond the few-shot paradigm . Extended Abstracts of the 2021 CHI Conference on Human Factors in Computing Systems

    work page arXiv 2021

  59. [59]

    Subhro Roy and Dan Roth. 2015. https://doi.org/10.18653/v1/D15-1202 Solving general arithmetic word problems . EMNLP

    work page doi:10.18653/v1/d15-1202 2015

  60. [60]

    Subhro Roy, Tim Vieira, and Dan Roth. 2015. https://doi.org/10.1162/tacl_a_00118 Reasoning about Quantities in Natural Language . TACL

    work page doi:10.1162/tacl_a_00118 2015

  61. [61]

    Mohammed Saeed, Naser Ahmadi, Preslav Nakov, and Paolo Papotti. 2021. https://doi.org/10.18653/v1/2021.emnlp-main.110 R ule BERT : Teaching soft rules to pre-trained language models . EMNLP

    work page doi:10.18653/v1/2021.emnlp-main.110 2021

  62. [62]

    Multitask Prompted Training Enables Zero-Shot Task Generalization

    Victor Sanh, Albert Webson, Colin Raffel, Stephen H. Bach, Lintang Sutawika, Zaid Alyafeai, Antoine Chaffin, Arnaud Stiegler, Teven Le Scao, Arun Raja, et al. 2022. https://arxiv.org/abs/2110.08207 Multitask prompted training enables zero-shot task generalization . ICLR

    work page internal anchor Pith review arXiv 2022

  63. [63]

    Jianhao Shen, Yichun Yin, Lin Li, Lifeng Shang, Xin Jiang, Ming Zhang, and Qun Liu. 2021. https://aclanthology.org/2021.findings-emnlp.195 Generate & rank: A multi-task framework for math word problems . In Findings of the Association for Computational Linguistics: EMNLP 2021

    work page 2021

  64. [64]

    Alon Talmor, Jonathan Herzig, Nicholas Lourie, and Jonathan Berant. 2019. https://doi.org/10.18653/v1/N19-1421 C ommonsense QA : A question answering challenge targeting commonsense knowledge . NAACL

    work page doi:10.18653/v1/n19-1421 2019

  65. [65]

    Alon Talmor, Oyvind Tafjord, Peter Clark, Yoav Goldberg, and Jonathan Berant. 2020. https://arxiv.org/abs/2006.06609 Leap-of-thought: Teaching pre-trained models to systematically reason over implicit knowledge . NeurIPS

    work page arXiv 2020

  66. [66]

    Alon Talmor, Ori Yoran, Ronan Le Bras, Chandra Bhagavatula, Yoav Goldberg, Yejin Choi, and Jonathan Berant. 2021. https://arxiv.org/abs/2201.05320 Commonsense QA 2.0: E xposing the limits of ai through gamification . NeurIPS Track on Datasets and Benchmarks

    work page arXiv 2021

  67. [67]

    Yi Tay, Mostafa Dehghani, Vinh Q Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, and Donald Metzler. 2022. https://arxiv.org/abs/2205.05131 Unifying language learning paradigms . arXiv preprint arXiv:2205.05131

    work page arXiv 2022

  68. [68]

    Romal Thoppilan, Daniel De Freitas, Jamie Hall, Noam Shazeer, Apoorv Kulshreshtha, Heng-Tze Cheng, Alicia Jin, Taylor Bos, Leslie Baker, Yu Du, et al. 2022. https://arxiv.org/abs/2201.08239 La MDA : Language models for dialog applications . arXiv preprint arXiv:2201.08239

    work page Pith review arXiv 2022

  69. [69]

    Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc Le, Ed Chi, and Denny Zhou. 2022 a . https://arxiv.org/abs/2203.11171 Self-consistency improves chain of thought reasoning in language models . arXiv preprint arXiv:2203.11171

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  70. [70]

    Yizhong Wang, Swaroop Mishra, Pegah Alipoormolabashi, Yeganeh Kordi, Amirreza Mirzaei, Anjana Arunkumar, Arjun Ashok, Arut Selvan Dhanasekaran, Atharva Naik, David Stap, et al. 2022 b . https://arxiv.org/abs/2204.07705 Benchmarking generalization via in-context instructions on 1,600+ language tasks . arXiv preprint arXiv:2204.07705

    work page arXiv 2022

  71. [71]

    Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M

    Jason Wei, Maarten Bosma, Vincent Y. Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M. Dai, and Quoc V. Le. 2022 a . https://openreview.net/forum?id=gEZrGCozdqR Finetuned language models are zero-shot learners . ICLR

    work page 2022

  72. [72]

    Jason Wei, Yi Tay, Rishi Bommasani, Colin Raffel, Barret Zoph, Sebastian Borgeaud, Dani Yogatama, Maarten Bosma, Denny Zhou, Donald Metzler, et al. 2022 b . https://openreview.net/forum?id=yzkSU5zdwD Emergent abilities of large language models . Transactions on Machine Learning Research

    work page 2022

  73. [73]

    Sarah Wiegreffe, Jack Hessel, Swabha Swayamdipta, Mark Riedl, and Yejin Choi. 2022. https://arxiv.org/abs/2112.08674 Reframing human- AI collaboration for generating free-text explanations . NAACL

    work page arXiv 2022

  74. [74]

    Sarah Wiegreffe and Ana Marasovi \'c . 2021. https://arxiv.org/abs/2102.12060 Teach me to explain: A review of datasets for explainable NLP . NeurIPS

    work page arXiv 2021

  75. [75]

    Sarah Wiegreffe, Ana Marasovi \'c , and Noah A. Smith. 2021. https://aclanthology.org/2021.emnlp-main.804 M easuring association between labels and free-text rationales . EMNLP

    work page 2021

  76. [76]

    Tongshuang Wu, Ellen Jiang, Aaron Donsbach, Jeff Gray, Alejandra Molina, Michael Terry, and Carrie J Cai. 2022 a . https://dl.acm.org/doi/abs/10.1145/3491101.3519729 Prompt C hainer: Chaining large language model prompts through visual programming . CHI Extended Abstracts

    work page doi:10.1145/3491101.3519729 2022

  77. [77]

    Tongshuang Wu, Michael Terry, and Carrie Jun Cai. 2022 b . https://dl.acm.org/doi/abs/10.1145/3491102.3517582 A I chains: Transparent and controllable human- AI interaction by chaining large language model prompts . CHI

    work page doi:10.1145/3491102.3517582 2022

  78. [78]

    Yujun Yan, Kevin Swersky, Danai Koutra, Parthasarathy Ranganathan, and Milad Hashemi. 2020. https://arxiv.org/abs/2006.08084 Neural execution engines: Learning to execute subroutines . NeurIPS

    work page arXiv 2020

  79. [79]

    Huihan Yao, Ying Chen, Qinyuan Ye, Xisen Jin, and Xiang Ren. 2021. https://proceedings.neurips.cc/paper/2021/hash/4b26dc4663ccf960c8538d595d0a1d3a-Abstract.html Refining language models with compositional explanations . NeurIPS

    work page 2021

  80. [80]

    Xi Ye and Greg Durrett. 2022. https://arxiv.org/abs/2205.03401 The unreliability of explanations in few-shot in-context learning . arXiv preprint arXiv:2205.03401

    work page arXiv 2022

Showing first 80 references.