mlsanityData to 7 Oct 2026

Inverse Scaling: When Bigger Isn't Better

arXiv 2306.09479 ↗First posted 15 Jun 2023cs.CL cs.AI cs.CYcited by 205

Not on the map. Why?

Cited by 35 in LLM reasoning, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Work on scaling laws has found that large language models (LMs) show predictable improvements to overall loss with increased scale (model size, training data, and compute). Here, we present evidence for the claim that LMs may show inverse scaling, or worse task performance with increased scale, e.g., due to flaws in the training objective and data. We present empirical evidence of inverse scaling on 11 datasets collected by running a public contest, the Inverse Scaling Prize, with a substantial prize pool. Through analysis of the datasets, along with other examples found in the literature, we identify four potential causes of inverse scaling: (i) preference to repeat memorized sequences over following in-context instructions, (ii) imitation of undesirable patterns in the training data, (iii) tasks containing an easy distractor task which LMs could focus on, rather than the harder real task, and (iv) correct but misleading few-shot demonstrations of the task. We release the winning datasets at https://inversescaling.com/data to allow for further investigation of inverse scaling. Our tasks have helped drive the discovery of U-shaped and inverted-U scaling trends, where an initial trend reverses, suggesting that scaling trends are less reliable at predicting the behavior of larger-scale models than previously understood. Overall, our results suggest that there are tasks for which increased model scale alone may not lead to progress, and that more careful thought needs to go into the data and objectives for training language models.

Key facts. Cited by 205 papers in the map (data through 7 Oct 2026). Builds on 46 of its 68 references that are in the map. First posted on arXiv 15 Jun 2023.

Builds on

46 of 68 references.

References · 46

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2020Language Models are Few-Shot Learnerscited by 32,199
  3. 2022Training language models to follow instructions with human feedbackcited by 15,712
  4. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,302
  5. 2021Evaluating Large Language Models Trained on Codecited by 8,053
  6. 2020Measuring Massive Multitask Language Understandingcited by 7,768
  7. 2020Scaling Laws for Neural Language Modelscited by 6,411
  8. 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,304
  9. 2018GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understandingcited by 4,854
  10. 2021On the Opportunities and Risks of Foundation Modelscited by 3,961
  11. 2017Deep reinforcement learning from human preferencescited by 3,846
  12. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,634
34 more
  1. 2022OPT: Open Pre-trained Transformer Language Modelscited by 3,464
  2. 2022Training Compute-Optimal Large Language Modelscited by 2,927
  3. 2021TruthfulQA: Measuring How Models Mimic Human Falsehoodscited by 2,919
  4. 2020Learning to summarize from human feedbackcited by 2,562
  5. 2022Emergent Abilities of Large Language Modelscited by 2,483
  6. 2018Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answeringcited by 2,114
  7. 2022Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language modelscited by 1,970
  8. 2019SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systemscited by 1,718
  9. 2021Scaling Language Models: Methods, Analysis & Insights from Training Gophercited by 1,199
  10. 2022Scaling Laws for Reward Model Overoptimizationcited by 1,056
  11. 2021A General Language Assistant as a Laboratory for Alignmentcited by 1,023
  12. 2016The LAMBADA dataset: Word prediction requiring a broad discourse contextcited by 960
  13. 2022In-context Learning and Induction Headscited by 930
  14. 2021Show Your Work: Scratchpads for Intermediate Computation with Language Modelscited by 871
  15. 2022Discovering Language Model Behaviors with Model-Written Evaluationscited by 805
  16. 2022Quantifying Memorization Across Neural Language Modelscited by 755
  17. 2021BBQ: A Hand-Built Bias Benchmark for Question Answeringcited by 683
  18. 2021True Few-Shot Learning with Language Modelscited by 343
  19. 2022Deduplicating Training Data Mitigates Privacy Risks in Language Modelscited by 309
  20. 2019Risks from Learned Optimization in Advanced Machine Learning Systemscited by 250
  21. 2022The Alignment Problem from a Deep Learning Perspectivecited by 246
  22. 2023Pretraining Language Models with Human Preferencescited by 245
  23. 2022Predictability and Surprise in Large Generative Modelscited by 209
  24. 2021Process for Adapting Language Models to Society (PALMS) with Values-Targeted Datasetscited by 179
  25. 2023The Capacity for Moral Self-Correction in Large Language Modelscited by 171
  26. 2021Alignment of Language Agentscited by 167
  27. 2022Goal Misgeneralization: Why Correct Specifications Aren't Enough For Correct Goalscited by 120
  28. 2019Counterfactual Story Reasoning and Generationcited by 87
  29. 2021Identifying and Mitigating Spurious Correlations for Improving Robustness in NLP Modelscited by 76
  30. 2022Inverse scaling can become U-shapedcited by 76
  31. 2023The Larger They Are, the Harder They Fail: Language Models do not Recognize Identifier Swaps in Pythoncited by 29
  32. 2022Uncontrolled Lexical Exposure Leads to Overestimation of Compositional Generalization in Pretrained Modelscited by 26
  33. 2022COMPS: Conceptual Minimal Pair Sentences for testing Robust Property Knowledge and its Inheritance in Pre-trained Language Modelscited by 15
  34. 2023Beyond Positive Scaling: How Negation Impacts Scaling Trends of Language Modelscited by 12

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 33; 2024: 57; 2025: 62; 2026: 53.

205 papers.

Sort:

By area:

  1. 2023Measuring Faithfulness in Chain-of-Thought Reasoningcited by 668
  2. 2023Weak-to-Strong Generalization: Eliciting Strong Capabilities With Weak Supervisioncited by 445
  3. 2025Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscationcited by 338
  4. 2023Jailbreak in pieces: Compositional Adversarial Attacks on Multi-Modal Language Modelscited by 260
  5. 2023Challenges and Applications of Large Language Modelscited by 251
  6. 2023Reasoning or Reciting? Exploring the Capabilities and Limitations of Language Models Through Counterfactual Taskscited by 243
  7. 2023Do LLMs exhibit human-like response biases? A case study in survey designcited by 176
  8. 2023Embers of Autoregression: Understanding Large Language Models Through the Problem They are Trained to Solvecited by 175
  9. 2024CycleResearcher: Improving Automated Research via Automated Reviewcited by 142
  10. 2024AgentDojo: A Dynamic Environment to Evaluate Prompt Injection Attacks and Defenses for LLM Agentscited by 129
  11. 2024A Tale of Tails: Model Collapse as a Change of Scaling Lawscited by 127
  12. 2025BIG-Bench Extra Hardcited by 123
38 more
  1. 2023Frontier AI Regulation: Managing Emerging Risks to Public Safetycited by 95
  2. 2024The Task Shield: Enforcing Task Alignment to Defend Against Indirect Prompt Injection in LLM Agentscited by 89
  3. 2025Findings of the BabyLM Challenge: Sample-Efficient Pretraining on Developmentally Plausible Corporacited by 86
  4. 2024Beyond Accuracy: Evaluating the Reasoning Behavior of Large Language Models -- A Surveycited by 82
  5. 2023In-context Pretraining: Language Modeling Beyond Document Boundariescited by 76
  6. 2026Unifying Group-Relative and Self-Distillation Policy Optimization via Sample Routingcited by 65
  7. 2025MCPTox: A Benchmark for Tool Poisoning Attack on Real-World MCP Serverscited by 63
  8. 2025The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systemscited by 57
  9. 2024Safetywashing: Do AI Safety Benchmarks Actually Measure Safety Progress?cited by 56
  10. 2023Paloma: A Benchmark for Evaluating Language Model Fitcited by 53
  11. 2023Value FULCRA: Mapping Large Language Models to the Multidimensional Spectrum of Basic Human Valuescited by 49
  12. 2024Scaling Laws for Downstream Task Performance of Large Language Modelscited by 49
  13. 2025Inverse Scaling in Test-Time Computecited by 49
  14. 2023Managing extreme AI risks amid rapid progresscited by 48
  15. 2023Hoodwinked: Deception and Cooperation in a Text-Based Game for Language Modelscited by 46
  16. 2024ALI-Agent: Assessing LLMs' Alignment with Human Values via Agent-based Evaluationcited by 46
  17. 2023From Instructions to Intrinsic Human Values -- A Survey of Alignment Goals for Big Modelscited by 45
  18. 2024Evaluating Mathematical Reasoning of Large Language Models: A Focus on Error Identification and Correctioncited by 45
  19. 2023MoCa: Measuring Human-Language Model Alignment on Causal and Moral Judgment Taskscited by 44
  20. 2024CHAMP: A Competition-level Dataset for Fine-Grained Analyses of LLMs' Mathematical Reasoning Capabilitiescited by 43
  21. 2025How Do Large Language Monkeys Get Their Power (Laws)?cited by 40
  22. 2024SaulLM-54B & SaulLM-141B: Scaling Up Domain Adaptation for the Legal Domaincited by 39
  23. 2025MCP Security Bench (MSB): Benchmarking Attacks Against Model Context Protocol in LLM Agentscited by 37
  24. 2023SoK: Memorization in General-Purpose Large Language Modelscited by 36
  25. 2023The Consensus Game: Language Model Generation via Equilibrium Searchcited by 35
  26. 2024Vibe-Eval: A hard evaluation suite for measuring progress of multimodal language modelscited by 35
  27. 2024Style Outweighs Substance: Failure Modes of LLM Judges in Alignment Benchmarkingcited by 35
  28. 2026When Single-Agent with Skills Replace Multi-Agent Systems and When They Failcited by 34
  29. 2024Fewer Truncations Improve Language Modelingcited by 33
  30. 2024Benchmarks as Microscopes: A Call for Model Metrologycited by 33
  31. 2024On the Relationship between Truth and Political Bias in Language Modelscited by 31
  32. 2025Scaling Laws Are Unreliable for Downstream Tasks: A Reality Checkcited by 27
  33. 2023Denevil: Towards Deciphering and Navigating the Ethical Values of Large Language Models via Instruction Learningcited by 26
  34. 2024Are More LLM Calls All You Need? Towards Scaling Laws of Compound Inference Systemscited by 24
  35. 2024On Behalf of the Stakeholders: Trends in NLP Model Interpretability in the Era of LLMscited by 24
  36. 2023When can transformers reason with abstract symbols?cited by 23
  37. 2024Astraios: Parameter-Efficient Instruction Tuning Code Large Language Modelscited by 23
  38. 2024CLAVE: An Adaptive Framework for Evaluating Values of LLM Generated Responsescited by 23

50 most cited of 205 All 205 (JSON)