[2405.10938] Observational Scaling Laws and the Predictability of Language Model Performance

Abstract:Understanding how language model performance varies with scale is critical to benchmark and algorithm development. Scaling laws are one approach to building this understanding, but the requirement of training models across many different scales has limited their use. We propose an alternative, observational approach that bypasses model training and instead builds scaling laws from ~100 publically available models. Building a single scaling law from multiple model families is challenging due to large variations in their training compute efficiencies and capabilities. However, we show that these variations are consistent with a simple, generalized scaling law where language model performance is a function of a low-dimensional capability space, and model families only vary in their efficiency in converting training compute to capabilities. Using this approach, we show the surprising predictability of complex scaling phenomena: we show that several emergent phenomena follow a smooth, sigmoidal behavior and are predictable from small models; we show that the agent performance of models such as GPT-4 can be precisely predicted from simpler non-agentic benchmarks; and we show how to predict the impact of post-training interventions like Chain-of-Thought and Self-Consistency as language model capabilities continue to improve.

[2405.10938] Observational Scaling Laws and the Predictability of Language Model Performance Skip to main content arXiv is now an independent nonprofit! Learn more × Search arXiv Press Enter to search · Advanced search --> Computer Science > Machine Learning arXiv:2405.10938 (cs) [Submitted on 17 May 2024 ( v1 ), last revised 1 Oct 2024 (this version, v3)] Title: Observational Scaling Laws and the Predictability of Language Model Performance Authors: Yangjun Ruan , Chris J. Maddison , Tatsunori Hashimoto View a PDF of the paper titled Observational Scaling Laws and the Predictabil

related reading

• [2510.13786] The Art of Scaling Reinforcement Learning Compute for LLMsarxiv.org
• [2305.18290] Direct Preference Optimization: Your Language Model is Secretly a Reward Modelarxiv.org
• [2309.10668] Language Modeling Is Compressionarxiv.org
• arxiv.org/pdf/2505.24832arxiv.org
• [2503.10965] Auditing language models for hidden objectivesarxiv.org
• [2605.12715] Scaling Laws for Mixture Pretraining Under Data Constraintsarxiv.org
• Discovering Language Model Behaviors with Model-Written Evaluations — LessWronglesswrong.com
• [2503.04715] Predictable Scale: Part I, Step Law -- Optimal Hyperparameter Scaling Law in Large Language Model Pretrainingarxiv.org
• [2511.08579] Training Language Models to Explain Their Own Computationsarxiv.org
• [2602.05970] Inverse Depth Scaling From Most Layers Being Similararxiv.org
• 2308.03958arxiv.org
• [2509.14786] Pre-training under infinite computearxiv.org