Inference Scaling Laws: An Empirical Analysis of Compute-Optimal Inference for Problem-Solving with Language Models

• URL: http://arxiv.org/abs/2408.00724v2
• Date: Mon, 14 Oct 2024 13:41:35 GMT
• Title: Inference Scaling Laws: An Empirical Analysis of Compute-Optimal Inference for Problem-Solving with Language Models
• Authors: Yangzhen Wu, Zhiqing Sun, Shanda Li, Sean Welleck, Yiming Yang,
• Abstract summary: We study inference scaling laws and compute-optimal inference for large language models (LLMs) training. As a first step towards understanding and designing compute-optimal inference methods, we studied cost-performance trade-offs for inference strategies. Our findings indicate smaller models (e.g., Llemma-7B) can outperform larger models given the same computation budgets.
• Score: 46.959380978972206
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: While the scaling laws of large language models (LLMs) training have been extensively studied, optimal inference configurations of LLMs remain underexplored. We study inference scaling laws and compute-optimal inference, focusing on the trade-offs between model sizes and generating additional tokens with different inference strategies. As a first step towards understanding and designing compute-optimal inference methods, we studied cost-performance trade-offs for inference strategies such as greedy search, majority voting, best-of-$n$, weighted voting, and two different tree search algorithms, using different model sizes and compute budgets. Our findings indicate smaller models (e.g., Llemma-7B) can outperform larger models given the same computation budgets, and that smaller models paired with advanced inference algorithms yield Pareto-optimal cost-performance trade-offs. For instance, the Llemma-7B model, equipped with our novel tree search algorithm, consistently outperforms Llemma-34B with standard majority voting on the MATH benchmark across all FLOPs budgets. We hope these findings contribute to a broader understanding of inference scaling laws for LLMs.

Related papers

• EVOLvE: Evaluating and Optimizing LLMs For Exploration [76.66831821738927]
  Large language models (LLMs) remain under-studied in scenarios requiring optimal decision-making under uncertainty. We measure LLMs' (in)ability to make optimal decisions in bandits, a state-less reinforcement learning setting relevant to many applications. Motivated by the existence of optimal exploration algorithms, we propose efficient ways to integrate this algorithmic knowledge into LLMs.
  arXiv  Detail & Related papers  (2024-10-08T17:54:03Z)
• LLaMA-Berry: Pairwise Optimization for O1-like Olympiad-Level Mathematical Reasoning [56.273799410256075]
  The framework combines Monte Carlo Tree Search (MCTS) with iterative Self-Refine to optimize the reasoning path. The framework has been tested on general and advanced benchmarks, showing superior performance in terms of search efficiency and problem-solving capability.
  arXiv  Detail & Related papers  (2024-10-03T18:12:29Z)
• Are Protein Language Models Compute Optimal? [0.0]
  We investigate the optimal ratio between model parameters and training tokens within a fixed compute budget. Our study reveals that pLM sizes scale sublinearly with compute budget, showing diminishing returns in performance as model size increases. This work paves the way towards more compute-efficient pLMs, democratizing their training and practical application in computational biology.
  arXiv  Detail & Related papers  (2024-06-11T13:32:11Z)
• MindStar: Enhancing Math Reasoning in Pre-trained LLMs at Inference Time [51.5039731721706]
  MindStar is a purely inference-based searching method for large language models. It formulates reasoning tasks as searching problems and proposes two search ideas to identify the optimal reasoning paths. It significantly enhances the reasoning abilities of open-source models, such as Llama-2-13B and Mistral-7B, and achieves comparable performance to GPT-3.5 and Grok-1.
  arXiv  Detail & Related papers  (2024-05-25T15:07:33Z)
• Optimising Calls to Large Language Models with Uncertainty-Based Two-Tier Selection [80.63946798650653]
  Decision centers on whether to use a large LLM with better performance or a smaller one with reduced costs. We propose a simpler solution; we use only the uncertainty of the generations of the small LLM as the decision criterion. Our experiments reveal this simple solution optimally balances cost and performance, outperforming existing methods on 25 out of 27 experimental setups.
  arXiv  Detail & Related papers  (2024-05-03T14:38:59Z)
• $\mathbf{(N,K)}$-Puzzle: A Cost-Efficient Testbed for Benchmarking Reinforcement Learning Algorithms in Generative Language Model [50.636423457653066]
  We present a generalized version of the 24-Puzzle: the $(N,K)$-Puzzle, which challenges language models to reach a target value $K$ with $N$ integers. We evaluate the effectiveness of established RL algorithms such as Proximal Policy Optimization (PPO), alongside novel approaches like Identity Policy Optimization (IPO) and Direct Policy Optimization (DPO)
  arXiv  Detail & Related papers  (2024-03-11T22:24:14Z)
• Proof of Swarm Based Ensemble Learning for Federated Learning Applications [3.2536767864585663]
  In federated learning it is not feasible to apply centralised ensemble learning directly due to privacy concerns. Most distributed consensus algorithms, such as Byzantine fault tolerance (BFT), do not normally perform well in such applications. We propose PoSw, a novel distributed consensus algorithm for ensemble learning in a federated setting.
  arXiv  Detail & Related papers  (2022-12-28T13:53:34Z)
• Optimizing Binary Decision Diagrams with MaxSAT for classification [3.2894524838755608]
  A growing interest in explainable artificial intelligence motivates the need for interpretable machine learning (ML) models. Recently, several exact methods for computing such models are proposed to overcome weaknesses of traditional methods. In this paper, we first propose SAT-based models for learning optimal Binary decision diagrams (BDDs) Then, we lift the encoding to a MaxSAT model to learn optimal BDDs in limited depths. Finally, we tackle the fragmentation problem by introducing a method to merge compatible subtrees for the BDDs found via the MaxSAT model.
  arXiv  Detail & Related papers  (2022-03-21T23:17:37Z)
• Pretrained Cost Model for Distributed Constraint Optimization Problems [37.79733538931925]
  Distributed Constraint Optimization Problems (DCOPs) are an important subclass of optimization problems. We propose a novel directed acyclic graph schema representation for DCOPs and leverage the Graph Attention Networks (GATs) to embed graph representations. Our model, GAT-PCM, is then pretrained with optimally labelled data in an offline manner, so as to boost a broad range of DCOP algorithms.
  arXiv  Detail & Related papers  (2021-12-08T09:24:10Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.