Input-Time Scaling

• URL: http://arxiv.org/abs/2508.13654v4
• Date: Fri, 12 Sep 2025 07:04:59 GMT
• Title: Input-Time Scaling
• Authors: Rapheal Huang, Weilong Guo,
• Abstract summary: Current Large Language Models (LLMs) are usually post-trained on large-scale carefully curated datasets.<n>We present a new scaling paradigm, Input-Time Scaling, to complement previous scaling methods by putting resources on queries.<n>We are also surprised to find that seemingly low data quality datasets can perform better.
• Score: 2.2917707112773598
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Current Large Language Models (LLMs) are usually post-trained on large-scale carefully curated datasets (data & training scaling) and doing reasoning in test time (inference time scaling). In this work, we present a new scaling paradigm, Input-Time Scaling, to complement previous scaling methods by putting resources on queries (input time). During training and testing, we utilize meta-knowledge from LLMs to refine inputs with different strategies. We also discover a new phenomenon, train-test co-design. It requires us to apply query strategies during training and testing as a whole. Only applying strategies on training or testing would seriously degrade the performance gained. We are also surprised to find that seemingly low data quality datasets can perform better. We can get the best performance even by adding irrelevant information to the queries, with randomly selected 1k examples from a minimally filtered dataset. These findings contradict the widely held inductive bias, "garbage in, garbage out". Curating datasets with seemingly high-quality data can even potentially limit the performance ceiling. In addition, models trained on more data with similar quality (15k VS 1k) perform worse, the intuition of simply scaling the size should also be carefully inspected. The good news is that our findings are compatible with the Less is More phenomenon. 1K examples are enough to invoke high-level reasoning ability. With experiments on Qwen2.5-32B-Instruct, we are able to reach SOTA performance among 32B models on AIME24(76.7%) and AIME25(76.7%) pass@1. We can further achieve AIME24(76.7%) and AIME25(80%) with a majority vote of three models. Starting from DeepSeek-R1-Distill-Qwen-32B, the result would be 90.0% on AIME24 and 80.0% on AIME25. To facilitate reproducibility and further research, we are working on open-source our datasets, data pipelines, evaluation results, and checkpoints.

Related papers

• Progressive Data Dropout: An Embarrassingly Simple Approach to Faster Training [26.65053392031144]
  We propose a series of alternative training paradigms that leverage insights from hard-data-mining and dropout.<n>The proposed Progressive Data Dropout reduces the number of effective epochs to as little as 12.4% of the baseline.<n>Surprisingly, the proposed method improves accuracy by up to 4.82%.
  arXiv  Detail & Related papers  (2025-05-28T13:26:52Z)
• DataDecide: How to Predict Best Pretraining Data with Small Experiments [67.95896457895404]
  We release models, data, and evaluations in DataDecide -- the most extensive open suite of models over differences in data and scale.<n>We conduct controlled pretraining experiments across 25 corpora with differing sources, deduplication, and filtering up to 100B tokens, model sizes up to 1B parameters, and 3 random seeds.
  arXiv  Detail & Related papers  (2025-04-15T17:02:15Z)
• LIMR: Less is More for RL Scaling [25.477841726836836]
  We introduce Learning Impact Measurement (LIM), an automated method to evaluate and prioritize training samples.<n>Our method achieves comparable or even superior performance using only 1,389 samples versus the full 8,523 samples dataset.<n>For reproducible research and future innovation, we are open-sourcing LIMR, including implementation of LIM, training and evaluation code, curated datasets, and trained models.
  arXiv  Detail & Related papers  (2025-02-17T15:13:29Z)
• What Do Learning Dynamics Reveal About Generalization in LLM Reasoning? [83.83230167222852]
  We find that a model's generalization behavior can be effectively characterized by a training metric we call pre-memorization train accuracy. By connecting a model's learning behavior to its generalization, pre-memorization train accuracy can guide targeted improvements to training strategies.
  arXiv  Detail & Related papers  (2024-11-12T09:52:40Z)
• Improving Pretraining Data Using Perplexity Correlations [56.41097718862742]
  We present a framework that selects high-quality pretraining data without any LLM training of our own.<n>We build a new statistical framework for data selection centered around estimates of perplexity-benchmark correlations.<n>Our approach outperforms DSIR on every benchmark, while matching the best data selector found in DataComp-LM.
  arXiv  Detail & Related papers  (2024-09-09T17:23:29Z)
• Zephyr: Direct Distillation of LM Alignment [59.03530095974505]
  We aim to produce a smaller language model that is aligned to user intent. Previous research has shown that applying supervised fine-tuning (dSFT) on larger models significantly improves task accuracy. We apply distilled direct preference optimization (dDPO) to learn a chat model with significantly improved intent alignment.
  arXiv  Detail & Related papers  (2023-10-25T19:25:16Z)
• Large-scale Dataset Pruning with Dynamic Uncertainty [28.60845105174658]
  The state of the art of many learning tasks, e.g., image classification, is advanced by collecting larger datasets and then training larger models on them. In this paper, we investigate how to prune the large-scale datasets, and thus produce an informative subset for training sophisticated deep models with negligible performance drop.
  arXiv  Detail & Related papers  (2023-06-08T13:14:35Z)
• Delving Deeper into Data Scaling in Masked Image Modeling [145.36501330782357]
  We conduct an empirical study on the scaling capability of masked image modeling (MIM) methods for visual recognition. Specifically, we utilize the web-collected Coyo-700M dataset. Our goal is to investigate how the performance changes on downstream tasks when scaling with different sizes of data and models.
  arXiv  Detail & Related papers  (2023-05-24T15:33:46Z)
• Self-Supervised Pre-Training for Transformer-Based Person Re-Identification [54.55281692768765]
  Transformer-based supervised pre-training achieves great performance in person re-identification (ReID) Due to the domain gap between ImageNet and ReID datasets, it usually needs a larger pre-training dataset to boost the performance. This work aims to mitigate the gap between the pre-training and ReID datasets from the perspective of data and model structure.
  arXiv  Detail & Related papers  (2021-11-23T18:59:08Z)
• A contextual analysis of multi-layer perceptron models in classifying hand-written digits and letters: limited resources [0.0]
  We extensively test an end-to-end vanilla neural network (MLP) approach in pure numpy without any pre-processing or feature extraction done beforehand. We show that basic data mining operations can significantly improve the performance of the models in terms of computational time.
  arXiv  Detail & Related papers  (2021-07-05T04:30:37Z)
• Comparing Test Sets with Item Response Theory [53.755064720563]
  We evaluate 29 datasets using predictions from 18 pretrained Transformer models on individual test examples. We find that Quoref, HellaSwag, and MC-TACO are best suited for distinguishing among state-of-the-art models. We also observe span selection task format, which is used for QA datasets like QAMR or SQuAD2.0, is effective in differentiating between strong and weak models.
  arXiv  Detail & Related papers  (2021-06-01T22:33:53Z)
• Omni-supervised Facial Expression Recognition via Distilled Data [120.11782405714234]
  We propose omni-supervised learning to exploit reliable samples in a large amount of unlabeled data for network training. We experimentally verify that the new dataset can significantly improve the ability of the learned FER model. To tackle this, we propose to apply a dataset distillation strategy to compress the created dataset into several informative class-wise images.
  arXiv  Detail & Related papers  (2020-05-18T09:36:51Z)

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.