English
Related papers

Related papers: Forecasting Downstream Performance of LLMs With Pr…

200 papers

Given the prohibitive cost of pre-training large language models, it is essential to leverage smaller proxy models to optimize datasets before scaling up. However, this approach becomes challenging for reasoning capabilities, which exhibit…

Machine Learning · Computer Science 2026-02-27 Woosung Koh , Juyoung Suk , Sungjun Han , Se-Young Yun , Jamin Shin

Most LLM benchmarks score how well a model responds to explicit requests. They leave unmeasured a different conversational ability: noticing and acting on needs the user has implied but not said. We call this \emph{conversational…

Machine Learning · Computer Science 2026-05-12 Sepehr Harfi , Ahmad Salimi , Dongming Shen , Alex Smola

Generative reward models (GRMs) for vision-language models (VLMs) often evaluate outputs via a three-stage pipeline: rubric generation, criterion-based scoring, and a final verdict. However, the intermediate rubric is rarely optimized…

Computer Vision and Pattern Recognition · Computer Science 2026-03-19 Weijie Qiu , Dai Guan , Junxin Wang , Zhihang Li , Yongbo Gai , Mengyu Zhou , Erchao Zhao , Xiaoxi Jiang , Guanjun Jiang

Recent advances in speech large language models (speech LLMs) have enabled seamless spoken interactions, but these systems still struggle with complex reasoning tasks. Previously, chain-of-thought (CoT) prompting or fine-tuning has been to…

Computation and Language · Computer Science 2025-10-10 Yi-Jen Shih , Desh Raj , Chunyang Wu , Wei Zhou , SK Bong , Yashesh Gaur , Jay Mahadeokar , Ozlem Kalinli , Mike Seltzer

Autoregressive pretraining has become the de facto paradigm for learning general-purpose representations in large language models (LLMs). However, linear probe performance across downstream perception tasks shows substantial variability,…

Computation and Language · Computer Science 2025-05-26 Yu-Ang Cheng , Leyang Hu , Hai Huang , Randall Balestriero

Evaluating the robustness of LLMs to adversarial attacks is crucial for safe deployment, yet current red-teaming methods are often prohibitively expensive. We compare the ability of fast proxy metrics to predict the real-world robustness of…

Cryptography and Security · Computer Science 2025-02-18 Tim Beyer , Jan Schuchardt , Leo Schwinn , Stephan Günnemann

Data teams at frontier AI companies routinely train small proxy models to make critical decisions about pretraining data recipes for full-scale training runs. However, the community has a limited understanding of whether and when…

Machine Learning · Computer Science 2026-04-14 Jiachen T. Wang , Tong Wu , Kaifeng Lyu , James Zou , Dawn Song , Ruoxi Jia , Prateek Mittal

Language models have consistently grown to compress more world knowledge into their parameters, but the knowledge that can be pretrained into them is upper-bounded by their parameter size. Especially the capacity of Small Language Models…

Computation and Language · Computer Science 2026-05-18 Szilvia Ujváry , Louis Béthune , Pierre Ablin , João Monteiro , Marco Cuturi , Michael Kirchhof

Empirical researchers increasingly use upstream machine-learning (ML) methods to construct proxies for latent target variables from complex, unstructured data. A naive plug-in use of such proxies in downstream econometric models, however,…

Econometrics · Economics 2026-04-14 Lixiong Li

Prompt optimization is a practical and widely applicable alternative to fine tuning for improving large language model performance. Yet many existing methods evaluate candidate prompts by sampling full outputs, often coupled with self…

Computation and Language · Computer Science 2025-09-19 Chenzhuo Zhao , Ziqian Liu , Xinda Wang , Junting Lu , Chaoyi Ruan

In many scientific domains, including experimentation, researchers rely on measurements of proxy outcomes to achieve faster and more frequent reads, especially when the primary outcome of interest is challenging to measure directly. While…

Methodology · Statistics 2026-05-08 Steven Wilkins-Reeves , Alexandra N. M. Darmon , Deeksha Sinha

Language model pretraining involves training on extensive corpora, where data quality plays a pivotal role. In this work, we aim to directly estimate the contribution of data during pretraining and select pretraining data in an efficient…

Computation and Language · Computer Science 2025-08-05 Kashun Shum , Yuzhen Huang , Hongjian Zou , Qi Ding , Yixuan Liao , Xiaoxin Chen , Qian Liu , Junxian He

Humans ponder before articulating complex sentence elements, enabling deeper cognitive processing through focused effort. In this work, we introduce this pondering process into language models by repeatedly invoking the forward process…

Computation and Language · Computer Science 2026-02-23 Boyi Zeng , Shixiang Song , Siyuan Huang , Yixuan Wang , He Li , Ziwei He , Xinbing Wang , Zhiyu Li , Zhouhan Lin

Cross-entropy loss is a common choice when it comes to multiclass classification tasks and language modeling in particular. Minimizing this loss results in language models of very good quality. We show that it is possible to fine-tune these…

Computation and Language · Computer Science 2019-01-16 Vadim Popov , Mikhail Kudinov

As Language Models (LMs) increasingly operate as autonomous agents, accurately forecasting their capabilities becomes crucial for societal preparedness. We evaluate six forecasting methods that predict downstream capabilities of LM agents.…

Computation and Language · Computer Science 2025-03-04 Govind Pimpale , Axel Højmark , Jérémy Scheurer , Marius Hobbhahn

Training Large Language Models (LLMs) typically involves a two-stage pipeline at the output layer: hidden states are projected into vocabulary logits via a linear transformation (lm_head), followed by cross-entropy loss computation against…

Machine Learning · Computer Science 2025-11-25 Jianbing Dong , Jianbin Chang

Decoding strategies play a central role in shaping the reasoning ability of large language models (LLMs). Traditional methods such as greedy decoding and beam search often suffer from error propagation, while sampling-based approaches…

Existing high-dimensional statistical methods are largely established for analyzing individual-level data. In this work, we study estimation and inference for high-dimensional linear models where we only observe "proxy data", which include…

Methodology · Statistics 2022-01-12 Sai Li , T. Tony Cai , Hongzhe Li

Multimodal models often experience a significant performance drop when one or more modalities are missing during inference. To address this challenge, we propose a simple yet effective approach that enhances robustness to missing modalities…

Computer Vision and Pattern Recognition · Computer Science 2025-10-28 Md Kaykobad Reza , Ameya Patil , Mashhour Solh , M. Salman Asif

The cross entropy loss is widely used due to its effectiveness and solid theoretical grounding. However, as training progresses, the loss tends to focus on hard to classify samples, which may prevent the network from obtaining gains in…

Machine Learning · Computer Science 2021-09-14 Barak Battash , Lior Wolf , Tamir Hazan