AuthorsZhen Zhang†, Changyi Yang‡§, Zijie Xia§, Zhen Yang, Chengzhi Liu†, Zhaotiao Weng†, Yepeng Liu†, Haobo Chen†, Jin Pan¶§, Chenyang Zhao§, Yuheng Bu†, Alkesh Patel, Zhe Gan, Xin Eric Wang†
Token serves as the fundamental unit of computation in modern autoregressive models, and generation length directly influences both inference cost and reasoning performance. Despite its importance, existing approaches lack fine-grained length modeling, operating primarily at the coarse-grained sequence level. In this paper, we introduce the Length Value Model (LenVM), a token-level framework that models the remaining generation length at each decoding step. By formulating length modeling as a value estimation problem and assigning a constant negative reward to each generated token, LenVM predicts a bounded, discounted return that serves as a monotone proxy for the remaining generation horizon. This formulation yields supervision that is annotation-free, dense, unbiased, and scalable. Experiments on LLMs and VLMs demonstrate that LenVM provides a highly effective signal at inference time. On the LIFEBench exact length matching task, applying LenVM to a 7B model improves the length score from 30.9 to 64.8, significantly outperforming frontier closed-source models. Furthermore, LenVM enables continuous control over the trade off between performance and efficiency. On GSM8K at a budget of 200 tokens, LenVM maintains 63 percent accuracy compared to 6 percent for token budget baseline. It also accurately predicts total generation length from the prompt boundary. Finally, LenVM’s token-level values offer an interpretable view of generation dynamics, revealing how specific tokens shift reasoning toward shorter or longer regimes. Results demonstrate that LenVM supports a broad range of applications, including length control, prediction, and interpretation of generation dynamics. They suggest that generation length can be effectively modeled as a token-level value signal, highlighting the potential of LenVM as a general framework for length modeling and as a length-specific value signal that could support future RL training.
- † University of California, Santa Barbara
- ‡ Carnegie Mellon University
- § LMSYS Org
- ¶ University of Wisconsin–Madison
Related readings and updates.
A Formal Framework for Understanding Length Generalization in Transformers
A major challenge for transformers is generalizing to sequences longer than those observed during training. While previous works have empirically shown that transformers can either succeed or fail at length generalization depending on the task, theoretical understanding of this phenomenon remains limited. In this work, we introduce a rigorous theoretical framework to analyze length generalization in causal transformers with learnable absolute…
Dataset Decomposition: Faster LLM Training with Variable Sequence Length Curriculum
Large language models (LLMs) are commonly trained on datasets consisting of fixed-length token sequences. These datasets are created by randomly concatenating documents of various lengths and then chunking them into sequences of a predetermined target length (concat-and-chunk). Recent attention implementations mask cross-document attention, reducing the effective length of a chunk of tokens. Additionally, training on long sequences becomes…
