arXiv subjects

Noah A. Smith

Publications and source records attributed to Noah A. Smith.

At least 37 records · Page 2Linked to original sources

arXiv2026

RewardBench 2: Advancing Reward Model Evaluation

Reward models are used throughout the post-training of language models to capture nuanced signals from preference data and provide a training target for optimization across instruction following, reasoning, safety, and more domains. The community has begun establishing best practices for evaluating reward models, from the development of benchmarks that test capabilities in specific skill areas to others that test agreement with human preferences. At the same time, progress in evaluation has not been mirrored by the effectiveness of reward models in downstream tasks -- simpler direct alignment algorithms are reported to work better in many cases. This paper introduces RewardBench 2, a new multi-skill reward modeling benchmark designed to bring new, challenging data for accuracy-based reward model evaluation -- models score about 20 points on average lower on RewardBench 2 compared to the first RewardBench -- while being highly correlated with downstream performance. Compared to most other benchmarks, RewardBench 2 sources new human prompts instead of existing prompts from downstream evaluations, facilitating more rigorous evaluation practices. In this paper, we describe our benchmark construction process and report how existing models perform on it, while quantifying how performance on the benchmark correlates with downstream use of the models in both inference-time scaling algorithms, like best-of-N sampling, and RLHF training algorithms like proximal policy optimization.

cs.CL↗

arXiv2026

Train Separately, Merge Together: Modular Post-Training with Mixture-of-Experts

Extending a fully post-trained language model with new domain capabilities is fundamentally limited by monolithic training paradigms: retraining from scratch is expensive and scales poorly, while continued training often degrades existing capabilities. We present BAR (Branch-Adapt-Route), which trains independent domain experts, each through its own mid-training, supervised finetuning, and reinforcement learning pipeline, and composes them via a Mixture-of-Experts architecture with lightweight router training. Unlike retraining approaches that mix all domains and require full reprocessing for any update (with cost scaling quadratically), BAR enables updating individual experts independently with linear cost scaling and no degradation to existing domains. At the 7B scale, with experts for math, code, tool use, and safety, BAR achieves an overall score of 49.1 (averaged across 7 evaluation categories), matching or exceeding re-training baselines (47.8 without mid-training, 50.5 with). We further show that modular training provides a structural advantage: by isolating each domain, it avoids the catastrophic forgetting that occurs when late-stage RL degrades capabilities from earlier training stages, while significantly reducing the cost and complexity of updating or adding a domain. Together, these results suggest that decoupled, expert-based training is a scalable alternative to monolithic retraining for extending language models.

cs.LG↗

arXiv2026

Reading Between the Lines: The One-Sided Conversation Problem

Conversational AI is constrained in many real-world settings where only one side of a dialogue can be recorded, such as telemedicine, call centers, and smart glasses. We formalize this as the one-sided conversation problem (1SC): inferring and learning from one side of a conversation. We study two tasks: (1) reconstructing the missing speaker's turns for real-time use cases, and (2) generating summaries from one-sided transcripts. Evaluating prompting and finetuned models on MultiWOZ, DailyDialog, and Candor with both human A/B testing and LLM-as-a-judge metrics, we find that access to one future turn and information about utterance length improves reconstruction, placeholder prompting helps to mitigate hallucination, and while large models generate promising reconstructions with prompting, smaller models require finetuning. Further, high-quality summaries can be generated without reconstructing missing turns. We present 1SC as a novel challenge and report promising results that mark a step toward privacy-aware conversational AI.

cs.CL↗

arXiv2026

Olmo 3

We introduce Olmo 3, a family of state-of-the-art, fully-open language models at the 7B and 32B parameter scales. Olmo 3 model construction targets long-context reasoning, function calling, coding, instruction following, general chat, and knowledge recall. This release includes the entire model flow, i.e., the full lifecycle of the family of models, including every stage, checkpoint, data point, and dependency used to build it. Our flagship model, Olmo 3 Think 32B, is the strongest fully-open thinking model released to-date.

cs.CL↗

arXiv2026

Meta-Reinforcement Learning with Self-Reflection for Agentic Search

This paper introduces MR-Search, an in-context meta reinforcement learning (RL) formulation for agentic search with self-reflection. Instead of optimizing a policy within a single independent episode with sparse rewards, MR-Search trains a policy that conditions on past episodes and adapts its search strategy across episodes. MR-Search learns to learn a search strategy with self-reflection, allowing search agents to improve in-context exploration at test-time. Specifically, MR-Search performs cross-episode exploration by generating explicit self-reflections after each episode and leveraging them as additional context to guide subsequent attempts, thereby promoting more effective exploration during test-time. We further introduce a multi-turn RL algorithm that estimates a dense relative advantage at the turn level, enabling fine-grained credit assignment on each episode. Empirical results across various benchmarks demonstrate the advantages of MR-Search over baselines based RL, showing strong generalization and relative improvements of 9.2% to 19.3% across eight benchmarks. Our code and data are available at https://github.com/tengxiao1/MR-Search.

cs.LG↗

arXiv2026

Bolmo: Byteifying the Next Generation of Language Models

Recent advances in generative AI have been largely driven by large language models (LLMs), deep neural networks that operate over discrete units called tokens. To represent text, the vast majority of LLMs use words or word fragments as the tokens, known as subword tokenization. Subword tokenization obscures fine-grained information, which is problematic, especially for scientific data - such as computer code or biological sequences - where meaning depends on the individual characters. Models that instead operate directly on the byte encoding of text avoid these limitations, but until now they have lagged behind subword-based models in performance. Here we introduce Bolmo, a family of fully open byte-level LLMs that approach the capabilities of subword-based systems. Using a two-stage conversion procedure, we transform existing subword-based models into byte-level models with minimal additional training. The resulting models outperform prior byte-level approaches and excel on character-level reasoning tasks, while remaining competitive across standard benchmarks. By efficiently processing byte-level information, these models achieve practical inference speeds and can be adapted at low cost using the existing ecosystem around the source LLM. Our results remove a long-standing performance barrier to end-to-end byte-level language modeling, demonstrating that models operating on raw text encodings can scale competitively while offering advantages in domains requiring fine-grained textual understanding.

cs.CL↗

arXiv2026

BASS: Benchmarking Audio LMs for Musical Structure and Semantic Reasoning

Music understanding is a complex task that often requires reasoning over both structural and semantic elements of audio. We introduce BASS, designed to evaluate music understanding and reasoning in audio language models across four broad categories: structural segmentation, lyric transcription, musicological analysis, and artist collaboration. BASS comprises 2658 questions spanning 12 tasks, 1993 unique songs and covering over 138 hours of music from a wide range of genres and tracks, crafted to assess musicological knowledge and reasoning in real-world scenarios. We evaluate 14 open-source and frontier multimodal LMs, finding that even state-of-the-art models struggle on higher-level reasoning tasks such as structural segmentation and artist collaboration, while performing best on lyric transcription. Our analysis reveals that current models leverage linguistic priors effectively but remain limited in reasoning over musical structure, vocal, and musicological attributes. BASS provides an evaluation framework with widespread applications in music recommendation and search and has the potential to guide the development of audio LMs.

cs.SD↗

arXiv2026

Broken Tokens? Your Language Model can Secretly Handle Non-Canonical Tokenizations

Modern tokenizers employ deterministic algorithms to map text into a single "canonical" token sequence, yet the same string can be encoded as many non-canonical tokenizations using the tokenizer vocabulary. In this work, we investigate the robustness of LMs to text encoded with non-canonical tokenizations entirely unseen during training. Surprisingly, when evaluated across 20 benchmarks, we find that instruction-tuned models retain up to 93.4% of their original performance when given a randomly sampled tokenization, and 90.8% with character-level tokenization. We see that overall stronger models tend to be more robust, and robustness diminishes as the tokenization departs farther from the canonical form. Motivated by these results, we then identify settings where non-canonical tokenization schemes can *improve* performance, finding that character-level segmentation improves string manipulation and code understanding tasks by up to +14%, and right-aligned digit grouping enhances large-number arithmetic by +33%. Finally, we investigate the source of this robustness, finding that it arises in the instruction-tuning phase. We show that while both base and post-trained models grasp the semantics of non-canonical tokenizations (perceiving them as containing misspellings), base models try to mimic the imagined mistakes and degenerate into nonsensical output, while post-trained models are committed to fluent responses. Overall, our findings suggest that models are less tied to their tokenizer than previously believed, and demonstrate the promise of intervening on tokenization at inference time to boost performance.

cs.CL↗

arXiv2026

Are you going to finish that? A Practical Study of the Partial Token Problem

Language models (LMs) are trained over sequences of tokens, whereas users interact with LMs via text. This mismatch gives rise to the partial token problem, which occurs when a user ends their prompt in the middle of the expected next-token, leading to distorted next-token predictions. Although this issue has been studied using arbitrary character prefixes, its prevalence and severity in realistic prompts respecting word boundaries remains underexplored. In this work, we identify three domains where token and "word" boundaries often do not line up: languages that do not use whitespace, highly compounding languages, and code. In Chinese, for example, up to 25% of word boundaries do not line up with token boundaries, making even natural, word-complete prompts susceptible to this problem. We systematically construct semantically natural prompts ending with a partial tokens; in experiments, we find that they comprise a serious failure mode: frontier LMs consistently place three orders of magnitude less probability on the correct continuation compared to when the prompt is "backed-off" to be token-aligned. This degradation does not diminish with scale and often worsens for larger models. Finally, we evaluate inference-time mitigations to the partial token problem and validate the effectiveness of recent exact solutions. Overall, we demonstrate the scale and severity of probability distortion caused by tokenization in realistic use cases, and provide practical recommentions for model inference providers.

cs.CL↗

arXiv2026

Infini-gram mini: Exact n-gram Search at the Internet Scale with FM-Index

Language models are trained mainly on massive text data from the Internet, and it becomes increasingly important to understand this data source. Exact-match search engines enable searching in large text corpora - counting string appearances and retrieving the enclosing documents - yet the high storage overhead hinders their application on Internet-scale data. We present infini-gram mini, an efficient and scalable system that can make petabyte-level text corpora searchable. Based on the FM-index data structure (Ferragina and Manzini, 2000), which simultaneously indexes and compresses text, our system creates indexes with size only 44% of the corpus. Infini-gram mini greatly improves upon the best existing implementation of FM-index in terms of indexing speed (18$\times$) and memory use during both indexing (3.2$\times$ reduction) and querying (down to a negligible amount). We index 83TB of Internet text in 99 days with a single CPU node with 128 vCPUs (or 19 hours if using 137 such nodes). We show one important use case of infini-gram mini in a large-scale analysis of benchmark contamination. We find several core LM evaluation benchmarks to be heavily contaminated in Internet crawls (up to 74.2% in GSM8K), which could lead to overestimating the capabilities of language models if trained on such data. We host a benchmark contamination bulletin to share the contamination rate of many core and community-contributed benchmarks. We also release a web interface and an API endpoint to serve general search queries on infini-gram mini indexes.

cs.CL↗

arXiv2025

Sample, Don't Search: Rethinking Test-Time Alignment for Language Models

Increasing test-time computation has emerged as a promising direction for improving language model performance, particularly in scenarios where model finetuning is impractical or impossible due to computational constraints or private model weights. However, existing test-time search methods using a reward model (RM) often degrade in quality as compute scales, due to the over-optimization of what are inherently imperfect reward proxies. We introduce QAlign, a new test-time alignment approach. As we scale test-time compute, QAlign converges to sampling from the optimal aligned distribution for each individual prompt. By adopting recent advances in Markov chain Monte Carlo for text generation, our method enables better-aligned outputs without modifying the underlying model or even requiring logit access. We demonstrate the effectiveness of QAlign on mathematical reasoning benchmarks (GSM8K and GSM-Symbolic) using a task-specific RM, showing consistent improvements over existing test-time compute methods like best-of-n and majority voting. Furthermore, when applied with more realistic RMs trained on the Tulu 3 preference dataset, QAlign outperforms direct preference optimization (DPO), best-of-n, majority voting, and weighted majority voting on a diverse range of datasets (GSM8K, MATH500, IFEval, MMLU-Redux, and TruthfulQA). A practical solution to aligning language models at test time using additional computation without degradation, our approach expands the limits of the capability that can be obtained from off-the-shelf language models without further training.

cs.CL↗

arXiv2025

2 OLMo 2 Furious

We present OLMo 2, the next generation of our fully open language models. OLMo 2 includes a family of dense autoregressive language models at 7B, 13B and 32B scales with fully released artifacts -- model weights, full training data, training code and recipes, training logs and thousands of intermediate checkpoints. In this work, we describe our modified model architecture and training recipe, focusing on techniques for achieving better training stability and improved per-token efficiency. Our updated pretraining data mixture introduces a new, specialized data mix called Dolmino Mix 1124, which significantly improves model capabilities across many downstream task benchmarks when introduced via late-stage curriculum training (i.e. specialized data during the annealing phase of pretraining). Finally, we incorporate best practices from Tülu 3 to develop OLMo 2-Instruct, focusing on permissive data and extending our final-stage reinforcement learning with verifiable rewards (RLVR). Our OLMo 2 base models sit at the Pareto frontier of performance to training compute, often matching or outperforming open-weight only models like Llama 3.1, Qwen 2.5, and Gemma 2 while using fewer FLOPs and with fully transparent training data, code, and recipe. Our fully open OLMo 2-Instruct models are competitive with open-weight only models of comparable size and even some proprietary models like GPT-3.5 Turbo and GPT 4o Mini.

cs.CL↗

arXiv2025

Fluid Language Model Benchmarking

Language model (LM) benchmarking faces several challenges: comprehensive evaluations are costly, benchmarks often fail to measure the intended capabilities, and evaluation quality can degrade due to labeling errors and benchmark saturation. Although various strategies have been proposed to mitigate these issues, they tend to address individual aspects in isolation, neglecting broader questions about overall evaluation quality. Here, we introduce Fluid Benchmarking, a new evaluation approach that advances LM benchmarking across multiple dimensions. Inspired by psychometrics, Fluid Benchmarking is based on the insight that the relative value of benchmark items depends on an LM's capability level, suggesting that evaluation should adapt to each LM. Methodologically, Fluid Benchmarking estimates an item response model based on existing LM evaluation results and uses the inferred quantities to select evaluation items dynamically, similar to computerized adaptive testing in education. In our experiments, we compare Fluid Benchmarking against the common practice of random item sampling as well as more sophisticated baselines, including alternative methods grounded in item response theory. We examine four dimensions -- efficiency, validity, variance, and saturation -- and find that Fluid Benchmarking achieves superior performance in all of them (e.g., higher validity and less variance on MMLU with fifty times fewer items). Our analysis shows that the two components of Fluid Benchmarking have distinct effects: item response theory, used to map performance into a latent ability space, increases validity, while dynamic item selection reduces variance. Overall, our results suggest that LM benchmarking can be substantially improved by moving beyond static evaluation.

cs.CL↗

arXiv2025

SuperBPE: Space Travel for Language Models

The assumption across nearly all language model (LM) tokenization schemes is that tokens should be subwords, i.e., contained within word boundaries. While providing a seemingly reasonable inductive bias, is this common practice limiting the potential of modern LMs? Whitespace is not a reliable delimiter of meaning, as evidenced by multi-word expressions (e.g., "by the way"), crosslingual variation in the number of words needed to express a concept (e.g., "spacesuit helmet" in German is "raumanzughelm"), and languages that do not use whitespace at all (e.g., Chinese). To explore the potential of tokenization beyond subwords, we introduce a "superword" tokenizer, SuperBPE, which incorporates a simple pretokenization curriculum into the byte-pair encoding (BPE) algorithm to first learn subwords, then superwords that bridge whitespace. This brings dramatic improvements in encoding efficiency: when fixing the vocabulary size to 200k, SuperBPE encodes a fixed piece of text with up to 33% fewer tokens than BPE on average. In experiments, we pretrain 8B transformer LMs from scratch while fixing the model size, vocabulary size, and train compute, varying *only* the algorithm for learning the vocabulary. Our model trained with SuperBPE achieves an average +4.0% absolute improvement over the BPE baseline across 30 downstream tasks (including +8.2% on MMLU), while simultaneously requiring 27% less compute at inference time. In analysis, we find that SuperBPE results in segmentations of text that are more uniform in per-token difficulty. Qualitatively, this may be because SuperBPE tokens often capture common multi-word expressions that function semantically as a single unit. SuperBPE is a straightforward, local modification to tokenization that improves both encoding efficiency and downstream performance, yielding better language models overall.

cs.CL↗

arXiv2025

FlexOlmo: Open Language Models for Flexible Data Use

We introduce FlexOlmo, a new class of language models (LMs) that supports (1) distributed training without data sharing, where different model parameters are independently trained on closed datasets, and (2) data-flexible inference, where these parameters along with their associated data can be flexibly included or excluded from model inferences with no further training. FlexOlmo employs a mixture-of-experts (MoE) architecture where each expert is trained independently on closed datasets and later integrated through a new domain-informed routing without any joint training. FlexOlmo is trained on FlexMix, a corpus we curate comprising publicly available datasets alongside seven domain-specific sets, representing realistic approximations of closed sets. We evaluate models with up to 37 billion parameters (20 billion active) on 31 diverse downstream tasks. We show that a general expert trained on public data can be effectively combined with independently trained experts from other data owners, leading to an average 41% relative improvement while allowing users to opt out of certain data based on data licensing or permission requirements. Our approach also outperforms prior model merging methods by 10.1% on average and surpasses the standard MoE trained without data restrictions using the same training FLOPs. Altogether, this research presents a solution for both data owners and researchers in regulated industries with sensitive or protected data. FlexOlmo enables benefiting from closed data while respecting data owners' preferences by keeping their data local and supporting fine-grained control of data access during inference.

cs.CL↗

arXiv2025

Evaluating $n$-Gram Novelty of Language Models Using Rusty-DAWG

How novel are texts generated by language models (LMs) relative to their training corpora? In this work, we investigate the extent to which modern LMs generate $n$-grams from their training data, evaluating both (i) the probability LMs assign to complete training $n$-grams and (ii) $n$-novelty, the proportion of $n$-grams generated by an LM that did not appear in the training data (for arbitrarily large $n$). To enable arbitrary-length $n$-gram search over a corpus in constant time w.r.t. corpus size, we develop Rusty-DAWG, a novel search tool inspired by indexing of genomic data. We compare the novelty of LM-generated text to human-written text and explore factors that affect generation novelty, focusing on the Pythia models. We find that, for $n > 4$, LM-generated text is less novel than human-written text, though it is more novel for smaller $n$. Larger LMs and more constrained decoding strategies both decrease novelty. Finally, we show that LMs complete $n$-grams with lower loss if they are more frequent in the training data. Overall, our results reveal factors influencing the novelty of LM-generated text, and we release Rusty-DAWG to facilitate further pretraining data research.

cs.CL↗

arXiv2025

Establishing Task Scaling Laws via Compute-Efficient Model Ladders

We develop task scaling laws and model ladders to predict the individual task performance of pretrained language models (LMs) in the overtrained setting. Standard power laws for language modeling loss cannot accurately model task performance. Therefore, we leverage a two-step prediction approach: (1) use model and data size to predict an intermediate loss, then (2) use it to predict task performance. We train a set of small-scale "ladder" models, collect data points to fit the parameterized functions of the two prediction steps, and make predictions for two target models: a 7B model trained to 4T tokens and a 13B model trained to 5T tokens. Training the ladder models only costs 1% of the compute used for the target models. On four multiple-choice tasks formatted as ranked classification, we can predict the accuracy of both target models within 2 points of absolute error. We find that tasks with higher prediction error also have higher variance in the metrics over model checkpoints. We also contrast multiple design choices for predicting accuracy, and present recommendations for extending our method to new models and tasks.

cs.CL↗

arXiv2025

How Performance Pressure Influences AI-Assisted Decision Making

Many domains now employ AI-based decision-making aids, and although the potential for AI systems to assist with decision making is much discussed, human-AI collaboration often underperforms due to factors such as (mis)trust in the AI system and beliefs about AI being incapable of completing subjective tasks. One potential tool for influencing human decision making is performance pressure, which hasn't been much studied in interaction with human-AI decision making. In this work, we examine how pressure and explainable AI (XAI) techniques interact with AI advice-taking behavior. Using an inherently low-stakes task (spam review classification), we demonstrate effective and simple methods to apply pressure and influence human AI advice-taking behavior by manipulating financial incentives and imposing time limits. Our results show complex interaction effects, with different combinations of pressure and XAI techniques either improving or worsening AI advice taking behavior. We conclude by discussing the implications of these interactions, strategies to effectively use pressure, and encourage future research to incorporate pressure analysis.

cs.HC↗