Natural Language Processing Papers of the Week

On downstream accuracy evaluations, these predictions fail when the weaker models produced by the cheaper runs score near zero accuracy, because these scores give no trend to extrapolate to larger scales. We introduce hinted scaling laws, which use privileged information to turn a benchmark into a continuum of easier versions so that even the small models score above zero.

As a result, tool interfaces may not match real ones, tool compositions are not grounded in execution, and tasks rarely span more than one environment. We propose ToolSpace, a tool-centric framework that reverses this order.

We introduce CoworkOS, a general-purpose harness for long-horizon work that separates task strategy from execution management: the model decides actions, delegation, and acceptance, while the runtime maintains state and feedback.

We propose TurnWise, a turn-level GRPO framework that aligns learning with individual streaming decisions.

Beyond surveying existing methods, we propose a unified framework for evaluating mechanistic interpretability claims and characterise six research orientations by their goals, assumptions, and trade-offs.

With the world as an evaluation, we found that frontier AI-Scientists can already surpass the human researchers 74% of the time.

We introduce Strategist, a meta-agent that evolves inference strategies across tasks.

We introduce TB-MathModel (the Terminal-style Mathematical Modeling Bench), the first benchmark to make complete contest-style modeling problems agentic, verifiable, and terminal at once, over 18 real CUMCM and MCM problems (2023--2025).

We attribute this failure to a fundamental modality mismatch: reasoning is performed in a discrete token space, whereas image generation operates in a continuous space. To bridge this gap, we introduce a Latent Reasoning framework in Continuous space (LARC), which alleviates the token-level bottleneck by reasoning over continuous hidden states.

We introduce IdeaTaste, a benchmark that scores an LLM by how closely the research proposals it writes from a literature snapshot frozen at 2025 match the work the same communities published in 2026.

While post-training approaches can mitigate this through explicit training, they demand expensive real-world robotics data and require computationally intensive post-training for every individual base model. To resolve these limitations, we introduce D&A (Describe & Assess): our approach decouples progress estimation into an objective, task-grounded textual description of scene dynamics followed by an assessment phase using explicit rubrics.

Other artifacts--ontology, graph, wiki, timeline, mind map, memory--pay off in deployment (governance, integration, onboarding, temporal orientation, continuity) where accuracy benchmarks cannot see them. We describe RAGFlow's pipeline and harness (open source at this https URL), which turns the loop into five disciplined stations and keeps the context entering the model the smallest sufficient set; together they form an agent-native database: a store that hands the agent compiled, governed context instead of a ranked list of passages.

Besides, agentic RL does not uniformly elevate single-step decision quality; rather, its strength concentrates in error recovery and long-horizon resilience. To mitigate the deployment risks caused by this budget dependence, we provide a preliminary discussion on the benefits and limitations of process supervision signals and introduce BudgetMix RL as an alternative that varies interaction budgets during training.

We propose LoCo, a local preference learning method for reasoning compression.

We present Drex, a family of four open decision models with 8B (DLM), 9B (v1.5), 35B with 3B active (v2.0), and 27B (Max) parameters.

Yet, how this feedback shapes learning and generalization remains poorly understood. We conduct a mechanistic analysis of OPD's learning dynamics, distinguishing the two feedback loop stages: conversion of the distillation signal into parameter updates, and broadcast through which these updates change predictions across prefixes and reshape the rollout distribution.

In this paper, we derive an asymptotically subquadratic sparse attention method based on a hierarchical, tree-structured indexer, which we call Zyphra Hierarchical Attention (ZHA).

Other ranking losses yield similar gains, so the benefit comes from ordinal supervision itself. Analysis shows that O$^{3}$PD departs further from the teacher's probabilities than OPD, yet reproduces its ranking more faithfully and yields a stronger student.

This paper presents TurnTurk-GPT, a text-based turn-taking predictor for Turkish obtained by fine-tuning a Turkish GPT-2 model with the TurnGPT objective on synthetic and real dialogue, and evaluates it on real Turkish conversation against rule baselines, a zero-shot Turkish language model and a widely used multilingual end-of-turn detector.