arXiv subjects

Brandon Wang

Publications and source records attributed to Brandon Wang.

11 recordsLinked to original sources

arXiv2026

onepot-Bench 0: towards lab-aware in silico chemistry benchmarks

Language models are playing an increasingly important role in laboratory science, performing tasks such as experiment planning, execution, and post-hoc analysis. However, precisely measuring their abilities is difficult, as scientific capabilities require a mixture of both problem-solving skills and domain-specific intuition. Existing evaluations rarely measure the capabilities required to make reliable decisions in a physical laboratory and often rely on public data that may have appeared in model training corpora. We introduce onepot-Bench 0, a proprietary benchmark suite for evaluating language models on synthetic chemistry capabilities relevant to wet-lab execution. onepot-Bench 0 comprises three complementary evaluations: ChemAbacus measures tool-free cheminformatics literacy and numerical reasoning; SynthRefusal characterizes safety and refusal behavior across a variety of benign, controlled, and designer-drug targets; and SynthBench evaluates reaction-outcome prediction and catalyst selection using private experimental data generated in our laboratory. Together, these evaluations probe basic competency, reliability, and deeper knowledge, all skills which are required for reliable performance in the lab.

cs.LG↗

arXiv2026

dnaHNet: A Scalable and Hierarchical Foundation Model for Genomic Sequence Learning

Genomic foundation models have the potential to decode DNA syntax, yet face a fundamental tradeoff in their input representation. Standard fixed-vocabulary tokenizers fragment biologically meaningful motifs such as codons and regulatory elements, while nucleotide-level models preserve biological coherence but incur prohibitive computational costs for long contexts. We introduce dnaHNet, a state-of-the-art tokenizer-free autoregressive model that segments and models genomic sequences end-to-end. Using a differentiable dynamic chunking mechanism, dnaHNet compresses raw nucleotides into latent tokens adaptively, balancing compression with predictive accuracy. Pretrained on prokaryotic genomes, dnaHNet outperforms leading architectures including StripedHyena2 in scaling and efficiency. This recursive chunking yields quadratic FLOP reductions, enabling $>3 \times$ inference speedup over Transformers. On zero-shot tasks, dnaHNet achieves superior performance in predicting protein variant fitness and gene essentiality, while automatically discovering hierarchical biological structures without supervision. These results establish dnaHNet as a scalable, interpretable framework for next-generation genomic modeling.

cs.LG↗

arXiv2026

onepot CORE -- an enumerated chemical space to streamline drug discovery, enabled by automated small molecule synthesis and AI

The design-make-test-analyze cycle in early-stage drug discovery remains constrained primarily by the "make" step: small-molecule synthesis is slow, costly, and difficult to scale or automate across diverse chemotypes. Enumerated chemical spaces aim to reduce this bottleneck by predefining synthesizable regions of chemical space from available building blocks and reliable reactions, yet existing commercial spaces are still limited by long turnaround times, narrow reaction scope, and substantial manual decision-making in route selection and execution. Here we present the first version of onepot CORE, an enumerated chemical space containing 3.4B molecules and corresponding on-demand synthesis product enabled by an automated synthesis platform and an AI chemist, Phil, that designs, executes, and analyzes experiments. onepot CORE is constructed by (i) selecting a reaction set commonly used in medicinal chemistry, (ii) sourcing and curating building blocks from supplier catalogs, (iii) enumerating candidate products, and (iv) applying ML-based feasibility assessment to prioritize compounds for robust execution. In the current release, the space is supported by seven reactions. We describe an end-to-end workflow - from route selection and automated liquid handling through workup and purification. We further report validation across operational metrics (success rate, timelines, purity, and identity), including NMR confirmation for a representative set of synthesized compounds and assay suitability demonstrated using a series of DPP4 inhibitors. Collectively, onepot CORE illustrates a path toward faster, more reliable access to diverse small molecules, supporting accelerated discovery in pharmaceuticals and beyond.

physics.chem-ph↗

arXiv2025

OpenAI GPT-5 System Card

This is the system card published alongside the OpenAI GPT-5 launch, August 2025. GPT-5 is a unified system with a smart and fast model that answers most questions, a deeper reasoning model for harder problems, and a real-time router that quickly decides which model to use based on conversation type, complexity, tool needs, and explicit intent (for example, if you say 'think hard about this' in the prompt). The router is continuously trained on real signals, including when users switch models, preference rates for responses, and measured correctness, improving over time. Once usage limits are reached, a mini version of each model handles remaining queries. This system card focuses primarily on gpt-5-thinking and gpt-5-main, while evaluations for other models are available in the appendix. The GPT-5 system not only outperforms previous models on benchmarks and answers questions more quickly, but -- more importantly -- is more useful for real-world queries. We've made significant advances in reducing hallucinations, improving instruction following, and minimizing sycophancy, and have leveled up GPT-5's performance in three of ChatGPT's most common uses: writing, coding, and health. All of the GPT-5 models additionally feature safe-completions, our latest approach to safety training to prevent disallowed content. Similarly to ChatGPT agent, we have decided to treat gpt-5-thinking as High capability in the Biological and Chemical domain under our Preparedness Framework, activating the associated safeguards. While we do not have definitive evidence that this model could meaningfully help a novice to create severe biological harm -- our defined threshold for High capability -- we have chosen to take a precautionary approach.

cs.CL↗

arXiv2025

Dynamic Chunking for End-to-End Hierarchical Sequence Modeling

Major progress on language models (LMs) in recent years has largely resulted from moving away from specialized models designed for specific tasks, to general models based on powerful architectures (e.g. the Transformer) that learn everything from raw data. Despite this trend, pre-processing steps such as tokenization remain a barrier to true end-to-end foundation models. We introduce a collection of new techniques that enable a dynamic chunking mechanism which automatically learns content- and context- dependent segmentation strategies learned jointly with the rest of the model. Incorporating this into an explicit hierarchical network (H-Net) allows replacing the (implicitly hierarchical) tokenization-LM-detokenization pipeline with a single model learned fully end-to-end. When compute- and data- matched, an H-Net with one stage of hierarchy operating at the byte level outperforms a strong Transformer language model operating over BPE tokens. Iterating the hierarchy to multiple stages further increases its performance by modeling multiple levels of abstraction, demonstrating significantly better scaling with data and matching the token-based Transformer of twice its size. H-Nets pretrained on English show significantly increased character-level robustness, and qualitatively learn meaningful data-dependent chunking strategies without any heuristics or explicit supervision. Finally, the H-Net's improvement over tokenized pipelines is further increased in languages and modalities with weaker tokenization heuristics, such as Chinese and code, or DNA sequences (nearly 4x improvement in data efficiency over baselines), showing the potential of true end-to-end models that learn and scale better from unprocessed data.

cs.LG↗

arXiv2025

Physics and Computing Performance of the EggNet Tracking Pipeline

Particle track reconstruction is traditionally computationally challenging due to the combinatorial nature of the tracking algorithms employed. Recent developments have focused on novel algorithms with graph neural networks (GNNs), which can improve scalability. While most of these GNN-based methods require an input graph to be constructed before performing message passing, a one-shot approach called EggNet that directly takes detector spacepoints as inputs and iteratively apply graph attention networks with an evolving graph structure has been proposed. The graphs are gradually updated to improve the edge efficiency and purity, thus providing a better model performance. In this work, we evaluate the physics and computing performance of the EggNet tracking pipeline on the full TrackML dataset. We also explore different techniques to reduce constraints on computation memory and computing time.

physics.data-an↗

arXiv2025

Biquandles, quivers and virtual bridge indices

We investigate connections between biquandle colorings, quiver enhancements, and several notions of the bridge numbers $b_i(K)$ for virtual links, where $i=1,2$. We show that for any positive integers $m \leq n$, there exists a virtual link $K$ with $b_1(K) = m$ and $b_2(K) = n$, thereby answering a question posed by Nakanishi and Satoh. In some sense, this gap between the two formulations measures how far the knot is from being classical. We also use these bridge number analyses to systematically construct families of links in which quiver invariants can distinguish between links that share the same biquandle counting invariant.

math.GT↗

arXiv2024

Improving the functionality of non-stretching approximations

Entangled polymers are an important class of materials for their toughness, processability, and functionalizability. However, physically detailed modeling of highly entangled polymers can prove challenging, especially as one considers additional layers of physical or chemical complexity. To address these challenges, we present a series of generalizations for the useful "non-stretching" approximation, using asymptotic methods to formalize and expand the analysis. First, we rederive the popular non-stretching Rolie Poly model and extend it second order, reintroducing effects from finite chain stretching. Then, we extended the non-stretching framework to other special cases, accounting for flow-induced disentanglement, polydispersity, and reversible scission reactions. Benchmark calculations confirm that non-stretching models derived via systematic asymptotic methods provide excellent and improvable approximations for the rheology of well-entangled polymer constitutive equations with finite-time stretch relaxation dynamics.

cond-mat.soft↗

arXiv2024

Pathological Rheology of Non-Stretching Entangled Polymers: Finite-Time Blow-Up Predictions

The non-stretching approximation of polymer rheology simplifies a constitutive equation but fundamentally changes its behavior in fast flows, and the circumstances under which fast flows emerge cannot always be predicted a-priori. In this paper, we consider two simple flows for which shear rates are bounded in the original RP model but diverge to infinity in finite time for the non-stretching RP model. The disparity between the full and non-stretching models can be resolved by extending the non-stretching approximation to second order in accuracy.

cond-mat.soft↗

arXiv2024

EggNet: An Evolving Graph-based Graph Attention Network for Particle Track Reconstruction

Track reconstruction is a crucial task in particle experiments and is traditionally very computationally expensive due to its combinatorial nature. Recently, graph neural networks (GNNs) have emerged as a promising approach that can improve scalability. Most of these GNN-based methods, including the edge classification (EC) and the object condensation (OC) approach, require an input graph that needs to be constructed beforehand. In this work, we consider a one-shot OC approach that reconstructs particle tracks directly from a set of hits (point cloud) by recursively applying graph attention networks with an evolving graph structure. This approach iteratively updates the graphs and can better facilitate the message passing across each graph. Preliminary studies on the TrackML dataset show better track performance compared to the methods that require a fixed input graph.

physics.data-an↗

arXiv2019

Urban sensing as a random search process

We study a new random search process: the \textit{taxi-drive}. The motivation for this process comes from urban sensing, in which sensors are mounted on moving vehicles such as taxis, allowing urban environments to be opportunistically monitored. Inspired by the movements of real taxis, the taxi-drive is composed of both random and regular parts; passengers are brought to randomly chosen locations via deterministic (i.e. shortest paths) routes. We show through a numerical study that this hybrid motion endows the taxi-drive with advantageous spreading properties. In particular, on certain graph topologies it offers reduced cover times compared to persistent random walks.

physics.soc-ph↗