Yawen Zhang

Diji Yang, Kezhen Chen, Jinmeng Rao et al.

Visual language tasks require AI models to comprehend and reason with both visual and textual content. Driven by the power of Large Language Models (LLMs), two prominent methods have emerged: (1) the hybrid integration between LLMs and Vision-Language Models (VLMs), where visual inputs are firstly converted into language descriptions by VLMs, serving as inputs for LLMs to generate final answer(s); (2) visual feature alignment in language space, where visual inputs are encoded as embeddings and projected to LLMs' language space via further supervised fine-tuning. The first approach provides light training costs and interpretability but is hard to be optimized in an end-to-end fashion. The second approach presents decent performance, but feature alignment usually requires large amounts of training data and lacks interpretability. To tackle this dilemma, we propose a novel approach, Inner Monologue Multi-Modal Optimization (IMMO), to solve complex vision language problems by simulating inner monologue processes, a cognitive process in which an individual engages in silent verbal communication with themselves. We enable LLMs and VLMs to interact through natural language conversation and propose to use a two-stage training process to learn how to do the inner monologue (self-asking questions and answering questions). IMMO is evaluated on two popular tasks and the results suggest by emulating the cognitive phenomenon of internal dialogue, our approach can enhance reasoning and explanation abilities, contributing to the more effective fusion of vision and language models. More importantly, instead of using predefined human-crafted monologues, IMMO learns this process within the deep learning models, promising wider applicability to many different AI problems beyond vision language tasks.

Gabriel Tseng, Yawen Zhang, Favyen Bastani et al.

We present a set of improvements to the OlmoEarth family. These improvements allow us to cut compute costs during training ($1.7 \times$ reduction in GPU hours required to train our Base models) and inference ($2.9\times$ reductions in MACs on Sentinel-2 tasks), while maintaining the models' overall performance. All training code is available at github.com/allenai/olmoearth_pretrain.

Yawen Zhang, Michael Hannigan, Qin Lv

Air pollution is a major global environmental health threat, in particular for people who live or work near pollution sources. Areas adjacent to pollution sources often have high ambient pollution concentrations, and those areas are commonly referred to as air pollution hotspots. Detecting and characterizing pollution hotspots are of great importance for air quality management, but are challenging due to the high spatial and temporal variability of air pollutants. In this work, we explore the use of mobile sensing data (i.e., air quality sensors installed on vehicles) to detect pollution hotspots. One major challenge with mobile sensing data is uneven sampling, i.e., data collection can vary by both space and time. To address this challenge, we propose a two-step approach to detect hotspots from mobile sensing data, which includes local spike detection and sample-weighted clustering. Essentially, this approach tackles the uneven sampling issue by weighting samples based on their spatial frequency and temporal hit rate, so as to identify robust and persistent hotspots. To contextualize the hotspots and discover potential pollution source characteristics, we explore a variety of cross-domain urban data and extract features from them. As a soft-validation of the extracted features, we build hotspot inference models for cities with and without mobile sensing data. Evaluation results using real-world mobile sensing air quality data as well as cross-domain urban data demonstrate the effectiveness of our approach in detecting and inferring pollution hotspots. Furthermore, the empirical analysis of hotspots and source features yields useful insights regarding neighborhood pollution sources.

Yunkai Zhang, Yawen Zhang, Ming Zheng et al.

Time-series data is critical across many scientific and industrial domains, including environmental analysis, agriculture, transportation, and finance. However, mining insights from this data typically requires deep domain expertise, a process that is both time-consuming and labor-intensive. In this paper, we propose \textbf{Insight Miner}, a large-scale multimodal model (LMM) designed to generate high-quality, comprehensive time-series descriptions enriched with domain-specific knowledge. To facilitate this, we introduce \textbf{TS-Insights}\footnote{Available at \href{https://huggingface.co/datasets/zhykoties/time-series-language-alignment}{https://huggingface.co/datasets/zhykoties/time-series-language-alignment}.}, the first general-domain dataset for time series and language alignment. TS-Insights contains 100k time-series windows sampled from 20 forecasting datasets. We construct this dataset using a novel \textbf{agentic workflow}, where we use statistical tools to extract features from raw time series before synthesizing them into coherent trend descriptions with GPT-4. Following instruction tuning on TS-Insights, Insight Miner outperforms state-of-the-art multimodal models, such as LLaVA \citep{liu2023llava} and GPT-4, in generating time-series descriptions and insights. Our findings suggest a promising direction for leveraging LMMs in time series analysis, and serve as a foundational step toward enabling LLMs to interpret time series as a native input modality.

Kexue Fu, Jiaye Leng, Yawen Zhang et al.

Balancing scientific exposition and narrative engagement is a central challenge in science communication. To examine how to achieve balance, we conducted a formative study with four science communicators and a literature review of science communication practices, focusing on their workflows and strategies. These insights revealed how creators iteratively shift between exposition and engagement but often lack structured support. Building on this, we developed SpatialBalancing, a co-writing system that connects human spatial reasoning with the linguistic intelligence of large language models. The system visualizes revision trade-offs in a dual-axis space, where users select strategy-based labels to generate, compare, and refine versions during the revision process. This spatial externalization transforms revision into spatial navigation, enabling intentional iterations that balance scientific rigor with narrative appeal. In a within-subjects study (N=16), SpatialBalancing enhanced metacognitive reflection, flexibility, and creative exploration, demonstrating how coupling spatial reasoning with linguistic generation fosters monitoring in iterative science communication writing.

Chongyang Gao, Kezhen Chen, Jinmeng Rao et al. · deepmind

Parameter-efficient tuning (PEFT) techniques like low-rank adaptation (LoRA) offer training efficiency on Large Language Models, but their impact on model performance remains limited. Recent efforts integrate LoRA and Mixture-of-Experts (MoE) to improve the performance of PEFT methods. Despite promising results, research on improving the efficiency of LoRA with MoE is still in its early stages. Recent studies have shown that experts in the MoE architecture have different strengths and also exhibit some redundancy. Does this statement also apply to parameter-efficient MoE? In this paper, we introduce a novel parameter-efficient MoE method, \textit{\textbf{M}oE-L\textbf{o}RA with \textbf{L}ayer-wise Expert \textbf{A}llocation (MoLA)} for Transformer-based models, where each model layer has the flexibility to employ a varying number of LoRA experts. We investigate several architectures with varying layer-wise expert configurations. Experiments on six well-known NLP and commonsense QA benchmarks demonstrate that MoLA achieves equal or superior performance compared to all baselines. We find that allocating more LoRA experts to higher layers further enhances the effectiveness of models with a certain number of experts in total. With much fewer parameters, this allocation strategy outperforms the setting with the same number of experts in every layer. This work can be widely used as a plug-and-play parameter-efficient tuning approach for various applications. The code is available at https://github.com/GCYZSL/MoLA.

Jiaying Lu, Jinmeng Rao, Kezhen Chen et al.

Large Vision-Language Models (LVLMs) offer remarkable benefits for a variety of vision-language tasks. However, a challenge hindering their application in real-world scenarios, particularly regarding safety, robustness, and reliability, is their constrained semantic grounding ability, which pertains to connecting language to the physical-world entities or concepts referenced in images. Therefore, a crucial need arises for a comprehensive study to assess the semantic grounding ability of widely used LVLMs. Despite the significance, sufficient investigation in this direction is currently lacking. Our work bridges this gap by designing a pipeline for generating large-scale evaluation datasets covering fine-grained semantic information, such as color, number, material, etc., along with a thorough assessment of seven popular LVLMs' semantic grounding ability. Results highlight prevalent misgrounding across various aspects and degrees. To address this issue, we propose a data-centric enhancement method that aims to improve LVLMs' semantic grounding ability through multimodal instruction tuning on fine-grained conversations. Experiments on enhanced LVLMs demonstrate notable improvements in addressing misgrounding issues.

Henry Herzog, Favyen Bastani, Yawen Zhang et al.

Earth observation data presents a unique challenge: it is spatial like images, sequential like video or text, and highly multimodal. We present OlmoEarth: a multimodal, spatio-temporal foundation model that employs a novel self-supervised learning formulation, masking strategy, and loss all designed for the Earth observation domain. OlmoEarth achieves state-of-the-art performance compared to 12 other foundation models across a variety of research benchmarks and real-world tasks from external partners. When evaluating embeddings OlmoEarth achieves the best performance on 15 out of 24 tasks, and with full fine-tuning it is the best on 19 of 29 tasks. We deploy OlmoEarth as the backbone of an end-to-end platform for data collection, labeling, training, and inference of Earth observation models. The OlmoEarth Platform puts frontier foundation models and powerful data management tools into the hands of non-profits and NGOs working to solve the world's biggest problems. OlmoEarth source code, training data, and pre-trained weights are available at $\href{https://github.com/allenai/olmoearth_pretrain}{\text{https://github.com/allenai/olmoearth_pretrain}}$.

Diji Yang, Jinmeng Rao, Kezhen Chen et al.

Although the Retrieval-Augmented Generation (RAG) paradigms can use external knowledge to enhance and ground the outputs of Large Language Models (LLMs) to mitigate generative hallucinations and static knowledge base problems, they still suffer from limited flexibility in adopting Information Retrieval (IR) systems with varying capabilities, constrained interpretability during the multi-round retrieval process, and a lack of end-to-end optimization. To address these challenges, we propose a novel LLM-centric approach, IM-RAG, that integrates IR systems with LLMs to support multi-round RAG through learning Inner Monologues (IM, i.e., the human inner voice that narrates one's thoughts). During the IM process, the LLM serves as the core reasoning model (i.e., Reasoner) to either propose queries to collect more information via the Retriever or to provide a final answer based on the conversational context. We also introduce a Refiner that improves the outputs from the Retriever, effectively bridging the gap between the Reasoner and IR modules with varying capabilities and fostering multi-round communications. The entire IM process is optimized via Reinforcement Learning (RL) where a Progress Tracker is incorporated to provide mid-step rewards, and the answer prediction is further separately optimized via Supervised Fine-Tuning (SFT). We conduct extensive experiments with the HotPotQA dataset, a popular benchmark for retrieval-based, multi-step question-answering. The results show that our approach achieves state-of-the-art (SOTA) performance while providing high flexibility in integrating IR modules as well as strong interpretability exhibited in the learned inner monologues.

Patrick Alan Johnson, Gabriel Tseng, Yawen Zhang et al.

Wildfires are increasing in intensity and severity at an alarming rate. Recent advances in AI and publicly available satellite data enable monitoring critical wildfire risk factors globally, at high resolution and low latency. Live Fuel Moisture Content (LFMC) is a critical wildfire risk factor and is valuable for both wildfire research and operational response. However, ground-based LFMC samples are both labor intensive and costly to acquire, resulting in sparse and infrequent updates. In this work, we explore the use of a pretrained, highly-multimodal earth-observation model for generating large-scale spatially complete (wall-to-wall) LFMC maps. Our approach achieves significant improvements over previous methods using randomly initialized models (20 reduction in RMSE). We provide an automated pipeline that enables rapid generation of these LFMC maps across the United States, and demonstrate its effectiveness in two regions recently impacted by wildfire (Eaton and Palisades).

Patrick Beukema, Henry Herzog, Yawen Zhang et al.

We introduce a new dataset and algorithm for fast and efficient coastal distance calculations from Anywhere on Earth (AoE). Existing global coastal datasets are only available at coarse resolution (e.g. 1-4 km) which limits their utility. Publicly available satellite imagery combined with computer vision enable much higher precision. We provide a global coastline dataset at 10 meter resolution, a 100+ fold improvement in precision over existing data. To handle the computational challenge of querying at such an increased scale, we introduce a new library: Layered Iterative Geospatial Hierarchical Terrain-Oriented Unified Search Engine (Lighthouse). Lighthouse is both exceptionally fast and resource-efficient, requiring only 1 CPU and 2 GB of RAM to achieve millisecond online inference, making it well suited for real-time applications in resource-constrained environments.

Henry Herzog, Joshua Hansen, Yawen Zhang et al.

Unsustainable exploitation of the oceans exacerbated by global warming is threatening coastal communities worldwide. Accurate and timely monitoring of maritime activity is an essential step to effective governance and to inform future policy. In support of this complex global-scale effort, we built Atlantes, a deep learning based system that provides the first-ever real-time view of vessel behavior at global scale. Atlantes leverages a series of bespoke transformers to distill a high volume, continuous stream of GPS messages emitted by hundreds of thousands of vessels into easily quantifiable behaviors. The combination of low latency and high performance enables operationally relevant decision-making and successful interventions on the high seas where illegal and exploitative activity is too common. Atlantes is already in use by hundreds of organizations worldwide. Here we provide an overview of the model and infrastructure that enables this system to function efficiently and cost-effectively at global-scale and in real-time.

Xiaoyuan Guo, Kezhen Chen, Jinmeng Rao et al.

We present LOWA, a novel method for localizing objects with attributes effectively in the wild. It aims to address the insufficiency of current open-vocabulary object detectors, which are limited by the lack of instance-level attribute classification and rare class names. To train LOWA, we propose a hybrid vision-language training strategy to learn object detection and recognition with class names as well as attribute information. With LOWA, users can not only detect objects with class names, but also able to localize objects by attributes. LOWA is built on top of a two-tower vision-language architecture and consists of a standard vision transformer as the image encoder and a similar transformer as the text encoder. To learn the alignment between visual and text inputs at the instance level, we train LOWA with three training steps: object-level training, attribute-aware learning, and free-text joint training of objects and attributes. This hybrid training strategy first ensures correct object detection, then incorporates instance-level attribute information, and finally balances the object class and attribute sensitivity. We evaluate our model performance of attribute classification and attribute localization on the Open-Vocabulary Attribute Detection (OVAD) benchmark and the Visual Attributes in the Wild (VAW) dataset, and experiments indicate strong zero-shot performance. Ablation studies additionally demonstrate the effectiveness of each training step of our approach.

Qinggang Zhou, Yawen Zhang, Pengcheng Li et al.

The state-of-the-art deep learning algorithms rely on distributed training systems to tackle the increasing sizes of models and training data sets. Minibatch stochastic gradient descent (SGD) algorithm requires workers to halt forward/back propagations, to wait for gradients aggregated from all workers, and to receive weight updates before the next batch of tasks. This synchronous execution model exposes the overheads of gradient/weight communication among a large number of workers in a distributed training system. We propose a new SGD algorithm, DaSGD (Local SGD with Delayed Averaging), which parallelizes SGD and forward/back propagations to hide 100% of the communication overhead. By adjusting the gradient update scheme, this algorithm uses hardware resources more efficiently and reduces the reliance on the low-latency and high-throughput inter-connects. The theoretical analysis and the experimental results show its convergence rate O(1/sqrt(K)), the same as SGD. The performance evaluation demonstrates it enables a linear performance scale-up with the cluster size.

Songtao Lu, Yawen Zhang, Yunlong Wang et al.

Federated learning opens a number of research opportunities due to its high communication efficiency in distributed training problems within a star network. In this paper, we focus on improving the communication efficiency for fully decentralized federated learning over a graph, where the algorithm performs local updates for several iterations and then enables communications among the nodes. In such a way, the communication rounds of exchanging the common interest of parameters can be saved significantly without loss of optimality of the solutions. Multiple numerical simulations based on large, real-world electronic health record databases showcase the superiority of the decentralized federated learning compared with classic methods.

Qi Liu, Yawen Zhang, Qin Lv et al.

During summer, melt ponds have a significant influence on Arctic sea-ice albedo. The melt pond fraction (MPF) also has the ability to forecast the Arctic sea-ice in a certain period. It is important to retrieve accurate melt pond fraction (MPF) from satellite data for Arctic research. This paper proposes a satellite MPF retrieval model based on the multi-layer neural network, named MPF-NN. Our model uses multi-spectral satellite data as model input and MPF information from multi-site and multi-period visible imagery as prior knowledge for modeling. It can effectively model melt ponds evolution of different regions and periods over the Arctic. Evaluation results show that the MPF retrieved from MODIS data using the proposed model has an RMSE of 3.91% and a correlation coefficient of 0.73. The seasonal distribution of MPF is also consistent with previous results.