Despite the unprecedented progress in the field of 3D generation, current systems still often fail to produce high-quality 3D assets that are visually appealing and geometrically and semantically consistent across multiple viewpoints. To effectively assess the quality of the generated 3D data, there is a need for a reliable 3D evaluation tool. Unfortunately, existing 3D evaluation metrics often overlook the geometric quality of generated assets or merely rely on black-box multimodal large language models for coarse assessment. In this paper, we introduce Eval3D, a fine-grained, interpretable evaluation tool that can faithfully evaluate the quality of generated 3D assets based on various distinct yet complementary criteria. Our key observation is that many desired properties of 3D generation, such as semantic and geometric consistency, can be effectively captured by measuring the consistency among various foundation models and tools. We thus leverage a diverse set of models and tools as probes to evaluate the inconsistency of generated 3D assets across different aspects. Compared to prior work, Eval3D provides pixel-wise measurement, enables accurate 3D spatial feedback, and aligns more closely with human judgments. We comprehensively evaluate existing 3D generation models using Eval3D and highlight the limitations and challenges of current models. Project page: http://eval3d.github.io.

Jiao Sun, Deqing Fu, Yushi Hu, Su Wang, Royi Rassin, Da-Cheng Juan, Dana Alon, Charles Herrmann, Sjoerd Van Steenkiste, Ranjay Krishna, Cyrus Rashtchian. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

While recent Large Language Models (LLMs) have proven useful in answering user queries, they are prone to hallucination, and their responses often lack credibility due to missing references to reliable sources.An intuitive solution to these issues would be to include in-text citations referring to external documents as evidence.While previous works have directly prompted LLMs to generate in-text citations, their performances are far from satisfactory, especially when it comes to smaller LLMs.In this work, we propose an effective training framework using fine-grained rewards to teach LLMs to generate highly supportive and relevant citations, while ensuring the correctness of their responses.We also conduct a systematic analysis of applying these fine-grained rewards to common LLM training strategies, demonstrating its advantage over conventional practices.We conduct extensive experiments on Question Answering (QA) datasets taken from the ALCE benchmark and validate the model's generalizability using EXPERTQA.On LLaMA-2-7B, the incorporation of fine-grained rewards achieves the best performance among the baselines, even surpassing that of GPT-3.5-turbo.1

Human feedback plays a critical role in learning and refining reward models for text-to-image generation, but the optimal form the feedback should take for learning an accurate reward function has not been conclusively established. This paper investigates the effectiveness of fine-grained feedback which captures nuanced distinctions in image quality and prompt-alignment, compared to traditional coarse-grained feedback (for example, thumbs up/down or ranking between a set of options). While fine-grained feedback holds promise, particularly for systems catering to diverse societal preferences, we show that demonstrating its superiority to coarse-grained feedback is not automatic. Through experiments on real and synthetic preference data, we surface the complexities of building effective models due to the interplay of model choice, feedback type, and the alignment between human judgment and computational interpretation. We identify key challenges in eliciting and utilizing fine-grained feedback, prompting a reassessment of its assumed benefits and practicality. Our findings -- e.g., that fine-grained feedback can lead to worse models for a fixed budget, in some settings; however, in controlled settings with known attributes, fine grained rewards can indeed be more helpful -- call for careful consideration of feedback attributes and potentially beckon novel modeling approaches to appropriately unlock the potential value of fine-grained feedback in-the-wild.

Solving complex visual tasks such as “Who invented the musical instrument on the right?” involves a composition of skills: understanding space, recognizing instruments, and also retrieving prior knowledge. Recent work shows promise by decomposing such tasks using a large language model (LLM) into an executable program that invokes specialized vision models. However, generated programs are error-prone: they omit necessary steps, include spurious ones, and are unable to recover when the specialized models give incor-rect outputs. Moreover, they require loading multiple models, incurring high latency and computation costs. We propose Visual Program Distillation (VPD), an instruction tuning framework that produces a vision-language model (VLM) ca-pable of solving complex visual tasks with a single forward pass. VPD distills the reasoning ability of LLMs by using them to sample multiple candidate programs, which are then executed and verified to identify a correct one. It translates each correct program into a language description of the reasoning steps, which are then distilled into a VLM. Exten-sive experiments show that VPD improves the VLM's ability to count, understand spatial relations, and reason compositionally. Our VPD-trained PaLI-X outperforms all prior VLMs, achieving state-of-the-art performance across complex vision tasks, including MMBench, OK-VQA, A-OKVQA, TallyQA, POPE, and Hateful Memes. An evaluation with human annotators also confirms that VPD improves model response factuality and consistency. Finally, experiments on content moderation demonstrate that VPD is also helpful for adaptation to real-world applications with limited data.

Evaluating text-to-image models is notoriously difficult. A strong recent approach for assessing text-image faithfulness is based on QG/A (question generation and answering), which uses pre-trained foundational models to automatically generate a set of questions and answers from the prompt, and output images are scored based on whether these answers extracted with a visual question answering model are consistent with the prompt-based answers. This kind of evaluation is naturally dependent on the quality of the underlying QG and VQA models. We identify and address several reliability challenges in existing QG/A work: (a) QG questions should respect the prompt (avoiding hallucinations, duplications, and omissions) and (b) VQA answers should be consistent (not asserting that there is no motorcycle in an image while also claiming the motorcycle is blue). We address these issues with Davidsonian Scene Graph (DSG), an empirically grounded evaluation framework inspired by formal semantics, which is adaptable to any QG/A frameworks. DSG produces atomic and unique questions organized in dependency graphs, which (i) ensure appropriate semantic coverage and (ii) sidestep inconsistent answers. With extensive experimentation and human evaluation on a range of model configurations (LLM, VQA, and T2I), we empirically demonstrate that DSG addresses the challenges noted above. Finally, we present DSG-1k, an open-sourced evaluation benchmark that includes 1,060 prompts, covering a wide range of fine-grained semantic categories with a balanced distribution. We release the DSG-1k prompts and the corresponding DSG questions.

Humans draw to facilitate reasoning: we draw auxiliary lines when solving geometry problems; we mark and circle when reasoning on maps; we use sketches to amplify our ideas and relieve our limited-capacity working memory. However, such actions are missing in current multimodal language models (LMs). Current chain-of-thought and tool-use paradigms only use text as intermediate reasoning steps. In this work, we introduce Sketchpad, a framework that gives multimodal LMs a visual sketchpad and tools to draw on the sketchpad. The LM conducts planning and reasoning according to the visual artifacts it has drawn. Different from prior work, which uses text-to-image models to enable LMs to draw, Sketchpad enables LMs to draw with lines, boxes, marks, etc., which is closer to human sketching and better facilitates reasoning. \name can also use specialist vision models during the sketching process (e.g., draw bounding boxes with object detection models, draw masks with segmentation models), to further enhance visual perception and reasoning. We experiment on a wide range of math tasks (including geometry, functions, graph, chess) and complex visual reasoning tasks. Sketchpad substantially improves performance on all tasks over strong base models with no sketching, yielding an average gain of 12.7% on math tasks, and 8.6% on vision tasks. GPT-4o with Sketchpad sets a new state of the art on all tasks, including V*Bench (80.3%), BLINK spatial reasoning (83.9%), and visual correspondence (80.8%). We will release all code and data.

Aligning language models (LMs) to human preferences has emerged as a critical pursuit, enabling these models to better serve diverse user needs. Existing methods primarily focus on optimizing LMs for a single reward function, limiting their adaptability to varied objectives. Here, we propose $\textbf{multi-objective decoding~(MOD)}$, a decoding-time algorithm that outputs the next token from a linear combination of predictions of all base models, for any given weighting over different objectives. We exploit a common form among a family of $f$-divergence regularized alignment approaches (such as PPO, DPO, and their variants) to identify a closed-form solution by Legendre transform, and derive an efficient decoding strategy. Theoretically, we show why existing approaches can be sub-optimal even in natural settings and obtain optimality guarantees for our method. Empirical results demonstrate the effectiveness of the algorithm. For example, compared to a parameter-merging baseline, MOD achieves 12.8\% overall reward improvement when equally optimizing towards $3$ objectives. Moreover, we experiment with MOD on combining three fully-finetuned LMs of different model sizes, each aimed at different objectives such as safety, coding, and general user preference. Unlike traditional methods that require careful curation of a mixture of datasets to achieve comprehensive improvement, we can quickly experiment with preference weightings using MOD to find the best combination of models. Our best combination reduces toxicity on Toxigen to nearly 0\% and achieves 7.9--33.3\% improvement across three other metrics ($\textit{i.e.}$, Codex@1, GSM-COT, BBH-COT).

Knowledge-based visual question answering (VQA) involves questions that require world knowledge beyond the image to yield the correct answer. Large language models (LMs) like GPT-3 are particularly helpful for this task because of their strong knowledge retrieval and reasoning capabilities. To enable LM to understand images, prior work uses a captioning model to convert images into text. However, when summarizing an image in a single caption sentence, which visual entities to describe are often underspecified. Generic image captions often miss visual details essential for the LM to answer visual questions correctly. To address this challenge, we propose PromptCap (Prompt-guided image Captioning), a captioning model designed to serve as a better connector between images and black-box LMs. Different from generic captions, PromptCap takes a natural-language prompt to control the visual entities to describe in the generated caption. The prompt contains a question that the caption should aid in answering. To avoid extra annotation, PromptCap is trained by examples synthesized with GPT-3 and existing datasets. We demonstrate Prompt-Cap’s effectiveness on an existing pipeline in which GPT-3 is prompted with image captions to carry out VQA. Prompt-Cap outperforms generic captions by a large margin and achieves state-of-the-art accuracy on knowledge-based VQA tasks (60.4% on OK-VQA and 59.6% on A-OKVQA). Zero-shot results on WebQA show that PromptCap generalizes well to unseen domains.1

Despite thousands of researchers, engineers, and artists actively working on improving text-to-image generation models, systems often fail to produce images that accurately align with the text inputs. We introduce TIFA (Text-to-Image Faithfulness evaluation with question Answering), an automatic evaluation metric that measures the faithfulness of a generated image to its text input via visual question answering (VQA). Specifically, given a text input, we automatically generate several question-answer pairs using a language model. We calculate image faithfulness by checking whether existing VQA models can answer these questions using the generated image. TIFA is a reference-free metric that allows for fine-grained and interpretable evaluations of generated images. TIFA also has better correlations with human judgments than existing metrics. Based on this approach, we introduce TIFA v1.0, a benchmark consisting of 4K diverse text inputs and 25K questions across 12 categories (object, counting, etc.). We present a comprehensive evaluation of existing text-to-image models using TIFA v1.0 and highlight the limitations and challenges of current models. For instance, we find that current text-to-image models, despite doing well on color and material, still struggle in counting, spatial relations, and composing multiple objects. We hope our benchmark will help carefully measure the research progress in text-to-image synthesis and provide valuable insights for further research.1

Language models (LMs) often exhibit undesirable text generation behaviors, including generating false, toxic, or irrelevant outputs. Reinforcement learning from human feedback (RLHF)---where human preference judgments on LM outputs are transformed into a learning signal---has recently shown promise in addressing these issues. However, such holistic feedback conveys limited information on long text outputs; it does not indicate which aspects of the outputs influenced user preference; e.g., which parts contain what type(s) of errors. In this paper, we use fine-grained human feedback (e.g., which sentence is false, which sub-sentence is irrelevant) as an explicit training signal. We introduce Fine-Grained RLHF, a framework that enables training and learning from reward functions that are fine-grained in two respects: (1) density, providing a reward after every segment (e.g., a sentence) is generated; and (2) incorporating multiple reward models associated with different feedback types (e.g., factual incorrectness, irrelevance, and information incompleteness). We conduct experiments on detoxification and long-form question answering to illustrate how learning with this reward function leads to improved performance, supported by both automatic and human evaluation. Additionally, we show that LM behaviors can be customized using different combinations of fine-grained reward models. We release all data, collected human feedback, and codes at https://FineGrainedRLHF.github.io.

Query-by-example (QbE) speech search is the task of matching spoken queries to utterances within a search collection. In low-or zero-resource settings, QbE search is often addressed with approaches based on dynamic time warping (DTW). Recent work has found that methods based on acoustic word embeddings (AWEs) can improve both performance and search speed. However, prior work on AWE-based QbE has primarily focused on English data and with single-word queries. In this work, we generalize AWE training to spans of words, producing acoustic span embeddings (ASE), and explore the application of ASE to QbE with arbitrary-length queries in multiple unseen languages. We consider the commonly used setting where we have access to labeled data in other languages (in our case, several low-resource languages) distinct from the unseen test languages. We evaluate our approach on the QUESST 2015 QbE tasks, finding that multilingual ASE-based search is much faster than DTW-based search and outperforms the best previously published results on this task.

Acoustic word embeddings (AWEs) are vector representations of spoken word segments.AWEs can be learned jointly with embeddings of character sequences, to generate phonetically meaningful embeddings of written words, or acoustically grounded word embeddings (AGWEs).Such embeddings have been used to improve speech retrieval, recognition, and spoken term discovery.In this work, we extend this idea to multiple low-resource languages.We jointly train an AWE model and an AGWE model, using phonetically transcribed data from multiple languages.The pre-trained models can then be used for unseen zero-resource languages, or fine-tuned on data from low-resource languages.We also investigate distinctive features, as an alternative to phone labels, to better share cross-lingual information.We test our models on word discrimination tasks for twelve languages.When trained on eleven languages and tested on the remaining unseen language, our model outperforms traditional unsupervised approaches like dynamic time warping.After fine-tuning the pre-trained models on one hour or even ten minutes of data from a new language, performance is typically much better than training on only the target-language data.We also find that phonetic supervision improves performance over character sequences, and that distinctive feature supervision is helpful in handling unseen phones in the target language.