Pith. sign in

REVIEW 4 cited by

CLOSURE: Assessing Systematic Generalization of CLEVR Models

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 1912.05783 v2 pith:LLHKVKW3 submitted 2019-12-12 cs.AI cs.LG

classification cs.AIcs.LG
keywords modelsclevrclosuregeneralizationnovelsystematiccubeexperiments
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

The CLEVR dataset of natural-looking questions about 3D-rendered scenes has recently received much attention from the research community. A number of models have been proposed for this task, many of which achieved very high accuracies of around 97-99%. In this work, we study how systematic the generalization of such models is, that is to which extent they are capable of handling novel combinations of known linguistic constructs. To this end, we test models' understanding of referring expressions based on matching object properties (such as e.g. "another cube that is the same size as the brown cube") in novel contexts. Our experiments on the thereby constructed CLOSURE benchmark show that state-of-the-art models often do not exhibit systematicity after being trained on CLEVR. Surprisingly, we find that an explicitly compositional Neural Module Network model also generalizes badly on CLOSURE, even when it has access to the ground-truth programs at test time. We improve the NMN's systematic generalization by developing a novel Vector-NMN module architecture with vector-valued inputs and outputs. Lastly, we investigate how much few-shot transfer learning can help models that are pretrained on CLEVR to adapt to CLOSURE. Our few-shot learning experiments contrast the adaptation behavior of the models with intermediate discrete programs with that of the end-to-end continuous models.

Discussion (0). Sign in to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. ShapeCodeBench: A Renewable Benchmark for Perception-to-Program Reconstruction of Synthetic Shape Scenes

    cs.CV 2026-05 accept novelty 6.0 of 10

    ShapeCodeBench introduces a renewable benchmark for perception-to-program reconstruction of synthetic shapes, with evaluations showing low exact-match performance from current models and heuristics.

  2. Disentanglement-Based Equivariant Learning for Compositional VQA

    cs.CV 2026-06 unverdicted novelty 5.0 of 10

    DEAL disentangles concepts from images and text using causal interventions and enforces equivariance on compositional transformations to boost generalization in VQA, outperforming prior methods on CLEVR-CoGenT and GQA-SGL.

  3. Long Story Short: Disentangling Compositionality and Long-Caption Understanding in Contrastive VLMs

    cs.CV 2025-09 unverdicted novelty 5.0 of 10

    Empirical study shows bidirectional but sensitive relationship between compositionality and long-caption understanding in VLMs, promoted by high-quality grounded data and affected by architectural choices like frozen ...

  4. Modality-Aware Feature Matching in Visual and Vision-Language Applications: A Comprehensive Survey

    cs.CV 2025-07 conditional novelty 2.0 of 10

    A survey organizing feature matching research by modality, from SIFT to transformer-based dense matchers and vision-language models.

Pith tools