REVIEW 6 cited by
Does CLIP Bind Concepts? Probing Compositionality in Large Image 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
Signed reviews
read the original abstract
Large-scale neural network models combining text and images have made incredible progress in recent years. However, it remains an open question to what extent such models encode compositional representations of the concepts over which they operate, such as correctly identifying "red cube" by reasoning over the constituents "red" and "cube". In this work, we focus on the ability of a large pretrained vision and language model (CLIP) to encode compositional concepts and to bind variables in a structure-sensitive way (e.g., differentiating "cube behind sphere" from "sphere behind cube"). To inspect the performance of CLIP, we compare several architectures from research on compositional distributional semantics models (CDSMs), a line of research that attempts to implement traditional compositional linguistic structures within embedding spaces. We benchmark them on three synthetic datasets - single-object, two-object, and relational - designed to test concept binding. We find that CLIP can compose concepts in a single-object setting, but in situations where concept binding is needed, performance drops dramatically. At the same time, CDSMs also perform poorly, with best performance at chance level.
Forward citations
Cited by 6 Pith papers
-
Learning to Reason Iteratively and Parallelly for Complex Visual Reasoning Scenarios
A new attention-based reasoning module combining iterative steps with parallel operation slots improves accuracy on multiple visual question answering benchmarks while staying lightweight and partially interpretable.
-
Impact of Pretraining Word Co-occurrence on Compositional Generalization in Multimodal Models
The accuracy of CLIP and CLIP-based visual question answering models is strongly correlated with how often the concept pair in an image appears together in pretraining captions.
-
VladVA: Discriminative Fine-tuning of LVLMs
Fine-tuning LLaVA-1.5-7B with contrastive loss on short captions plus next-token prediction on long captions yields state-of-the-art zero-shot retrieval and compositionality.
-
Position: We Need An Algorithmic Understanding of Generative AI
The paper argues for a systematic algorithmic understanding of LLMs and presents a case study suggesting that Llama models do not implement BFS or DFS on graph navigation tasks.
-
On the rankability of visual embeddings
Visual embeddings from CLIP and other vision encoders encode ordinal attributes along linear directions, recoverable from as few as two extreme reference images, without full supervision.
-
VSC: Visual Search Compositional Text-to-Image Diffusion Model
VSC improves attribute binding by fusing visual prototypes of per-pair reference images into text embeddings via a lightweight MLP, with a segmentation-based localization loss.
Discussion (0). Continue with ORCID to comment.