IC-739GPT-3.5-turbo-0613 combined with CLIP produces more faithful concept-salience pseudo-labels than LLaMA-2-13B-Chat, InstructBLIP, or LLaVA-1.5B on most of five datasets

Aditya Chattopadhyay, Kwan Ho Ryan Chan, Rene Vidal

SourceBootstrapping Variational Information Pursuit with Large Language and Vision Models for Interpretable Image Classification

The paper compares four mechanisms for generating pseudo-labels indicating whether a semantic concept is salient for recognizing a given image class. GPT-3.5-turbo-0613 (with CLIP dot products) achieves 0.79–0.86 accuracy across ImageNet, Places365, CUB-200, CIFAR-10, and CIFAR-100. LLaMA-2-13B-Chat with CLIP is worse on most datasets (0.62 on CUB-200 vs 0.79 for GPT). InstructBLIP (FlanT5XL) is competitive with GPT+CLIP on most datasets but worse on CUB-200 (0.68 vs 0.79). LLaVA-1.5B is the weakest at 0.49–0.69. The authors note that GPT-4 was substituted for GPT-3.5 on CUB-200 due to domain-specificity, so that column may reflect GPT-4 performance.

Evidence
correlational
Key metric
GPT+CLIP accuracy: 0.86, 0.82, 0.79, 0.81, 0.79; LLaMA2+CLIP: 0.81, 0.75, 0.62, 0.78, 0.84; InstructBLIP (FlanT5XL): 0.85, 0.86, 0.68, 0.82, 0.84; LLaVA-1.5B: 0.63, 0.66, 0.49, 0.69, 0.66
Caveat
Ground truth is self-annotated by the authors (10,000 evaluations over two weeks). The CUB-200 column for GPT+CLIP uses GPT-4 rather than GPT-3.5-turbo-0613, as stated in the paper. The authors note that using VLMs (InstructBLIP, LLaVA) to generate pseudo-labels for the full ImageNet training set is computationally infeasible (~37,344 GPU hours for InstructBLIP vs ~600 for GPT+CLIP).
Model
GPT-3.5 / ChatGPT-3.5 GPT-3.5-turbo-0613, GPT-4 / ChatGPT4 / GPT-4 Code Interpreter / GPT-4 Technical Report, Llama 2 / Llama 2 base Llama-2-13B-Chat, InstructBLIP, LLaVA, CLIP / CLIP-ViT (LC)
Datasets
ImageNet-1k / ImageNet / ImageNet-1k-val / ImageNet-Val [eval], Places365 [eval], CUB-200-2011 / CUB200 [eval], CIFAR-10 [eval], CIFAR-100 [eval]
Related findings
IC-737, IC-738
Extraction
automatic-extraction