AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

Language Quantized AutoEncoders: Towards Unsupervised Text-Image Alignment

Hao Liu, Wilson Yan, Pieter Abbeel

NeurIPS 2023 · Vision-language and multimodal › CLIP and zero-shot · approach: Recaptioning and text-supervised pretraining · 5 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
remove L2 normalization in nearest-neighbor code search (ablation)LQAE · ImageNet linear classificationaccuracy: 35.60 -> 30.30Hurt
remove L2 normalization in nearest-neighbor code search (ablation)LQAE · miniImageNet 5-way avgaccuracy: 29.04 -> 27.42Hurt
use randomly initialized instead of pretrained RoBERTa (ablation)LQAE · miniImageNet 5-way avgaccuracy: 29.04 -> 0.51Hurt
add entropy regularization 0.5 on quantized codes (ablation)LQAE · ImageNet linear classificationaccuracy: 35.60 -> 30.70Hurt
remove BERT loss (weight 0.00) (ablation)LQAE · miniImageNet 2-way avgaccuracy: 53.97 -> 52.45No clear effect
raise BERT loss weight to 1.00 (ablation)LQAE · linear vs 5-way few-shotaccuracy: 35.60 -> 36.90 (linear); 29.04 -> 20.93 (5-way)Mixed
vector quantization before decoder input (decoder STE) (ablation)LQAE · miniImageNet 5-way avgaccuracy: 29.04 -> 30.01No clear effect
keep only 50% of LQAE codes in the GPT-3 prompt (ablation)LQAE · miniImageNet 5-way avgaccuracy: 29.04 -> 5.56Hurt
keep 75% of LQAE codes in the prompt (ablation)LQAE · miniImageNet 2-way avgaccuracy: 53.97 -> 50.56Hurt
use smaller GPT-3 (Curie 6.7B instead of Davinci 175B) (ablation)LQAE · miniImageNet 2-way avgaccuracy: 53.97 -> 46.55Hurt
use smallest GPT-3 (Babbage 1.3B) (ablation)LQAE · miniImageNet 5-way avgaccuracy: 29.04 -> 14.70Hurt
assign VQ-VAE tokens to RoBERTa codebook instead of LQAE training (ablation)LQAE · miniImageNet 2-way avgaccuracy: 53.97 -> 3.24Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy