Enregistré dans:
Détails bibliographiques
Auteurs principaux: Karamcheti, Siddharth, Nair, Suraj, Balakrishna, Ashwin, Liang, Percy, Kollar, Thomas, Sadigh, Dorsa
Format: Preprint
Publié: 2024
Sujets:
Accès en ligne:https://arxiv.org/abs/2402.07865
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866917679019851776
author Karamcheti, Siddharth
Nair, Suraj
Balakrishna, Ashwin
Liang, Percy
Kollar, Thomas
Sadigh, Dorsa
author_facet Karamcheti, Siddharth
Nair, Suraj
Balakrishna, Ashwin
Liang, Percy
Kollar, Thomas
Sadigh, Dorsa
contents Visually-conditioned language models (VLMs) have seen growing adoption in applications such as visual dialogue, scene understanding, and robotic task planning; adoption that has fueled a wealth of new models such as LLaVa, InstructBLIP, and PaLI-3. Despite the volume of new releases, key design decisions around image preprocessing, architecture, and optimization are under-explored, making it challenging to understand what factors account for model performance $-$ a challenge further complicated by the lack of objective, consistent evaluations. To address these gaps, we first compile a suite of standardized evaluations spanning visual question answering, object localization, and challenge sets that probe properties such as hallucination; evaluations that provide fine-grained insight VLM capabilities. Second, we rigorously investigate VLMs along key design axes, including pretrained visual representations and training from base vs. instruct-tuned language models, amongst others. We couple our analysis with three resource contributions: (1) a unified framework for evaluating VLMs, (2) optimized, flexible training code, and (3) checkpoints for all models, including a family of VLMs at the 7-13B scale that strictly outperform InstructBLIP and LLaVa v1.5, the state-of-the-art in open VLMs.
format Preprint
id arxiv_https___arxiv_org_abs_2402_07865
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Prismatic VLMs: Investigating the Design Space of Visually-Conditioned Language Models
Karamcheti, Siddharth
Nair, Suraj
Balakrishna, Ashwin
Liang, Percy
Kollar, Thomas
Sadigh, Dorsa
Computer Vision and Pattern Recognition
Artificial Intelligence
Computation and Language
Machine Learning
Visually-conditioned language models (VLMs) have seen growing adoption in applications such as visual dialogue, scene understanding, and robotic task planning; adoption that has fueled a wealth of new models such as LLaVa, InstructBLIP, and PaLI-3. Despite the volume of new releases, key design decisions around image preprocessing, architecture, and optimization are under-explored, making it challenging to understand what factors account for model performance $-$ a challenge further complicated by the lack of objective, consistent evaluations. To address these gaps, we first compile a suite of standardized evaluations spanning visual question answering, object localization, and challenge sets that probe properties such as hallucination; evaluations that provide fine-grained insight VLM capabilities. Second, we rigorously investigate VLMs along key design axes, including pretrained visual representations and training from base vs. instruct-tuned language models, amongst others. We couple our analysis with three resource contributions: (1) a unified framework for evaluating VLMs, (2) optimized, flexible training code, and (3) checkpoints for all models, including a family of VLMs at the 7-13B scale that strictly outperform InstructBLIP and LLaVa v1.5, the state-of-the-art in open VLMs.
title Prismatic VLMs: Investigating the Design Space of Visually-Conditioned Language Models
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Computation and Language
Machine Learning
url https://arxiv.org/abs/2402.07865