VILA$^2$: VILA Augmented VILA

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Fang, Yunhao, Zhu, Ligeng, Lu, Yao, Wang, Yan, Molchanov, Pavlo, Kautz, Jan, Cho, Jang Hyun, Pavone, Marco, Han, Song, Yin, Hongxu
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909374154276864
author Fang, Yunhao
Zhu, Ligeng
Lu, Yao
Wang, Yan
Molchanov, Pavlo
Kautz, Jan
Cho, Jang Hyun
Pavone, Marco
Han, Song
Yin, Hongxu
author_facet Fang, Yunhao
Zhu, Ligeng
Lu, Yao
Wang, Yan
Molchanov, Pavlo
Kautz, Jan
Cho, Jang Hyun
Pavone, Marco
Han, Song
Yin, Hongxu
contents While visual language model architectures and training infrastructures advance rapidly, data curation remains under-explored where quantity and quality become a bottleneck. Existing work either crawls extra Internet data with a loose guarantee of quality or distills from black-box proprietary models, e.g., GPT-4V / Gemini that are API frequency and performance bounded. This work enables a VLM to improve itself via data enhancement, exploiting its generative nature. We introduce a simple yet effective VLM augmentation scheme that includes a self-augment step and a specialist-augment step to iteratively improve data quality and hence, model performance. In the self-augment step, the instruction-finetuned VLM recaptions its pretraining caption datasets and then retrains from scratch leveraging refined data. Without any expensive human-in-the-loop annotation, we observe improvements in data quality and downstream accuracy boosts with three self-augmentation rounds -- a viable free lunch to the current VLM training recipe. When self-augmentation saturates, we augment the caption diversity by leveraging specialty skills picked up from instruction finetuning. We finetune VLM specialists from the self-augmented VLM with domain-specific experts, including spatial, grounding, and OCR, to fuse task-aware synthetic data into the pretraining stage. Data quality improvements and hallucination reductions are cross-checked by VLM (GPT-4V, Gemini) and human judges. Combining self-augmentation and specialist-augmented training, VILA$^2$ consistently improves the accuracy on a wide range of benchmarks over the prior art, producing a reusable pretraining dataset that is 300x more cost-efficient than human labeling.
format Preprint
id arxiv_https___arxiv_org_abs_2407_17453
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle VILA$^2$: VILA Augmented VILA
Fang, Yunhao
Zhu, Ligeng
Lu, Yao
Wang, Yan
Molchanov, Pavlo
Kautz, Jan
Cho, Jang Hyun
Pavone, Marco
Han, Song
Yin, Hongxu
Computer Vision and Pattern Recognition
While visual language model architectures and training infrastructures advance rapidly, data curation remains under-explored where quantity and quality become a bottleneck. Existing work either crawls extra Internet data with a loose guarantee of quality or distills from black-box proprietary models, e.g., GPT-4V / Gemini that are API frequency and performance bounded. This work enables a VLM to improve itself via data enhancement, exploiting its generative nature. We introduce a simple yet effective VLM augmentation scheme that includes a self-augment step and a specialist-augment step to iteratively improve data quality and hence, model performance. In the self-augment step, the instruction-finetuned VLM recaptions its pretraining caption datasets and then retrains from scratch leveraging refined data. Without any expensive human-in-the-loop annotation, we observe improvements in data quality and downstream accuracy boosts with three self-augmentation rounds -- a viable free lunch to the current VLM training recipe. When self-augmentation saturates, we augment the caption diversity by leveraging specialty skills picked up from instruction finetuning. We finetune VLM specialists from the self-augmented VLM with domain-specific experts, including spatial, grounding, and OCR, to fuse task-aware synthetic data into the pretraining stage. Data quality improvements and hallucination reductions are cross-checked by VLM (GPT-4V, Gemini) and human judges. Combining self-augmentation and specialist-augmented training, VILA$^2$ consistently improves the accuracy on a wide range of benchmarks over the prior art, producing a reusable pretraining dataset that is 300x more cost-efficient than human labeling.
title VILA$^2$: VILA Augmented VILA
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2407.17453