Papers › VILA: On Pre-training for Visual Language Models

VILA: On Pre-training for Visual Language Models

12 Dec 2023CVPR 2024 1arXiv:2312.07533archive 2025-07-28

Ji Lin, Hongxu Yin, Wei Ping, Yao Lu, Pavlo Molchanov, Andrew Tao, Huizi Mao, Jan Kautz, Mohammad Shoeybi, Song Han

Visual language models (VLMs) rapidly progressed with the recent success of large language models. There have been growing efforts on visual instruction tuning to extend the LLM with visual inputs, but lacks an in-depth study of the visual language pre-training process, where the model learns to perform joint modeling on both modalities. In this work, we examine the design options for VLM pre-training by augmenting LLM towards VLM through step-by-step controllable comparisons. We introduce three main findings: (1) freezing LLMs during pre-training can achieve decent zero-shot performance, but lack in-context learning capability, which requires unfreezing the LLM; (2) interleaved pre-training data is beneficial whereas image-text pairs alone are not optimal; (3) re-blending text-only instruction data to image-text data during instruction fine-tuning not only remedies the degradation of text-only tasks, but also boosts VLM task accuracy. With an enhanced pre-training recipe we build VILA, a Visual Language model family that consistently outperforms the state-of-the-art models, e.g., LLaVA-1.5, across main benchmarks without bells and whistles. Multi-modal pre-training also helps unveil appealing properties of VILA, including multi-image reasoning, enhanced in-context learning, and better world knowledge.

PaperPDFConference PDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

efficient-large-model/vila officialmentioned in papermentioned on GitHubpytorchApache-2.0 report
mit-han-lab/llm-awq officialmentioned in papermentioned on GitHubpytorch report
nvlabs/vila officialmentioned in papermentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

In-Context LearningLanguage ModellingVisual Question AnsweringWorld KnowledgeZero-Shot Video Question Answer

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Visual Question Answering MM-Vet VILA-13B GPT-4 score 45.7 #79 of 231 Archive leaderboard report
Zero-Shot Video Question Answer MSVD-QA VILA1.5-40B Accuracy 80.1 #4 of 28 Archive leaderboard report
Zero-Shot Video Question Answer Video-MME VILA-1.5 (34B) Accuracy (%) 64.1 #7 of 11 Archive leaderboard report
Zero-Shot Video Question Answer Video-MME (w/o subs) VILA-1.5 (34B) Accuracy (%) 61.4 #7 of 9 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections