Papers › PLLaVA : Parameter-free LLaVA Extension from Images to Videos for Video Dense Captioning

PLLaVA : Parameter-free LLaVA Extension from Images to Videos for Video Dense Captioning

25 Apr 2024arXiv 2024 4arXiv:2404.16994archive 2025-07-28

Lin Xu, Yilin Zhao, Daquan Zhou, Zhijie Lin, See Kiong Ng, Jiashi Feng

Vision-language pre-training has significantly elevated performance across a wide range of image-language applications. Yet, the pre-training process for video-related tasks demands exceptionally large computational and data resources, which hinders the progress of video-language models. This paper investigates a straight-forward, highly efficient, and resource-light approach to adapting an existing image-language pre-trained model for dense video understanding. Our preliminary experiments reveal that directly fine-tuning pre-trained image-language models with multiple frames as inputs on video datasets leads to performance saturation or even a drop. Our further investigation reveals that it is largely attributed to the bias of learned high-norm visual features. Motivated by this finding, we propose a simple but effective pooling strategy to smooth the feature distribution along the temporal dimension and thus reduce the dominant impacts from the extreme features. The new model is termed Pooling LLaVA, or PLLaVA in short. PLLaVA achieves new state-of-the-art performance on modern benchmark datasets for both video question-answer and captioning tasks. Notably, on the recent popular VideoChatGPT benchmark, PLLaVA achieves a score of 3.48 out of 5 on average of five evaluated dimensions, exceeding the previous SOTA results from GPT4V (IG-VLM) by 9%. On the latest multi-choice benchmark MVBench, PLLaVA achieves 58.1% accuracy on average across 20 sub-tasks, 14.5% higher than GPT4V (IG-VLM). Code is available at https://pllava.github.io/

PaperPDFConference PDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

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

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 2 code samples harvested from 1 repository linked to this paper: 0 ran · 2 unverified.

By repository: official repository: 2 samples from 1 repository, 0 ran (0 with no instrument failure; 0 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

magic-research/PLLaVA officialmentioned in papermentioned on GitHubpytorchSyntology: 2 samples, 0 ran · 2 unverified 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

2 samples harvested: 0 ran · 2 unverified. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

2unverified

Licence: 2 of the 2 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from magic-research/PLLaVA. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity.

magic-research/PLLaVA official repository 2 samples, 0 ran · 2 unverified
get_state_maybe_zero_3 tasks/train/train_pllava_nframe_accel.py unverified no licence file found · pointer only · 770e3b5004c211d6 · report
maybe_zero_3 in 74 papers tasks/train/train_pllava_nframe_accel.py unverified no licence file found · pointer only · 735025744c1ab0cf · report

Tasks

Dense CaptioningVideo Question AnsweringVideo UnderstandingVideo-based Generative Performance BenchmarkingVideo-based Generative Performance Benchmarking (Consistency)Video-based Generative Performance Benchmarking (Contextual Understanding)Video-based Generative Performance Benchmarking (Correctness of Information)Video-based Generative Performance Benchmarking (Detail Orientation))Video-based Generative Performance Benchmarking (Temporal Understanding)Zero-Shot Video Question Answer

1 archive task tag without a task page not shown.

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Video Question Answering MVBench PLLaVA Avg. 58.1 #11 of 22 Archive leaderboard report
Video Question Answering TVBench PLLaVA-34B Average Accuracy 42.3 #20 of 28 Archive leaderboard report
Video Question Answering TVBench PLLaVA-13B Average Accuracy 36.4 #25 of 28 Archive leaderboard report
Video Question Answering TVBench PLLaVA-7B Average Accuracy 34.9 #28 of 28 Archive leaderboard report
Video-based Generative Performance Benchmarking VideoInstruct PLLaVA-34B Consistency 3.25 #4 of 23 Archive leaderboard report
Video-based Generative Performance Benchmarking VideoInstruct PLLaVA-34B Contextual Understanding 3.90 #4 of 23 Archive leaderboard report
Video-based Generative Performance Benchmarking VideoInstruct PLLaVA-34B Correctness of Information 3.60 #4 of 23 Archive leaderboard report
Video-based Generative Performance Benchmarking VideoInstruct PLLaVA-34B Detail Orientation 3.20 #4 of 23 Archive leaderboard report
Video-based Generative Performance Benchmarking VideoInstruct PLLaVA-34B Temporal Understanding 2.67 #4 of 23 Archive leaderboard report
Video-based Generative Performance Benchmarking VideoInstruct PLLaVA-34B mean 3.32 #4 of 23 Archive leaderboard report
Video-based Generative Performance Benchmarking (Consistency) VideoInstruct PLLaVA-34B gpt-score 3.25 #5 of 18 Archive leaderboard report
Video-based Generative Performance Benchmarking (Contextual Understanding) VideoInstruct PLLaVA-34B gpt-score 3.9 #2 of 18 Archive leaderboard report
Video-based Generative Performance Benchmarking (Correctness of Information) VideoInstruct PLLaVA-34B gpt-score 3.60 #2 of 18 Archive leaderboard report
Video-based Generative Performance Benchmarking (Detail Orientation)) VideoInstruct PLLaVA-34B gpt-score 3.20 #2 of 18 Archive leaderboard report
Video-based Generative Performance Benchmarking (Temporal Understanding) VideoInstruct PLLaVA-34B gpt-score 2.67 #6 of 18 Archive leaderboard report
Zero-Shot Video Question Answer ActivityNet-QA PLLaVA (34B) Accuracy 60.9 #2 of 28 Archive leaderboard report
Zero-Shot Video Question Answer ActivityNet-QA PLLaVA (34B) Confidence Score 3.7 #2 of 28 Archive leaderboard report
Zero-Shot Video Question Answer MSRVTT-QA PLLaVA (34B) Accuracy 68.7 #2 of 30 Archive leaderboard report
Zero-Shot Video Question Answer MSRVTT-QA PLLaVA (34B) Confidence Score 3.6 #2 of 30 Archive leaderboard report
Zero-Shot Video Question Answer MSVD-QA PLLaVA (34B) Accuracy 79.9 #5 of 28 Archive leaderboard report
Zero-Shot Video Question Answer MSVD-QA PLLaVA (34B) Confidence Score 4.2 #5 of 28 Archive leaderboard report
Zero-Shot Video Question Answer TGIF-QA PLLaVA Accuracy 80.6 #4 of 14 Archive leaderboard report
Zero-Shot Video Question Answer TGIF-QA PLLaVA Confidence Score 4.3 #4 of 14 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