Tasks SotA Datasets Papers Methods Submit About

Papers With Code 2

A community resource for machine learning research: papers, code, benchmarks, and state-of-the-art results.

Explore

Notable Benchmarks All SotA Datasets Papers Methods

Community

Submit Results About

Data sourced from the PWC Archive (CC-BY-SA 4.0). Built by the community, for the community.

Models/TS-LLaVA-34B

TS-LLaVA-34B

Reported on 31 benchmarks across 6 tasks · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Reasoning24 results

Video Question AnsweringonNExT-QA
Accuracy· 2024-11-17
73.6
best: 85.5 (LinVT-Qwen2-VL (7B))
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video Question AnsweringonMVBench
Accuracy· 2024-11-17
52.6
best: 60.9 (InternVideo2-1B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video Question AnsweringonMSVD-QA
Accuracy· 2024-11-17
79.4
best: 80.3 (Tarsier (34B))
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video Question AnsweringonMSVD-QA
Confidence Score· 2024-11-17
4.1
best: 2.5 (Video LLaMA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video Question AnsweringonTGIF-QA
Accuracy· 2024-11-17
81
best: 82.5 (Tarsier (34B))
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video Question AnsweringonTGIF-QA
Confidence Score· 2024-11-17
4.2
best: 2.3 (Video Chat-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video Question AnsweringonMSRVTT-QA
Accuracy· 2024-11-17
66.2
best: 72.4 (Flash-VStream)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video Question AnsweringonMSRVTT-QA
Confidence Score· 2024-11-17
3.6
best: 1.8 (Video LLaMA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video Question AnsweringonIntentQA
Accuracy· 2024-11-17
67.9
best: 71.5 (ENTER)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video Question AnsweringonEgoSchema (subset)
Accuracy· 2024-11-17
57.8
best: 68.6 (Tarsier (34B))
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video Question AnsweringonActivityNet-QA
Accuracy· 2024-11-17
58.9
best: 61.6 (Tarsier (34B))
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video Question AnsweringonActivityNet-QA
Confidence Score· 2024-11-17
3.5
best: 1.1 (Video LLaMA)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Generative Visual Question AnsweringonVideoInstruct
mean· 2024-11-17
3.38
best: 3.73 (PPLLaVA-7B-dpo)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Generative Visual Question AnsweringonVideoInstruct
gpt-score· 2024-11-17
3.86
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Generative Visual Question AnsweringonVideoInstruct
gpt-score· 2024-11-17
3.55
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Generative Visual Question AnsweringonVideoInstruct
gpt-score· 2024-11-17
3.03
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Generative Visual Question AnsweringonVideoInstruct
gpt-score· 2024-11-17
2.77
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Generative Visual Question AnsweringonVideoInstruct
gpt-score· 2024-11-17
3.69
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video-based Generative Performance BenchmarkingonVideoInstruct
mean· 2024-11-17
3.38
best: 3.73 (PPLLaVA-7B-dpo)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video-based Generative Performance BenchmarkingonVideoInstruct
gpt-score· 2024-11-17
3.86
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video-based Generative Performance BenchmarkingonVideoInstruct
gpt-score· 2024-11-17
3.55
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video-based Generative Performance BenchmarkingonVideoInstruct
gpt-score· 2024-11-17
3.03
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video-based Generative Performance BenchmarkingonVideoInstruct
gpt-score· 2024-11-17
2.77
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Video-based Generative Performance BenchmarkingonVideoInstruct
gpt-score· 2024-11-17
3.69
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066

Natural Language Processing18 results

Question AnsweringonNExT-QA
Accuracy· 2024-11-17
73.6
best: 79.6 (VideoMultiAgent (GPT-4o))
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Question AnsweringonMVBench
Accuracy· 2024-11-17
52.6
best: 60.9 (InternVideo2-1B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Question AnsweringonMSVD-QA
Accuracy· 2024-11-17
79.4
best: 80.3 (Tarsier (34B))
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Question AnsweringonMSVD-QA
Confidence Score· 2024-11-17
4.1
best: 2.5 (Video LLaMA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Question AnsweringonTGIF-QA
Accuracy· 2024-11-17
81
best: 82.5 (Tarsier (34B))
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Question AnsweringonTGIF-QA
Confidence Score· 2024-11-17
4.2
best: 2.3 (Video Chat-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Question AnsweringonMSRVTT-QA
Accuracy· 2024-11-17
66.2
best: 72.4 (Flash-VStream)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Question AnsweringonMSRVTT-QA
Confidence Score· 2024-11-17
3.6
best: 1.8 (Video LLaMA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Question AnsweringonIntentQA
Accuracy· 2024-11-17
67.9
best: 71.5 (ENTER)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Question AnsweringonEgoSchema (subset)
Accuracy· 2024-11-17
57.8
best: 68.6 (Tarsier (34B))
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Question AnsweringonActivityNet-QA
Accuracy· 2024-11-17
58.9
best: 61.6 (Tarsier (34B))
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Question AnsweringonActivityNet-QA
Confidence Score· 2024-11-17
3.5
best: 1.1 (Video LLaMA)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Visual Question Answering (VQA)onVideoInstruct
mean· 2024-11-17
3.38
best: 3.73 (PPLLaVA-7B-dpo)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Visual Question Answering (VQA)onVideoInstruct
gpt-score· 2024-11-17
3.86
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Visual Question Answering (VQA)onVideoInstruct
gpt-score· 2024-11-17
3.55
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Visual Question Answering (VQA)onVideoInstruct
gpt-score· 2024-11-17
3.03
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Visual Question Answering (VQA)onVideoInstruct
gpt-score· 2024-11-17
2.77
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066
Visual Question Answering (VQA)onVideoInstruct
gpt-score· 2024-11-17
3.69
best: 4.21 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066

Other1 result

Video-based Generative Performance Benchmarking (Correctness of Information)onVideoInstruct
gpt-score· 2024-11-17
3.55
best: 3.85 (PPLLaVA-7B)
TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models arXiv:2411.11066