InstructBLIP-13B

Reported on 5 benchmarks across 4 tasks · 2 papers

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

Natural Language Processing4 results

Visual Question Answering (VQA)onIllusionVQA
Accuracy· 2024-03-23
34.25
best: 62.99 (GPT4-Vision 4-shot)
IllusionVQA: A Challenging Optical Illusion Dataset for Vision Language Models arXiv:2403.15952
Visual Question Answering (VQA)onBenchLMM
GPT-3.5 score· 2023-05-11
45.03
best: 58.37 (GPT-4V)
InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning arXiv:2305.06500
Instruction FollowingonLLaVA-Bench
avg score· 2023-05-11
58.2
best: 85.7 (CuMo-7B)
InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning arXiv:2305.06500
Visual Question AnsweringonBenchLMM
GPT-3.5 score· 2023-05-11
45.03
best: 58.37 (GPT-4V)
InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning arXiv:2305.06500

Computer Vision1 result

Object LocalizationonIllusionVQA
Accuracy· 2024-03-23
24.3
best: 49.7 (GPT4-Vision 4-shot+CoT)
IllusionVQA: A Challenging Optical Illusion Dataset for Vision Language Models arXiv:2403.15952