Image-to-image matching on Raven-50 (test)
45AccuracyGPT-4V + CoCoT
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-4V + CoCoTprompting_strategy=CoCoT, zero-shot=true, web_version_as_of=December 30, 20232024.01 | 45 | |
| GPT-4Vprompting_strategy=standard, zero-shot=true, web_version_as_of=December 30, 20232024.01 | 30 | |
| OPENFLAMINGO + CoCoTprompting_strategy=CoCoT, zero-shot=true, beam_search_width=32024.01 | 26 | |
| MMICL + CCoTprompting_strategy=CCoT, zero-shot=true, beam_search_width=82024.01 | 26 | |
| MMICL + CoCoTprompting_strategy=CoCoT, zero-shot=true, beam_search_width=82024.01 | 26 | |
| OPENFLAMINGOprompting_strategy=standard, zero-shot=true, beam_search_width=32024.01 | 24 | |
| OPENFLAMINGO + DDCoTprompting_strategy=DDCoT, zero-shot=true, beam_search_width=32024.01 | 24 | |
| OPENFLAMINGO + CCoTprompting_strategy=CCoT, zero-shot=true, beam_search_width=32024.01 | 24 | |
| MMICLprompting_strategy=standard, zero-shot=true, beam_search_width=82024.01 | 22 | |
| GEMINI + CoCoTprompting_strategy=CoCoT, zero-shot=true, temperature=0.4, TopK=32, TopP=1, max_length=40962024.01 | 22 | |
| GEMINI + CCoTprompting_strategy=CCoT, zero-shot=true, temperature=0.4, TopK=32, TopP=1, max_length=40962024.01 | 20 | |
| GEMINIprompting_strategy=standard, zero-shot=true, temperature=0.4, TopK=32, TopP=1, max_length=40962024.01 | 18 | |
| Random Choiceevaluation_type=baseline2024.01 | 17 | |
| GEMINI + DDCoTprompting_strategy=DDCoT, zero-shot=true, temperature=0.4, TopK=32, TopP=1, max_length=40962024.01 | 12 | |
| MMICL + DDCoTprompting_strategy=DDCoT, zero-shot=true, beam_search_width=82024.01 | 10 |