Image-to-caption generation on MS-COCO (val)
103CIDErFlamingo
Evaluation Results
| Method | Links | |
|---|---|---|
| FlamingoFinetuning status=Not finetuned, Model size=80B, In-context shots=4-shot2022.11 | 103 | |
| RA-CM3Finetuning status=Not finetuned, Model size=2.7B2022.11 | 89.1 | |
| FlamingoFinetuning status=Not finetuned, Model size=3B, In-context shots=4-shot2022.11 | 85 | |
| Retrieval BaselineFinetuning status=Not finetuned2022.11 | 84.1 | |
| PartiFinetuning status=Not finetuned2022.11 | 83.9 | |
| Vanilla CM3Finetuning status=Not finetuned2022.11 | 71.9 | |
| X-LXMERTFinetuning status=Not finetuned2022.11 | 55.8 | |
| minDALL-EFinetuning status=Not finetuned2022.11 | 48 | |
| ruDALL-E-XLFinetuning status=Not finetuned2022.11 | 38.7 | |
| DALL-E SmallFinetuning status=Not finetuned, Model size=Small2022.11 | 20.2 |