| output path: zeronlg-4langs-vc/evaluations_caption/vatex/en | |
| Creating model from zeronlg-4langs-vc | |
| Evaluation settings: {'lang': 'en', 'num_beams': 3, 'max_length': 20, 'min_length': 3, 'repetition_penalty': 1.0} | |
| Load dataset from data/annotations/vatex/en/test.json | |
| Load CLIP embs from data/annotations/vatex/clip-ViT-B-32_test.pkl | |
| There are 5792 vision inputs | |
| Save caption results to zeronlg-4langs-vc/evaluations_caption/vatex/en/test_captions.json | |
| Save detailed scores to zeronlg-4langs-vc/evaluations_caption/vatex/en/test_detailed_scores.json | |
| Save scores to zeronlg-4langs-vc/evaluations_caption/vatex/en/test_scores.json | |
| [test] Bleu_1 0.42168940639806163 | |
| [test] Bleu_2 0.24590583953294565 | |
| [test] Bleu_3 0.12531948515900065 | |
| [test] Bleu_4 0.06335825765892944 | |
| [test] METEOR 0.1165194842146988 | |
| [test] ROUGE_L 0.2934526708651432 | |
| [test] CIDEr 0.09084566361351876 | |
| [test] SPICE 0.03951599733101809 | |