metadata
library_name: transformers
license: mit
base_model: facebook/mbart-large-50
tags:
- generated_from_trainer
model-index:
- name: scenegraph_image_captioning
results: []
scenegraph_image_captioning
This model is a fine-tuned version of facebook/mbart-large-50 on an ViVG-Scene-Graph dataset.
It achieves the following results on the evaluation set:
- Loss: 3.3994
- Bleu4: 10.148
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 3e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 350
- num_epochs: 30
Training results
| Training Loss | Epoch | Step | Validation Loss | Bleu4 |
|---|---|---|---|---|
| 3.6494 | 1.0 | 4625 | 3.6095 | 8.3269 |
| 3.4809 | 2.0 | 9250 | 3.4953 | 9.0764 |
| 3.2703 | 3.0 | 13875 | 3.4439 | 9.185 |
| 3.3438 | 4.0 | 18500 | 3.4134 | 10.023 |
| 3.2965 | 5.0 | 23125 | 3.3911 | 10.0925 |
| 3.2711 | 6.0 | 27750 | 3.3830 | 10.4298 |
| 3.1469 | 7.0 | 32375 | 3.3745 | 10.5282 |
| 3.1113 | 8.0 | 37000 | 3.3760 | 10.292 |
| 3.0507 | 9.0 | 41625 | 3.3739 | 10.7014 |
| 3.0794 | 10.0 | 46250 | 3.3767 | 10.4073 |
| 3.0509 | 11.0 | 50875 | 3.3793 | 10.3126 |
| 3.0570 | 12.0 | 55500 | 3.3880 | 10.2934 |
| 2.9785 | 13.0 | 60125 | 3.3865 | 10.2243 |
| 2.9636 | 14.0 | 64750 | 3.3994 | 10.148 |
Framework versions
- Transformers 5.10.2
- Pytorch 2.11.0+cu128
- Datasets 4.0.0
- Tokenizers 0.22.2