IC_ver6d_coco_swin_gpt2_50Bpc_1e

This model is a fine-tuned version of VK246/IC_ver6c_coco_swin_gpt2_50Apc_1e on the coco dataset. It achieves the following results on the evaluation set:

• Loss: 0.7897
• Rouge1: 42.1846
• Rouge2: 16.6343
• Rougel: 38.2927
• Rougelsum: 38.2913
• Bleu: 10.4949
• Gen Len: 11.3063

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: 5e-05
• train_batch_size: 96
• eval_batch_size: 96
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Bleu	Gen Len
0.7741	0.17	500	0.8245	40.8216	15.539	37.1564	37.1436	9.6536	11.3063
0.7813	0.34	1000	0.8155	41.2707	15.9841	37.4357	37.4254	10.033	11.3063
0.782	0.51	1500	0.8061	41.6066	16.0222	37.802	37.8009	9.9619	11.3063
0.79	0.68	2000	0.7985	41.6881	16.2489	37.9303	37.9178	10.3074	11.3063
0.7888	0.85	2500	0.7929	42.1145	16.5388	38.2401	38.2324	10.3546	11.3063

Framework versions

• Transformers 4.30.2
• Pytorch 2.0.1+cu118
• Datasets 2.13.1
• Tokenizers 0.13.3