IC_ver6e_coco_swin_gpt2_50Apc_1e

This model is a fine-tuned version of VK246/IC_ver6d_coco_swin_gpt2_50Bpc_1e on the coco dataset. It achieves the following results on the evaluation set:

• Loss: 0.7783
• Cider: 19.1116
• Rouge1: 42.2076
• Rouge2: 16.6791
• Rougel: 38.4352
• Rougelsum: 38.4324
• Bleu-1: 42.9768
• Bleu-2: 25.0535
• Bleu-3: 15.8932
• Bleu-4: 10.5581
• Gen Len: 11.2806

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	Cider	Rouge1	Rouge2	Rougel	Rougelsum	Bleu-1	Bleu-2	Bleu-3	Bleu-4	Gen Len
0.7299	0.17	500	0.8169	15.1223	40.4746	15.1013	36.817	36.8166	41.7335	23.5713	14.621	9.566	11.2806
0.7243	0.34	1000	0.8063	15.7288	41.2081	15.8926	37.4018	37.4016	42.2656	24.2595	15.2602	10.0788	11.2806
0.7396	0.51	1500	0.7999	15.5164	41.6231	16.1665	38.0103	38.0119	42.0958	24.3223	15.2851	10.0869	11.2806
0.7507	0.68	2000	0.7879	15.3421	41.9871	16.4909	38.2491	38.2515	42.6606	24.7464	15.6329	10.3731	11.2806
0.7712	0.85	2500	0.7820	11.751	41.9906	16.5153	38.2624	38.2634	42.8539	24.8663	15.7151	10.3989	11.2806

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.1+cu118
• Datasets 2.14.4
• Tokenizers 0.13.3