IC_ver6N_coco_swin_gpt2_50B_1e

This model is a fine-tuned version of VK246/IC_ver6M_coco_swin_gpt2_50A_1e on the coco dataset. It achieves the following results on the evaluation set:

• Loss: 0.8687
• Cider: 72.5295
• Rouge1: 41.1356
• Rouge2: 15.6403
• Rougel: 37.2276
• Rougelsum: 37.2406
• Bleu-1: 42.1796
• Bleu-2: 24.0186
• Bleu-3: 14.9974
• Bleu-4: 9.8206
• 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	Cider	Rouge1	Rouge2	Rougel	Rougelsum	Bleu-1	Bleu-2	Bleu-3	Bleu-4	Gen Len
0.3812	0.34	1000	1.0009	68.2568	39.9649	14.5975	36.1435	36.1441	41.3234	23.0865	14.2103	9.2075	11.3063
0.5147	0.68	2000	0.8687	72.5295	41.1356	15.6403	37.2276	37.2406	42.1796	24.0186	14.9974	9.8206	11.3063

Framework versions

• Transformers 4.32.0
• Pytorch 2.0.1+cu118
• Datasets 2.14.4
• Tokenizers 0.13.3