README.md

metadata

library_name: transformers
license: mit
base_model: microsoft/git-large-r-coco
tags:
  - generated_from_trainer
datasets:
  - imagefolder
model-index:
  - name: git-large-r-coco-IDB_ADv1_COCO
    results: []

git-large-r-coco-IDB_ADv1_COCO

This model is a fine-tuned version of microsoft/git-large-r-coco on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.1340
• Meteor Score: {'meteor': 0.5103395242652348}

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: 128
• eval_batch_size: 128
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 1024
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• num_epochs: 200
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Meteor Score
4.8755	5.0	5	4.6208	{'meteor': 0.41234022080924504}
4.577	10.0	10	4.2632	{'meteor': 0.462553330633559}
4.228	15.0	15	3.9351	{'meteor': 0.4637035445872813}
3.8968	20.0	20	3.6117	{'meteor': 0.4716545164583618}
3.5731	25.0	25	3.2943	{'meteor': 0.4775515760416854}
3.2551	30.0	30	2.9844	{'meteor': 0.4827646049987953}
2.9433	35.0	35	2.6819	{'meteor': 0.4820540318646651}
2.6406	40.0	40	2.3893	{'meteor': 0.48387008867521647}
2.348	45.0	45	2.1093	{'meteor': 0.48688764217538394}
2.0685	50.0	50	1.8438	{'meteor': 0.48840003275775357}
1.8052	55.0	55	1.5954	{'meteor': 0.49229450416352066}
1.5592	60.0	60	1.3681	{'meteor': 0.49462336346473573}
1.3335	65.0	65	1.1642	{'meteor': 0.4943789886645904}
1.1308	70.0	70	0.9838	{'meteor': 0.4932081022324161}
0.9511	75.0	75	0.8281	{'meteor': 0.4949580448605414}
0.7953	80.0	80	0.6959	{'meteor': 0.4945236890902709}
0.6629	85.0	85	0.5849	{'meteor': 0.49613363555493917}
0.5518	90.0	90	0.4970	{'meteor': 0.49476521946537905}
0.4599	95.0	95	0.4203	{'meteor': 0.49694213467111825}
0.3856	100.0	100	0.3609	{'meteor': 0.5023234677593583}
0.3248	105.0	105	0.3137	{'meteor': 0.49291655975794224}
0.2756	110.0	110	0.2757	{'meteor': 0.49187607478517975}
0.236	115.0	115	0.2432	{'meteor': 0.4999076360911653}
0.2039	120.0	120	0.2196	{'meteor': 0.5054381333125716}
0.1782	125.0	125	0.2006	{'meteor': 0.4998272338605217}
0.1576	130.0	130	0.1864	{'meteor': 0.5095656338179543}
0.1411	135.0	135	0.1755	{'meteor': 0.5030929069103355}
0.1279	140.0	140	0.1653	{'meteor': 0.5097532440481348}
0.1173	145.0	145	0.1576	{'meteor': 0.5126165420799782}
0.1088	150.0	150	0.1516	{'meteor': 0.5168283983418568}
0.1023	155.0	155	0.1462	{'meteor': 0.5145210432669091}
0.097	160.0	160	0.1424	{'meteor': 0.5135483205500848}
0.0929	165.0	165	0.1399	{'meteor': 0.5099977164420265}
0.0899	170.0	170	0.1384	{'meteor': 0.5093303675700068}
0.0876	175.0	175	0.1369	{'meteor': 0.5097771482308939}
0.086	180.0	180	0.1357	{'meteor': 0.5080664663529372}
0.085	185.0	185	0.1347	{'meteor': 0.5101483486776783}
0.0843	190.0	190	0.1342	{'meteor': 0.5110798690668398}
0.0839	195.0	195	0.1340	{'meteor': 0.5102824562761434}
0.0838	200.0	200	0.1340	{'meteor': 0.5103395242652348}

Framework versions

• Transformers 4.46.1
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.20.2