detr-resnet-101_sgd_finetuned_food-roboflow

This model is a fine-tuned version of facebook/detr-resnet-101 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 2.9270

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: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
6.7367	0.77	50	6.2388
5.9103	1.54	100	5.4768
4.8918	2.31	150	4.3742
3.827	3.08	200	3.8198
3.4129	3.85	250	3.3190
3.0329	4.62	300	3.1967
2.8295	5.38	350	3.1841
2.8599	6.15	400	3.1155
2.7862	6.92	450	3.1008
2.7885	7.69	500	3.0490
2.6737	8.46	550	3.0872
2.7679	9.23	600	3.0429
2.6093	10.0	650	2.9775
2.6316	10.77	700	3.0016
2.5801	11.54	750	2.9701
2.6009	12.31	800	2.8919
2.5841	13.08	850	2.9398
2.5394	13.85	900	2.9266
2.5189	14.62	950	2.9270

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1