20240812

This model is a fine-tuned version of facebook/detr-resnet-50 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 1.8711

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: 20
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.4317	14.29	400	2.2595
1.922	28.57	800	1.8296
1.6594	42.86	1200	1.7387
1.423	57.14	1600	1.5079
1.285	71.43	2000	1.4590
1.1252	85.71	2400	1.4123
1.0256	100.0	2800	1.3712
0.8894	114.29	3200	1.3256
0.8398	128.57	3600	1.3442
0.7759	142.86	4000	1.3562
0.7035	157.14	4400	1.4312
0.6855	171.43	4800	1.4156
0.6538	185.71	5200	1.4837
0.6318	200.0	5600	1.4802
0.5742	214.29	6000	1.4791
0.5484	228.57	6400	1.5387
0.5344	242.86	6800	1.5534
0.512	257.14	7200	1.5829
0.5145	271.43	7600	1.5305
0.4935	285.71	8000	1.5981
0.5081	300.0	8400	1.6221
0.4352	314.29	8800	1.6413
0.4366	328.57	9200	1.5708
0.4174	342.86	9600	1.6348
0.4228	357.14	10000	1.5518
0.4032	371.43	10400	1.6675
0.397	385.71	10800	1.6444
0.3873	400.0	11200	1.6559
0.3801	414.29	11600	1.6167
0.3663	428.57	12000	1.6626
0.3493	442.86	12400	1.6778
0.3791	457.14	12800	1.6364
0.3463	471.43	13200	1.7587
0.341	485.71	13600	1.6680
0.3099	500.0	14000	1.7935
0.3393	514.29	14400	1.7037
0.3433	528.57	14800	1.7138
0.3111	542.86	15200	1.7046
0.3166	557.14	15600	1.6987
0.2909	571.43	16000	1.7040
0.3138	585.71	16400	1.7465
0.297	600.0	16800	1.7605
0.2899	614.29	17200	1.7669
0.2893	628.57	17600	1.7275
0.2906	642.86	18000	1.7863
0.2827	657.14	18400	1.7285
0.2927	671.43	18800	1.7736
0.2737	685.71	19200	1.8137
0.2724	700.0	19600	1.7758
0.265	714.29	20000	1.7858
0.2554	728.57	20400	1.8155
0.2589	742.86	20800	1.7439
0.2499	757.14	21200	1.8261
0.2631	771.43	21600	1.8170
0.254	785.71	22000	1.7619
0.2526	800.0	22400	1.7755
0.2462	814.29	22800	1.7828
0.2561	828.57	23200	1.8284
0.232	842.86	23600	1.8761
0.2437	857.14	24000	1.8398
0.2578	871.43	24400	1.8664
0.2446	885.71	24800	1.8846
0.241	900.0	25200	1.8541
0.2291	914.29	25600	1.8632
0.2474	928.57	26000	1.8613
0.2386	942.86	26400	1.8383
0.2378	957.14	26800	1.8783
0.2276	971.43	27200	1.8564
0.2296	985.71	27600	1.8669
0.2558	1000.0	28000	1.8711

Framework versions

• Transformers 4.37.2
• Pytorch 2.3.0
• Datasets 2.12.0
• Tokenizers 0.15.1