food_vit_model

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the food101 dataset. It achieves the following results on the evaluation set:

• Loss: 0.5959
• Accuracy: 0.8749

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.4008	1.0	592	0.5106	0.8690
0.313	2.0	1184	0.5045	0.8718
0.3444	3.0	1776	0.5029	0.8711
0.3305	4.0	2368	0.5049	0.8706
0.2993	5.0	2960	0.5123	0.8702
0.3658	6.0	3552	0.5202	0.8662
0.314	7.0	4144	0.5344	0.8633
0.2973	8.0	4736	0.5558	0.8589
0.3171	9.0	5328	0.5806	0.8566
0.2841	10.0	5920	0.5932	0.856
0.4034	11.0	6512	0.5770	0.8554
0.3231	12.0	7104	0.5455	0.8607
0.3162	13.0	7696	0.5420	0.8634
0.3706	14.0	8288	0.5591	0.8590
0.2857	15.0	8880	0.5284	0.8653
0.2647	16.0	9472	0.5680	0.8567
0.2411	17.0	10064	0.5492	0.8648
0.2566	18.0	10656	0.5716	0.8581
0.2338	19.0	11248	0.5842	0.8573
0.2862	20.0	11840	0.5735	0.8592
0.2689	21.0	12432	0.5669	0.8604
0.1892	22.0	13024	0.5747	0.8602
0.1801	23.0	13616	0.5581	0.8627
0.2258	24.0	14208	0.5717	0.8614
0.2215	25.0	14800	0.6046	0.8562
0.1443	26.0	15392	0.5758	0.8642
0.2143	27.0	15984	0.5805	0.8626
0.1699	28.0	16576	0.5843	0.8616
0.1787	29.0	17168	0.5740	0.8657
0.1702	30.0	17760	0.5718	0.8653
0.1703	31.0	18352	0.5703	0.8646
0.1692	32.0	18944	0.5918	0.8627
0.1643	33.0	19536	0.6041	0.8608
0.214	34.0	20128	0.5950	0.8624
0.1996	35.0	20720	0.5861	0.8637
0.1618	36.0	21312	0.6032	0.8622
0.181	37.0	21904	0.5915	0.8646
0.1641	38.0	22496	0.5697	0.8663
0.1233	39.0	23088	0.5987	0.8617
0.1469	40.0	23680	0.5944	0.8635
0.1492	41.0	24272	0.5893	0.8651
0.1616	42.0	24864	0.5717	0.8667
0.1359	43.0	25456	0.5897	0.8655
0.1318	44.0	26048	0.5920	0.8684
0.102	45.0	26640	0.5908	0.8683
0.1416	46.0	27232	0.5977	0.8625
0.1393	47.0	27824	0.6069	0.8648
0.1003	48.0	28416	0.5849	0.8682
0.121	49.0	29008	0.5880	0.8661
0.128	50.0	29600	0.5800	0.8693
0.1409	51.0	30192	0.6004	0.8663
0.1783	52.0	30784	0.5847	0.8678
0.1177	53.0	31376	0.5984	0.8683
0.097	54.0	31968	0.5973	0.8669
0.137	55.0	32560	0.5983	0.8668
0.1227	56.0	33152	0.5913	0.8689
0.1259	57.0	33744	0.5949	0.868
0.0947	58.0	34336	0.6065	0.8664
0.1184	59.0	34928	0.6098	0.8667
0.0996	60.0	35520	0.5958	0.8700
0.0977	61.0	36112	0.6019	0.8694
0.1295	62.0	36704	0.6012	0.8698
0.0842	63.0	37296	0.5993	0.8688
0.0784	64.0	37888	0.6074	0.8689
0.1183	65.0	38480	0.5853	0.8713
0.1215	66.0	39072	0.5962	0.8709
0.1069	67.0	39664	0.5786	0.8728
0.101	68.0	40256	0.5938	0.8691
0.1004	69.0	40848	0.5985	0.8716
0.0958	70.0	41440	0.5961	0.8721
0.0914	71.0	42032	0.6053	0.8704
0.0915	72.0	42624	0.5937	0.8713
0.0964	73.0	43216	0.6001	0.8703
0.0558	74.0	43808	0.5993	0.8697
0.0977	75.0	44400	0.6025	0.8706
0.1096	76.0	44992	0.6018	0.8706
0.0883	77.0	45584	0.5973	0.8733
0.0811	78.0	46176	0.6023	0.8741
0.0912	79.0	46768	0.6004	0.8733
0.0981	80.0	47360	0.5851	0.8730
0.0892	81.0	47952	0.5782	0.8754
0.1119	82.0	48544	0.5893	0.8727
0.1016	83.0	49136	0.5911	0.8722
0.0801	84.0	49728	0.5880	0.8755
0.107	85.0	50320	0.6088	0.8710
0.0763	86.0	50912	0.5912	0.8760
0.0667	87.0	51504	0.5974	0.8752
0.0485	88.0	52096	0.5903	0.8763
0.1002	89.0	52688	0.6097	0.8744
0.0786	90.0	53280	0.5853	0.8762
0.1067	91.0	53872	0.5874	0.8772
0.0618	92.0	54464	0.5847	0.8762
0.0667	93.0	55056	0.5803	0.8774
0.0702	94.0	55648	0.5812	0.8781
0.055	95.0	56240	0.5918	0.8761
0.0941	96.0	56832	0.5904	0.8766
0.0821	97.0	57424	0.5849	0.8762
0.0998	98.0	58016	0.5891	0.8757
0.0594	99.0	58608	0.5845	0.8778
0.0727	100.0	59200	0.5959	0.8749

Framework versions

• Transformers 4.30.2
• Pytorch 2.0.0+cu118
• Datasets 2.13.1
• Tokenizers 0.13.3