dino-base-2023_11_24-unfreeze

This model is a fine-tuned version of facebook/dinov2-base on the multilabel_complete_dataset dataset. It achieves the following results on the evaluation set:

• Loss: 0.2045
• F1 Micro: 0.6595
• F1 Macro: 0.5161
• Roc Auc: 0.7681
• Accuracy: 0.2735
• Learning Rate: 0.001

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: 0.01
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Micro	F1 Macro	Roc Auc	Accuracy	Rate
0.3961	1.0	536	0.3293	0.3296	0.0966	0.6005	0.0736	0.01
0.3418	2.0	1072	0.3382	0.3379	0.1283	0.6054	0.0439	0.01
0.3334	3.0	1608	0.3551	0.3905	0.2420	0.6319	0.0786	0.01
0.3323	4.0	2144	0.3213	0.2555	0.1099	0.5720	0.0625	0.01
0.3248	5.0	2680	0.3164	0.3355	0.1298	0.6024	0.0450	0.01
0.3235	6.0	3216	0.3346	0.2864	0.0806	0.5834	0.0239	0.01
0.32	7.0	3752	0.3029	0.4594	0.1968	0.6663	0.0682	0.01
0.3138	8.0	4288	0.2866	0.5468	0.2940	0.7240	0.0579	0.01
0.3052	9.0	4824	0.2807	0.4767	0.2993	0.6672	0.1225	0.01
0.3157	10.0	5360	0.2955	0.4752	0.2091	0.6733	0.0707	0.01
0.3119	11.0	5896	0.3405	0.4028	0.2160	0.6336	0.1361	0.01
0.3162	12.0	6432	0.4163	0.4899	0.2965	0.6863	0.0532	0.01
0.3184	13.0	6968	0.2964	0.5429	0.3299	0.7170	0.1047	0.01
0.3142	14.0	7504	0.3005	0.5253	0.3154	0.7072	0.0832	0.01
0.3104	15.0	8040	3.1991	0.1673	0.0674	0.4879	0.0	0.01
0.3042	16.0	8576	0.2820	0.4544	0.2746	0.6519	0.1583	0.001
0.2788	17.0	9112	0.2741	0.5744	0.3842	0.7205	0.1640	0.001
0.2724	18.0	9648	0.2424	0.5903	0.3936	0.7256	0.2072	0.001
0.2642	19.0	10184	0.2414	0.6021	0.4095	0.7347	0.2186	0.001
0.2597	20.0	10720	0.2269	0.6079	0.4156	0.7347	0.2251	0.001
0.2575	21.0	11256	0.2249	0.6231	0.4253	0.7463	0.2340	0.001
0.253	22.0	11792	0.2261	0.6291	0.4639	0.7521	0.2429	0.001
0.2491	23.0	12328	0.2163	0.6454	0.4856	0.7627	0.2537	0.001
0.2484	24.0	12864	0.2212	0.6262	0.4635	0.7460	0.2569	0.001
0.2465	25.0	13400	0.2118	0.6486	0.4780	0.7622	0.2772	0.001
0.241	26.0	13936	0.2106	0.6602	0.5159	0.7727	0.2558	0.001
0.2413	27.0	14472	0.2135	0.6390	0.4979	0.7536	0.2722	0.001
0.2385	28.0	15008	0.2182	0.6103	0.4596	0.7319	0.2772	0.001
0.2366	29.0	15544	0.2132	0.6615	0.5354	0.7758	0.2708	0.001
0.2345	30.0	16080	0.2069	0.6566	0.5122	0.7658	0.2747	0.001

Framework versions

• Transformers 4.34.1
• Pytorch 2.1.0+cu118
• Datasets 2.14.5
• Tokenizers 0.14.1