DinoAmoros-large-2025_05_06_37720-prova_bs16_freeze_monolabel

This model is a fine-tuned version of facebook/dinov2-large on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.9011
• F1 Micro: 0.4
• F1 Macro: 0.2667
• Accuracy: 0.4
• Learning Rate: 1e-05

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: 0.001
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 150
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Micro	F1 Macro	Accuracy	Rate
No log	1.0	1	3.1371	0.1	0.0667	0.1	0.001
No log	2.0	2	3.1559	0.1	0.05	0.1	0.001
No log	3.0	3	3.0740	0.2	0.1524	0.2	0.001
No log	4.0	4	3.0686	0.4	0.2333	0.4	0.001
No log	5.0	5	2.9377	0.5	0.2714	0.5	0.001
No log	6.0	6	2.8814	0.4	0.2286	0.4	0.001
No log	7.0	7	2.8164	0.3	0.1587	0.3	0.001
No log	8.0	8	2.7270	0.2	0.1333	0.2	0.001
No log	9.0	9	2.6143	0.4	0.2667	0.4	0.001
No log	10.0	10	2.4839	0.4	0.2667	0.4	0.001
No log	11.0	11	2.3587	0.4	0.2667	0.4	0.001
No log	12.0	12	2.2280	0.5	0.3048	0.5	0.001
No log	13.0	13	2.0928	0.5	0.3048	0.5	0.001
No log	14.0	14	2.0064	0.5	0.3048	0.5	0.001
No log	15.0	15	1.8491	0.5	0.3048	0.5	0.001
No log	16.0	16	1.7906	0.6	0.4048	0.6	0.001
No log	17.0	17	1.7055	0.6	0.4048	0.6	0.001
No log	18.0	18	1.6328	0.6	0.4048	0.6	0.001
No log	19.0	19	1.5969	0.6	0.4048	0.6	0.001
No log	20.0	20	1.5740	0.6	0.4048	0.6	0.001
No log	21.0	21	1.5900	0.6	0.4048	0.6	0.001
No log	22.0	22	1.6051	0.6	0.4048	0.6	0.001
No log	23.0	23	1.6009	0.6	0.4048	0.6	0.001
No log	24.0	24	1.5874	0.6	0.4048	0.6	0.001
No log	25.0	25	1.5908	0.6	0.4048	0.6	0.001
No log	26.0	26	1.5866	0.6	0.4048	0.6	0.001
No log	27.0	27	1.5541	0.6	0.4048	0.6	0.0001
No log	28.0	28	1.5380	0.6	0.4048	0.6	0.0001
No log	29.0	29	1.5012	0.6	0.4048	0.6	0.0001
No log	30.0	30	1.4620	0.6	0.4048	0.6	0.0001
No log	31.0	31	1.4533	0.6	0.4048	0.6	0.0001
No log	32.0	32	1.4629	0.6	0.4048	0.6	0.0001
No log	33.0	33	1.4447	0.6	0.4048	0.6	0.0001
No log	34.0	34	1.4385	0.6	0.4048	0.6	0.0001
No log	35.0	35	1.4305	0.6	0.4048	0.6	0.0001
No log	36.0	36	1.4212	0.6	0.4048	0.6	0.0001
No log	37.0	37	1.4205	0.6	0.4048	0.6	0.0001
No log	38.0	38	1.4142	0.6	0.4048	0.6	0.0001
No log	39.0	39	1.3952	0.6	0.4048	0.6	0.0001
No log	40.0	40	1.3966	0.6	0.4048	0.6	0.0001
No log	41.0	41	1.4088	0.6	0.4048	0.6	0.0001
No log	42.0	42	1.4072	0.6	0.4048	0.6	0.0001
No log	43.0	43	1.4033	0.6	0.4048	0.6	0.0001
No log	44.0	44	1.4202	0.6	0.4048	0.6	0.0001
No log	45.0	45	1.4144	0.5	0.3048	0.5	0.0001
No log	46.0	46	1.4181	0.5	0.3048	0.5	1e-05
No log	47.0	47	1.4177	0.5	0.3048	0.5	1e-05
No log	48.0	48	1.4321	0.5	0.3048	0.5	1e-05
No log	49.0	49	1.4261	0.5	0.3048	0.5	1e-05

Framework versions

• Transformers 4.48.0
• Pytorch 2.6.0+cu118
• Datasets 3.0.2
• Tokenizers 0.21.1