bert-italian-xxl-cased-ItaCoLA

This model is a fine-tuned version of dbmdz/bert-base-italian-xxl-cased on the ItaCoLA dataset. It achieves the following results on the evaluation set:

• Loss: 0.3049
• Accuracy: 0.8911

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: 32
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.4243	0.41	100	0.3805	0.8541
0.3806	0.82	200	0.3862	0.8573
0.3289	1.23	300	0.3537	0.8679
0.2967	1.64	400	0.3165	0.8827
0.2992	2.05	500	0.3382	0.8784
0.2273	2.46	600	0.3294	0.8816
0.2215	2.87	700	0.3049	0.8911
0.1718	3.28	800	0.3531	0.8911
0.1757	3.69	900	0.3903	0.8922
0.1698	4.1	1000	0.3871	0.8953
0.1307	4.51	1100	0.4255	0.8953
0.1426	4.92	1200	0.3729	0.8985
0.1136	5.33	1300	0.4939	0.8964
0.1163	5.74	1400	0.4004	0.8964
0.0936	6.15	1500	0.5116	0.8964
0.0973	6.56	1600	0.4808	0.8922
0.0899	6.97	1700	0.4813	0.8869
0.0687	7.38	1800	0.6046	0.8848
0.0709	7.79	1900	0.5940	0.8964
0.0694	8.2	2000	0.5791	0.8911
0.0732	8.61	2100	0.5577	0.8922
0.0714	9.02	2200	0.5249	0.8996
0.0531	9.43	2300	0.6098	0.8932
0.0713	9.84	2400	0.5610	0.8943

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.1+cu117
• Datasets 2.12.0
• Tokenizers 0.13.3