Roberta-CLS-URL

This model is a fine-tuned version of papluca/xlm-roberta-base-language-detection on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1817
• Accuracy: 0.9571
• F1: 0.9572

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: 2e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
No log	0.14	50	0.2787	0.8943	0.8942
No log	0.28	100	0.2332	0.9179	0.9180
No log	0.42	150	0.2369	0.9268	0.9269
No log	0.56	200	0.2071	0.9313	0.9314
No log	0.69	250	0.2017	0.9344	0.9343
No log	0.83	300	0.1953	0.9414	0.9415
No log	0.97	350	0.2031	0.9394	0.9394
0.2675	1.11	400	0.1915	0.9439	0.9439
0.2675	1.25	450	0.1696	0.9439	0.9440
0.2675	1.39	500	0.1747	0.9487	0.9488
0.2675	1.53	550	0.1958	0.9425	0.9427
0.2675	1.67	600	0.1608	0.9462	0.9461
0.2675	1.81	650	0.1547	0.9523	0.9524
0.2675	1.94	700	0.1668	0.9557	0.9557
0.1686	2.08	750	0.1709	0.9498	0.9499
0.1686	2.22	800	0.1605	0.9554	0.9555
0.1686	2.36	850	0.1703	0.9501	0.9501
0.1686	2.5	900	0.1603	0.9465	0.9466
0.1686	2.64	950	0.1742	0.9451	0.9451
0.1686	2.78	1000	0.1507	0.9546	0.9546
0.1686	2.92	1050	0.1423	0.9557	0.9557
0.1385	3.06	1100	0.1496	0.9574	0.9575
0.1385	3.19	1150	0.1590	0.9549	0.9549
0.1385	3.33	1200	0.1492	0.9523	0.9524
0.1385	3.47	1250	0.1390	0.9565	0.9566
0.1385	3.61	1300	0.1496	0.9529	0.9530
0.1385	3.75	1350	0.1425	0.9551	0.9552
0.1385	3.89	1400	0.1494	0.9521	0.9522
0.1221	4.03	1450	0.1541	0.9557	0.9557
0.1221	4.17	1500	0.1897	0.9532	0.9532
0.1221	4.31	1550	0.1595	0.9518	0.9519
0.1221	4.44	1600	0.1514	0.9554	0.9555
0.1221	4.58	1650	0.1553	0.9554	0.9555
0.1221	4.72	1700	0.1626	0.9543	0.9543
0.1221	4.86	1750	0.1509	0.9523	0.9523
0.1034	5.0	1800	0.1448	0.9554	0.9555
0.1034	5.14	1850	0.1685	0.9574	0.9574
0.1034	5.28	1900	0.1555	0.9551	0.9552
0.1034	5.42	1950	0.1595	0.9557	0.9557
0.1034	5.56	2000	0.1660	0.9565	0.9566
0.1034	5.69	2050	0.1511	0.9554	0.9555
0.1034	5.83	2100	0.1443	0.9565	0.9566
0.1034	5.97	2150	0.1526	0.9554	0.9554
0.0925	6.11	2200	0.1753	0.9540	0.9541
0.0925	6.25	2250	0.1503	0.9557	0.9557
0.0925	6.39	2300	0.1827	0.9518	0.9518
0.0925	6.53	2350	0.1486	0.9568	0.9568
0.0925	6.67	2400	0.1652	0.9568	0.9569
0.0925	6.81	2450	0.1544	0.9537	0.9538
0.0925	6.94	2500	0.1599	0.9551	0.9552
0.082	7.08	2550	0.1748	0.9568	0.9569
0.082	7.22	2600	0.1765	0.9582	0.9583
0.082	7.36	2650	0.1699	0.9568	0.9569
0.082	7.5	2700	0.1631	0.9563	0.9563
0.082	7.64	2750	0.1759	0.9602	0.9602
0.082	7.78	2800	0.1746	0.9565	0.9566
0.082	7.92	2850	0.1561	0.9568	0.9569
0.0742	8.06	2900	0.1721	0.9577	0.9577
0.0742	8.19	2950	0.1877	0.9563	0.9563
0.0742	8.33	3000	0.1896	0.9549	0.9549
0.0742	8.47	3050	0.1751	0.9577	0.9577
0.0742	8.61	3100	0.1812	0.9577	0.9577
0.0742	8.75	3150	0.1845	0.9577	0.9577
0.0742	8.89	3200	0.1844	0.9579	0.9580
0.0659	9.03	3250	0.1963	0.9571	0.9571
0.0659	9.17	3300	0.1861	0.9577	0.9577
0.0659	9.31	3350	0.1941	0.9585	0.9586
0.0659	9.44	3400	0.1900	0.9565	0.9566
0.0659	9.58	3450	0.1903	0.9565	0.9566
0.0659	9.72	3500	0.1836	0.9579	0.9580
0.0659	9.86	3550	0.1818	0.9565	0.9566
0.0631	10.0	3600	0.1817	0.9571	0.9572

Framework versions

• Transformers 4.39.3
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2