fine_tuned_xlm-roberta-large_2April

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

• Best F1: 75.9147
• Loss: 1.6634
• Exact: 39.3876
• F1: 56.7272
• Total: 3821
• Hasans Exact: 56.6152
• Hasans F1: 81.5886
• Hasans Total: 2653
• Noans Exact: 0.2568
• Noans F1: 0.2568
• Noans Total: 1168
• Best Exact: 60.4292
• Best Exact Thresh: 0.6508
• Best F1 Thresh: 0.9299

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: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Best F1	Validation Loss	Exact	F1	Total	Hasans Exact	Hasans F1	Hasans Total	Noans Exact	Noans F1	Noans Total	Best Exact	Best Exact Thresh	Best F1 Thresh
2.0619	0.26	500	54.8058	1.4812	32.2952	51.4988	3821	46.5134	74.1714	2653	0.0	0.0	1168	41.4551	0.7820	0.8887
1.3598	0.53	1000	65.8410	1.2289	35.9068	54.1665	3821	51.7150	78.0136	2653	0.0	0.0	1168	51.7666	0.8149	0.8993
1.2347	0.79	1500	67.6083	1.1868	37.2416	55.0317	3821	53.6374	79.2597	2653	0.0	0.0	1168	53.7032	0.7981	0.8571
1.0961	1.05	2000	71.4132	1.1491	38.9165	55.8262	3821	56.0498	80.4041	2653	0.0	0.0	1168	57.5504	0.7733	0.8679
0.9003	1.32	2500	72.3291	1.2053	38.4193	56.0447	3821	55.3336	80.7188	2653	0.0	0.0	1168	57.5766	0.8557	0.9662
0.8705	1.58	3000	71.7222	1.1239	38.5239	56.1153	3821	55.4844	80.8204	2653	0.0	0.0	1168	56.6344	0.7408	0.8452
0.8655	1.84	3500	73.6273	1.0855	39.0212	56.4204	3821	56.2005	81.2599	2653	0.0	0.0	1168	58.7543	0.7574	0.8970
0.7431	2.11	4000	74.8323	1.1817	39.6231	56.5930	3821	56.9544	81.3954	2653	0.2568	0.2568	1168	59.7226	0.8032	0.9219
0.5738	2.37	4500	74.4675	1.2047	38.8642	56.6456	3821	55.8236	81.4334	2653	0.3425	0.3425	1168	58.8851	0.7003	0.8792
0.5904	2.63	5000	74.7345	1.1571	38.5763	56.4366	3821	55.5221	81.2455	2653	0.0856	0.0856	1168	59.1206	0.7922	0.8458
0.5831	2.89	5500	74.9378	1.1537	39.7278	56.6778	3821	57.2182	81.6306	2653	0.0	0.0	1168	59.9581	0.7947	0.8767
0.4785	3.16	6000	75.0234	1.3432	39.3353	56.8437	3821	56.1628	81.3795	2653	1.1130	1.1130	1168	59.2515	0.7999	0.8258
0.3774	3.42	6500	74.8641	1.4903	39.3876	56.6856	3821	56.2759	81.1894	2653	1.0274	1.0274	1168	59.6964	0.7004	0.9268
0.3882	3.68	7000	75.0504	1.3418	38.7857	56.4315	3821	55.8613	81.2759	2653	0.0	0.0	1168	59.4609	0.6782	0.9456
0.3747	3.95	7500	75.3181	1.3673	39.4399	56.6455	3821	56.8036	81.5840	2653	0.0	0.0	1168	59.6441	0.7541	0.9554
0.2748	4.21	8000	75.5835	1.5103	39.2829	56.6922	3821	56.5398	81.6136	2653	0.0856	0.0856	1168	60.1152	0.7237	0.9792
0.2346	4.47	8500	75.8283	1.6566	40.0157	57.2731	3821	57.0675	81.9225	2653	1.2842	1.2842	1168	60.5339	0.7154	0.9550
0.2339	4.74	9000	75.7051	1.6699	39.0474	56.5165	3821	56.0874	81.2475	2653	0.3425	0.3425	1168	60.3245	0.8887	0.9799
0.23	5.0	9500	75.9147	1.6634	39.3876	56.7272	3821	56.6152	81.5886	2653	0.2568	0.2568	1168	60.4292	0.6508	0.9299

Framework versions

• Transformers 4.37.2
• Pytorch 1.13.1+cu117
• Datasets 2.16.1
• Tokenizers 0.15.2