xlm-roberta-large-finetuned-augument-visquad2-27-3-2023-3

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: 75.3631
• Loss: 2.0450
• Exact: 38.9165
• F1: 56.3720
• Total: 3821
• Hasans Exact: 55.9744
• Hasans F1: 81.1148
• Hasans Total: 2653
• Noans Exact: 0.1712
• Noans F1: 0.1712
• Noans Total: 1168
• Best Exact: 59.7749
• Best Exact Thresh: 0.5183
• Best F1 Thresh: 0.8690

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

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
0.8597	1.0	4221	66.4890	1.2255	36.1947	54.1414	3821	52.1297	77.9775	2653	0.0	0.0	1168	52.9704	0.8158	0.9074
0.4623	2.0	8443	70.0050	1.1813	37.8173	55.5970	3821	54.4666	80.0740	2653	0.0	0.0	1168	55.1950	0.7529	0.8275
0.2999	3.0	12664	75.0810	1.2417	39.8587	56.3329	3821	57.3690	81.0961	2653	0.0856	0.0856	1168	60.4030	0.9294	0.9459
0.1915	4.0	16886	74.7037	1.6500	38.7333	56.2476	3821	55.7482	80.9733	2653	0.0856	0.0856	1168	58.6496	0.7690	0.9767
0.1185	5.0	21105	75.3631	2.0450	38.9165	56.3720	3821	55.9744	81.1148	2653	0.1712	0.1712	1168	59.7749	0.5183	0.8690

Framework versions

• Transformers 4.27.3
• Pytorch 1.13.1+cu117
• Datasets 2.10.1
• Tokenizers 0.13.2