CNEC2_0_xlm-roberta-large

This model is a fine-tuned version of FacebookAI/xlm-roberta-large on the cnec dataset. It achieves the following results on the evaluation set:

• Loss: 0.2807
• Precision: 0.8544
• Recall: 0.8813
• F1: 0.8676
• Accuracy: 0.9631

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
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.7031	0.56	500	0.3041	0.6755	0.6917	0.6835	0.9328
0.2981	1.11	1000	0.2336	0.7821	0.8011	0.7915	0.9489
0.2327	1.67	1500	0.1969	0.8030	0.7947	0.7988	0.9518
0.1962	2.22	2000	0.1898	0.8152	0.8501	0.8323	0.9583
0.1683	2.78	2500	0.1690	0.8053	0.8401	0.8223	0.9585
0.1499	3.33	3000	0.1810	0.8319	0.8444	0.8381	0.9602
0.1376	3.89	3500	0.1888	0.8340	0.8591	0.8464	0.9599
0.1198	4.44	4000	0.2022	0.8089	0.8494	0.8287	0.9570
0.1089	5.0	4500	0.1930	0.8320	0.8448	0.8383	0.9578
0.0911	5.56	5000	0.1945	0.8412	0.8544	0.8478	0.9627
0.0945	6.11	5500	0.1961	0.8424	0.8430	0.8427	0.9606
0.0695	6.67	6000	0.2186	0.8289	0.8559	0.8422	0.9588
0.0628	7.22	6500	0.2016	0.8567	0.8723	0.8644	0.9629
0.0563	7.78	7000	0.2195	0.8528	0.8727	0.8626	0.9617
0.0504	8.33	7500	0.2301	0.8508	0.8730	0.8618	0.9609
0.0444	8.89	8000	0.2135	0.8486	0.8780	0.8631	0.9629
0.0386	9.44	8500	0.2347	0.8451	0.8838	0.8640	0.9625
0.0355	10.0	9000	0.2314	0.8499	0.8670	0.8584	0.9620
0.0305	10.56	9500	0.2467	0.8532	0.8709	0.8619	0.9627
0.0283	11.11	10000	0.2602	0.8440	0.8687	0.8562	0.9615
0.0217	11.67	10500	0.2639	0.8548	0.8777	0.8661	0.9632
0.0224	12.22	11000	0.2688	0.8504	0.8780	0.8640	0.9623
0.0194	12.78	11500	0.2661	0.8545	0.8798	0.8670	0.9629
0.0224	13.33	12000	0.2731	0.8512	0.8798	0.8653	0.9623
0.014	13.89	12500	0.2778	0.8537	0.8766	0.8650	0.9629
0.0146	14.44	13000	0.2783	0.8551	0.8798	0.8673	0.9629
0.0142	15.0	13500	0.2807	0.8544	0.8813	0.8676	0.9631

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0