2026-08-27-crf-classweights-clean

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

• Loss: 8.6272
• Precision: 0.7821
• Recall: 0.8211
• F1: 0.8011
• Accuracy: 0.9722
• Academicdiscipline F1: 0.3810
• Ambiguouslydefinedconcept F1: 0.7356
• Discoursephenomenon F1: 0.6549
• Graphemicphenomenon F1: 0.0
• Languagerelatedterm F1: 0.8691
• Languageresourceinformation F1: 0.7868
• Lexicalphenomenon F1: 0.6759
• Morphologicalphenomenon F1: 0.8191
• Morphosyntacticphenomenon F1: 0.8479
• New Tag F1: 0.8207
• Otherlinguisticterm F1: 0.7439
• Phonologicalphenomenon F1: 0.8649
• Semanticphenomenon F1: 0.7048
• Syntacticphenomenon F1: 0.7965
• Topnode Dummy F1: 0.6852
• Unclassifiedlinguisticconcept F1: 0.8787

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: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 8
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy	Academicdiscipline F1	Ambiguouslydefinedconcept F1	Discoursephenomenon F1	Graphemicphenomenon F1	Languagerelatedterm F1	Languageresourceinformation F1	Lexicalphenomenon F1	Morphologicalphenomenon F1	Morphosyntacticphenomenon F1	New Tag F1	Otherlinguisticterm F1	Phonologicalphenomenon F1	Semanticphenomenon F1	Syntacticphenomenon F1	Topnode Dummy F1	Unclassifiedlinguisticconcept F1
315.3845	1.0	667	14.5431	0.5680	0.5182	0.5420	0.9473	0.0	0.0	0.0	0.0	0.7149	0.2805	0.0176	0.5985	0.6561	0.4946	0.4232	0.7031	0.0	0.5345	0.1853	0.0
60.1323	2.0	1334	8.8849	0.7186	0.7187	0.7187	0.9637	0.5	0.5479	0.5388	0.0	0.7927	0.7110	0.5450	0.7505	0.7886	0.7353	0.6519	0.8232	0.5540	0.6947	0.5443	0.7627
28.8042	3.0	2001	7.9844	0.7165	0.7966	0.7544	0.9663	0.5	0.6	0.5954	0.0	0.8207	0.7217	0.6040	0.7752	0.8127	0.7662	0.6958	0.8428	0.6908	0.7262	0.6268	0.85
19.7129	4.0	2668	7.6316	0.7587	0.7994	0.7785	0.9695	0.4211	0.7314	0.6038	0.0	0.8359	0.756	0.6786	0.8241	0.8171	0.7872	0.7268	0.8528	0.6925	0.7694	0.6604	0.8559
16.1557	5.0	3335	8.0953	0.7406	0.8337	0.7844	0.9691	0.3077	0.7442	0.6696	0.0	0.8418	0.7829	0.6262	0.7988	0.8349	0.8207	0.7372	0.8581	0.7079	0.7754	0.6591	0.8583
10.1011	6.0	4002	8.4333	0.7617	0.8333	0.7959	0.9707	0.3478	0.7619	0.7043	0.0	0.8461	0.7824	0.6582	0.7984	0.8348	0.8173	0.7417	0.8669	0.7135	0.7932	0.6897	0.875
7.3063	7.0	4669	8.6272	0.7821	0.8211	0.8011	0.9722	0.3810	0.7356	0.6549	0.0	0.8691	0.7868	0.6759	0.8191	0.8479	0.8207	0.7439	0.8649	0.7048	0.7965	0.6852	0.8787
6.1858	8.0	5336	9.1583	0.7745	0.8212	0.7972	0.9713	0.3636	0.7640	0.6667	0.0	0.8618	0.7841	0.6771	0.8175	0.8432	0.8139	0.7404	0.8646	0.7028	0.7952	0.6811	0.8729

Framework versions

• Transformers 4.57.6
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.22.2