Goader_liberta-large-deprel

This model is a fine-tuned version of Goader/liberta-large on the universal_dependencies dataset. It achieves the following results on the evaluation set:

• Loss: 0.5901
• : {'precision': 0.42857142857142855, 'recall': 0.21428571428571427, 'f1': 0.2857142857142857, 'number': 14}
• Arataxis: {'precision': 0.5066666666666667, 'recall': 0.3486238532110092, 'f1': 0.41304347826086957, 'number': 109}
• Arataxis:discourse: {'precision': 0.4117647058823529, 'recall': 0.3684210526315789, 'f1': 0.3888888888888889, 'number': 19}
• Arataxis:rel: {'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 7}
• Ark: {'precision': 0.8214285714285714, 'recall': 0.7777777777777778, 'f1': 0.7990074441687344, 'number': 207}
• Ase: {'precision': 0.8921023359288098, 'recall': 0.8044132397191575, 'f1': 0.8459915611814346, 'number': 997}
• Bj: {'precision': 0.8446389496717724, 'recall': 0.7524366471734892, 'f1': 0.7958762886597938, 'number': 513}
• Bl: {'precision': 0.8267090620031796, 'recall': 0.7084468664850136, 'f1': 0.7630227439471754, 'number': 734}
• C: {'precision': 0.8328690807799443, 'recall': 0.7310513447432763, 'f1': 0.7786458333333331, 'number': 409}
• Cl: {'precision': 0.7894736842105263, 'recall': 0.3409090909090909, 'f1': 0.4761904761904762, 'number': 44}
• Cl:adv: {'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 4}
• Cl:relcl: {'precision': 0.7611940298507462, 'recall': 0.7083333333333334, 'f1': 0.7338129496402879, 'number': 144}
• Comp: {'precision': 0.7708333333333334, 'recall': 0.7602739726027398, 'f1': 0.7655172413793104, 'number': 146}
• Comp:sp: {'precision': 0.7931034482758621, 'recall': 0.5897435897435898, 'f1': 0.676470588235294, 'number': 39}
• Dvcl: {'precision': 0.8210526315789474, 'recall': 0.7027027027027027, 'f1': 0.7572815533980582, 'number': 111}
• Dvcl:sp: {'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 3}
• Dvcl:svc: {'precision': 1.0, 'recall': 0.2, 'f1': 0.33333333333333337, 'number': 5}
• Dvmod: {'precision': 0.8085585585585585, 'recall': 0.7638297872340426, 'f1': 0.7855579868708972, 'number': 470}
• Dvmod:det: {'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 2}
• Et: {'precision': 0.8657407407407407, 'recall': 0.7824267782426778, 'f1': 0.8219780219780219, 'number': 239}
• Et:numgov: {'precision': 0.75, 'recall': 0.75, 'f1': 0.75, 'number': 12}
• Et:nummod: {'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 3}
• Iscourse: {'precision': 0.753731343283582, 'recall': 0.6352201257861635, 'f1': 0.6894197952218429, 'number': 159}
• Islocated: {'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 5}
• Ixed: {'precision': 0.75, 'recall': 0.2608695652173913, 'f1': 0.3870967741935483, 'number': 23}
• Lat:abs: {'precision': 1.0, 'recall': 0.5, 'f1': 0.6666666666666666, 'number': 2}
• Lat:foreign: {'precision': 0.625, 'recall': 0.25, 'f1': 0.35714285714285715, 'number': 20}
• Lat:name: {'precision': 0.631578947368421, 'recall': 0.43636363636363634, 'f1': 0.5161290322580645, 'number': 55}
• Lat:range: {'precision': 0.75, 'recall': 0.5, 'f1': 0.6, 'number': 12}
• Lat:repeat: {'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 3}
• Lat:title: {'precision': 0.616, 'recall': 0.47530864197530864, 'f1': 0.5365853658536586, 'number': 162}
• Mod: {'precision': 0.7972646822204345, 'recall': 0.7155234657039711, 'f1': 0.7541856925418569, 'number': 1385}
• Obj: {'precision': 0.4090909090909091, 'recall': 0.6, 'f1': 0.4864864864864865, 'number': 15}
• Ocative: {'precision': 0.25, 'recall': 1.0, 'f1': 0.4, 'number': 1}
• Oeswith: {'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 1}
• Ompound: {'precision': 0.6764705882352942, 'recall': 0.3898305084745763, 'f1': 0.49462365591397844, 'number': 59}
• Onj: {'precision': 0.7439024390243902, 'recall': 0.5831739961759083, 'f1': 0.6538049303322616, 'number': 523}
• Oot: {'precision': 0.9379310344827586, 'recall': 0.9066666666666666, 'f1': 0.9220338983050848, 'number': 600}
• Op: {'precision': 0.7592592592592593, 'recall': 0.7454545454545455, 'f1': 0.7522935779816514, 'number': 55}
• Ppos: {'precision': 0.4262295081967213, 'recall': 0.30952380952380953, 'f1': 0.3586206896551724, 'number': 84}
• Rphan: {'precision': 0.6, 'recall': 0.23076923076923078, 'f1': 0.33333333333333337, 'number': 13}
• Subj: {'precision': 0.8660084626234132, 'recall': 0.8143236074270557, 'f1': 0.8393711551606288, 'number': 754}
• Ummod: {'precision': 0.6153846153846154, 'recall': 0.6, 'f1': 0.6075949367088608, 'number': 40}
• Ummod:gov: {'precision': 0.7352941176470589, 'recall': 0.625, 'f1': 0.6756756756756757, 'number': 40}
• Unct: {'precision': 0.8604790419161676, 'recall': 0.7418688693856479, 'f1': 0.7967840310507347, 'number': 1937}
• Ux: {'precision': 0.6875, 'recall': 0.6111111111111112, 'f1': 0.6470588235294118, 'number': 18}
• Xpl: {'precision': 0.8333333333333334, 'recall': 0.7142857142857143, 'f1': 0.7692307692307692, 'number': 7}
• Overall Precision: 0.8253
• Overall Recall: 0.7232
• Overall F1: 0.7709
• Overall Accuracy: 0.8090

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: 5e-05
• train_batch_size: 32
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Framework versions

• Transformers 4.39.3
• Pytorch 1.11.0a0+17540c5
• Datasets 2.21.0
• Tokenizers 0.15.2