This model is a fine-tuned version of allenai/longformer-base-4096 on the stab-gurevych-essays dataset. It achieves the following results on the evaluation set:

Loss: 0.2532
Claim: {'precision': 0.6492678318375059, 'recall': 0.5891555936562366, 'f1-score': 0.6177528089887641, 'support': 4666.0}
Majorclaim: {'precision': 0.8648648648648649, 'recall': 0.8003280032800328, 'f1-score': 0.8313458262350937, 'support': 2439.0}
O: {'precision': 1.0, 'recall': 0.9992669218864544, 'f1-score': 0.9996333265430841, 'support': 12277.0}
Premise: {'precision': 0.8896274845333685, 'recall': 0.9276645391531123, 'f1-score': 0.9082479422140097, 'support': 14571.0}
Accuracy: 0.8979
Macro avg: {'precision': 0.8509400453089349, 'recall': 0.8291037644939591, 'f1-score': 0.839244975995238, 'support': 33953.0}
Weighted avg: {'precision': 0.8947265686653585, 'recall': 0.8978882572968515, 'f1-score': 0.8958462048390052, 'support': 33953.0}

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: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Claim	Majorclaim	O	Premise	Accuracy	Macro avg	Weighted avg
No log	1.0	41	0.3662	{'precision': 0.5429831006612784, 'recall': 0.15837976853836264, 'f1-score': 0.24522979923676788, 'support': 4666.0}	{'precision': 0.6182052106786748, 'recall': 0.7880278802788028, 'f1-score': 0.6928622927180966, 'support': 2439.0}	{'precision': 0.9983710702068741, 'recall': 0.9984523906491813, 'f1-score': 0.9984117287721441, 'support': 12277.0}	{'precision': 0.81162452775356, 'recall': 0.9583419120170201, 'f1-score': 0.8789023162134945, 'support': 14571.0}	0.8507	{'precision': 0.7427959773250968, 'recall': 0.7258004878708417, 'f1-score': 0.7038515342351258, 'support': 33953.0}	{'precision': 0.8283375336305401, 'recall': 0.8506759343798781, 'f1-score': 0.8216687720926173, 'support': 33953.0}
No log	2.0	82	0.2911	{'precision': 0.6262440103206782, 'recall': 0.3641234462066009, 'f1-score': 0.4604960021683155, 'support': 4666.0}	{'precision': 0.7612877895563408, 'recall': 0.7949979499794998, 'f1-score': 0.7777777777777778, 'support': 2439.0}	{'precision': 0.9998368545558365, 'recall': 0.9983709375254541, 'f1-score': 0.9991033583306163, 'support': 12277.0}	{'precision': 0.8484239990264086, 'recall': 0.9569006931576419, 'f1-score': 0.8994033220448315, 'support': 14571.0}	0.8788	{'precision': 0.808948163364816, 'recall': 0.7785982567172992, 'f1-score': 0.7841951150803853, 'support': 33953.0}	{'precision': 0.8663805444019677, 'recall': 0.878803051276765, 'f1-score': 0.8663997903530639, 'support': 33953.0}
No log	3.0	123	0.2556	{'precision': 0.6437802907915994, 'recall': 0.5124303471924561, 'f1-score': 0.5706443914081145, 'support': 4666.0}	{'precision': 0.8149312377210216, 'recall': 0.8503485034850349, 'f1-score': 0.8322632423756019, 'support': 2439.0}	{'precision': 1.0, 'recall': 0.9994298281339089, 'f1-score': 0.9997148327697886, 'support': 12277.0}	{'precision': 0.8788900414937759, 'recall': 0.9303410884633861, 'f1-score': 0.9038839806634439, 'support': 14571.0}	0.8921	{'precision': 0.8344003925015993, 'recall': 0.8231374418186965, 'f1-score': 0.8266266118042372, 'support': 33953.0}	{'precision': 0.8857774841763904, 'recall': 0.8921450240037699, 'f1-score': 0.8875948888942388, 'support': 33953.0}
No log	4.0	164	0.2579	{'precision': 0.5919405320813772, 'recall': 0.6485212173167595, 'f1-score': 0.6189404786254857, 'support': 4666.0}	{'precision': 0.8280930992241732, 'recall': 0.8314883148831488, 'f1-score': 0.8297872340425532, 'support': 2439.0}	{'precision': 1.0, 'recall': 0.9988596562678179, 'f1-score': 0.9994295028524858, 'support': 12277.0}	{'precision': 0.9086276452685965, 'recall': 0.88106512936655, 'f1-score': 0.8946341463414634, 'support': 14571.0}	0.8881	{'precision': 0.8321653191435366, 'recall': 0.839983579458569, 'f1-score': 0.835697840465497, 'support': 33953.0}	{'precision': 0.8923608226344708, 'recall': 0.8881394869378259, 'f1-score': 0.8899813710116259, 'support': 33953.0}
No log	5.0	205	0.2532	{'precision': 0.6492678318375059, 'recall': 0.5891555936562366, 'f1-score': 0.6177528089887641, 'support': 4666.0}	{'precision': 0.8648648648648649, 'recall': 0.8003280032800328, 'f1-score': 0.8313458262350937, 'support': 2439.0}	{'precision': 1.0, 'recall': 0.9992669218864544, 'f1-score': 0.9996333265430841, 'support': 12277.0}	{'precision': 0.8896274845333685, 'recall': 0.9276645391531123, 'f1-score': 0.9082479422140097, 'support': 14571.0}	0.8979	{'precision': 0.8509400453089349, 'recall': 0.8291037644939591, 'f1-score': 0.839244975995238, 'support': 33953.0}	{'precision': 0.8947265686653585, 'recall': 0.8978882572968515, 'f1-score': 0.8958462048390052, 'support': 33953.0}

Framework versions

Transformers 4.46.0
Pytorch 2.5.0+cu124
Datasets 3.0.2
Tokenizers 0.20.1

Theoreticallyhugo/longformer-sep_tok

longformer-sep_tok

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

Theoreticallyhugo/
longformer-sep_tok