litbank-coref-mem-large-triple

This model is a fine-tuned version of google/flan-t5-large on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0078

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: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 4
• 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: 7

Training results

Training Loss	Epoch	Step	Validation Loss
3.9003	0.2896	500	0.0298
0.0364	0.5792	1000	0.0206
0.0171	0.8688	1500	0.0175
0.0149	1.1581	2000	0.0154
0.0122	1.4477	2500	0.0142
0.0117	1.7373	3000	0.0128
0.0115	2.0266	3500	0.0118
0.0091	2.3162	4000	0.0114
0.0091	2.6058	4500	0.0108
0.0087	2.8955	5000	0.0101
0.0079	3.1848	5500	0.0101
0.0075	3.4744	6000	0.0097
0.0072	3.7640	6500	0.0094
0.0071	4.0533	7000	0.0092
0.0063	4.3429	7500	0.0089
0.0066	4.6325	8000	0.0085
0.0063	4.9221	8500	0.0084
0.006	5.2114	9000	0.0082
0.0056	5.5010	9500	0.0080
0.0056	5.7906	10000	0.0080
0.0053	6.0799	10500	0.0080
0.0054	6.3695	11000	0.0079
0.0052	6.6591	11500	0.0079
0.0051	6.9487	12000	0.0078

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu121
• Datasets 3.3.1
• Tokenizers 0.21.0