metadata

license: apache-2.0
base_model: studio-ousia/luke-base
tags:
  - generated_from_trainer
metrics:
  - accuracy
  - precision
  - recall
  - f1
model-index:
  - name: luke-base-fp16-paper-setup
    results: []

luke-base-fp16-paper-setup

This model is a fine-tuned version of studio-ousia/luke-base on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.3800
Accuracy: 0.8426
Precision: 0.8465
Recall: 0.8370
F1: 0.8417

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: 1e-05
train_batch_size: 32
eval_batch_size: 32
seed: 42
optimizer: Adam with betas=(0.9,0.98) and epsilon=1e-06
lr_scheduler_type: linear
num_epochs: 3
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.5192	1.0	1074	0.4464	0.8063	0.8263	0.7756	0.8001
0.3708	2.0	2148	0.3782	0.8315	0.8365	0.8241	0.8303
0.3191	3.0	3222	0.3800	0.8426	0.8465	0.8370	0.8417

Framework versions

Transformers 4.36.2
Pytorch 2.1.0+cu121
Datasets 2.16.1
Tokenizers 0.15.0