bubai567/nifty_bert

This model is a fine-tuned version of nlpaueb/legal-bert-small-uncased on an unknown dataset. It achieves the following results on the evaluation set:

Train Loss: 0.0529
Train Accuracy: 0.9889
Validation Loss: 0.0324
Validation Accuracy: 0.9933
Epoch: 1

Model description

This model has been trained using 30 days of Nifty index data at 30-minute intervals. In this training dataset, signal values are represented as follows: 1 for peak signals, -1 for valley signals, and 0 for stay signals. The Smooth Z-Score method is employed to extract training samples. For example, a 30x10 input sample might look like this: 0 0 -1 -1 0 0 1 0 0 1. This model uses these samples to predict the direction of the Nifty index for the next 30 minutes.

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': None, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': False, 'is_legacy_optimizer': False, 'learning_rate': {'module': 'keras.optimizers.schedules', 'class_name': 'PolynomialDecay', 'config': {'initial_learning_rate': 2e-05, 'decay_steps': 608, 'end_learning_rate': 0.0, 'power': 1.0, 'cycle': False, 'name': None}, 'registered_name': None}, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False}
training_precision: float32

Training results

Train Loss	Train Accuracy	Validation Loss	Validation Accuracy	Epoch
0.1528	0.9651	0.0509	0.9914	0
0.0529	0.9889	0.0324	0.9933	1

Framework versions

Transformers 4.33.2
TensorFlow 2.13.0
Datasets 2.14.5
Tokenizers 0.13.3