Project is the transfer learning of VGG-19 on 525 Bird Species.
Table of Contents
Fine-tuning DistilBERT, BERT, and Large BERT cased models for sentiment analysis of financial data and creating a user interface that allows a user to select one of the models and enter a sentence for analysis.
Required Libraries/Frameworks:
Tensorflow (configured for CUDA), Numpy, and Pandas
Required Hardware:
NVidia GPU
- Clone the Repo
git clone https://github.com/zjshermanburke/TransferLearning_VGG19- Download the dataset from Kaggle Birds 525 Species
- Extract dataset in Data Folder so that the structure follows "Data/train", "Data/valid", "Data/test"
Adjust the options in cell below "Establishing Hyperparameters" and modify output layers as desired in cell below "Creating Our Output Layers"
Hyperparameters
BATCH_SIZE = 64
EPOCHS =50
LEARNING_RATE = 0.0001
IMAGE_SIZE = [224, 224]Output Layers
K = len(folders)
x = Flatten()(pretrained_model.output)
# x = Dense(1024, activation='relu')(x)
# x = Dense(512, activation='relu')(x)
# x = Dense(256, activation='relu')(x)
x = Dense(K, activation='softmax')(x)Distributed under the Unlicense license. See LICENSE.txtfor more information.
Zachary Sherman-Burke - LinkedIn - zjshermanburke@gmail.com
Project Link: https://github.com/zjshermanburke/TransferLearning_VGG19