https://github.com/gurbaaz27/dl-models

Under Model Zoo summer project.
https://github.com/gurbaaz27/dl-models

cnn convolutional-neural-networks deep-learning gan generative-adversarial-network image-captioning infogan-pytorch machine-learning model-zoo pytorch resnet

Last synced: 27 days ago
JSON representation

Under Model Zoo summer project.

• Host: GitHub
• URL: https://github.com/gurbaaz27/dl-models
• Owner: gurbaaz27
• Created: 2020-05-17T08:43:39.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2020-07-15T07:34:08.000Z (over 4 years ago)
• Last Synced: 2024-11-14T12:11:55.842Z (3 months ago)
• Topics: cnn, convolutional-neural-networks, deep-learning, gan, generative-adversarial-network, image-captioning, infogan-pytorch, machine-learning, model-zoo, pytorch, resnet
• Language: Jupyter Notebook
• Homepage:
• Size: 3.54 MB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Deep Learning Models

Under the summer project at my institute **IIT Kanpur** "Model Zoo", we are learning and practising to implement and summarise various deep learning architectures and models.

Here is a list of the implemenations:

1. **Resnet** : [Original Paper](https://arxiv.org/abs/1512.03385) (*Implemented*)

   * Based on CNN, aimed to tackle the difficulty of training deeper networks without rise in training and testing error.

2. **InfoGAN** : [Original Paper](https://arxiv.org/abs/1606.03657) (*Implemented*)

   * Based on DCGan, aimed at disentangled representation so that network learns systematically various meaningful features of      the input sample. 

3. **ALBERT** : [Original Paper](https://arxiv.org/abs/1909.11942) (*Under implementation*)

   * Based on BERT, it addresses the GPU/TPU memory limitations and training complexities of the the former by proposing two        parameter-reduction techniques, which lowers memory, increases training speed and even achieves greater accuracy. 

4. **Image Captioning** : [Original Paper](https://cs.stanford.edu/people/karpathy/cvpr2015.pdf) (*Implemented*)

    * Based on a novel combination of CNN over images and bidirectional RNN (LSTM) over captions, using output of encoder CNN as embeddings for the decoder RNN to generate

rich descriptions of image regions.

5. **GloVe** : [Original Paper](https://nlp.stanford.edu/pubs/glove.pdf) (*Implemented*)

    * GloVe is an unsupervised learning algorithm for obtaining vector representations for words. Training is performed on aggregated global word-word co-occurrence statistics from a corpus, and the resulting representations showcase interesting linear substructures of the word vector space.