https://github.com/legalaspro/rnn_gru_lstm_experiments
RNN, GRU, LSTM implementation using PyTorch and Numpy
https://github.com/legalaspro/rnn_gru_lstm_experiments
Last synced: 3 months ago
JSON representation
RNN, GRU, LSTM implementation using PyTorch and Numpy
- Host: GitHub
- URL: https://github.com/legalaspro/rnn_gru_lstm_experiments
- Owner: legalaspro
- Created: 2025-02-03T11:18:41.000Z (4 months ago)
- Default Branch: main
- Last Pushed: 2025-02-03T18:55:13.000Z (4 months ago)
- Last Synced: 2025-02-03T19:38:58.645Z (4 months ago)
- Language: Jupyter Notebook
- Size: 3.14 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: readme.md
Awesome Lists containing this project
README
# Deep Reinforcement Learning with RNN
In this project, I explored the blog by Andrej Karpathy about Recurrent Neural Networks (RNNs). Additionally, I delved into Gated Recurrent Units (GRUs) and Long Short-Term Memory (LSTM) networks out of my own interest. I implemented these models using PyTorch and also performed backpropagation from scratch using Numpy. A valuable lesson learned with Numpy was the importance of setting the initial weights correctly.
## Implementations
1. **Recurrent Neural Networks (RNNs)**
- Studied the theoretical concepts from Karpathy's blog.
- Implemented RNNs using PyTorch.
- Performed backpropagation from scratch using Numpy.
2. **Gated Recurrent Units (GRUs)**
- Explored the architecture and benefits of GRUs.
- Implemented GRUs using PyTorch.
- Conducted backpropagation from scratch using Numpy.
3. **Long Short-Term Memory (LSTM) Networks**
- Investigated the structure and advantages of LSTMs.
- Implemented LSTMs using PyTorch.
- Executed backpropagation from scratch using Numpy.
## Conclusion
Through this project, I gained a deeper understanding of various recurrent neural network architectures and their implementations. The hands-on experience with PyTorch and Numpy reinforced my learning and provided valuable insights into the workings of RNNs, GRUs, and LSTMs. A valuable lesson learned with Numpy was the importance of setting the initial weights correctly.