jpan127 / nn-compression-research

Researching optimal compression for neural network models.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

nn-compression-research

Team

We are a team of Computer Engineering students at San Jose State University, under the guidance of Professor Hyeran Jeon and Professor Xuan Guan, researching optimal compression methods for neural network models.

Current Team Members: Jonathan Pan, Spoorthi Mysore Shivakumar

Previous Contributors: Mina Yi, Jaswanth Bhimanapalli

Directory Structure

.
├── Huffman       # Files for huffman encoding overlay
│   ├── bitstream       # .bit and .tcl files for overlay
│   ├── hls             # HLS synthesis and co-simulation reports
│   ├── ip              # IP files and AXI Lite memory map
│   ├── notebook        # .ipynb Pynq notebook
│   ├── src             # HLS source code
│   └── vivado          # Vivado block diagram, summary, and memory map
├── LZW           # Files for lzw overlay
│   ├── bitstream       # .bit and .tcl files for overlay
│   ├── hls             # HLS synthesis and co-simulation reports
│   ├── ip              # IP files and AXI Lite memory map
│   ├── notebook        # .ipynb Pynq notebook
│   ├── src             # HLS source code
│   └── vivado          # Vivado block diagram, summary, and memory map
├── RLE           # Files for run length encoding overlay
│   ├── bitstream       # .bit and .tcl files for overlay
│   ├── hls             # HLS synthesis and co-simulation reports
│   ├── ip              # IP files and AXI Lite memory map
│   ├── notebook        # .ipynb Pynq notebook
│   ├── src             # HLS source code
│   └── vivado          # Vivado block diagram, summary, and memory map
└── test          # Testing compression ratios with software compressors
    ├── programs        # Windows executables with a single input file parameter
    └── results         # Text results after running compression on various files  

About

Researching optimal compression for neural network models.


Languages

Language:Jupyter Notebook 73.9%Language:Tcl 9.9%Language:VHDL 8.0%Language:Verilog 5.0%Language:C 1.7%Language:C++ 1.6%Language:Makefile 0.0%