LLM Kit: Tools & Best Practices 🧙

1. What is This¶

In summary, this is an area for sharing experimental experiences. We aim to compare different frameworks, parameters, and datasets in a fair environment and share various methods and potential issues. Specifically:

• Framework Comparison: Compare different frameworks in a fair environment to ensure the results are just and comparable.
• Parameter Optimization: Explore the impact of different parameter settings on model performance and share optimization experiences.
• Dataset Analysis: Analyze the characteristics of different datasets and provide suggestions for data preprocessing and augmentation.
• Method Sharing: Share different experimental methods, including but not limited to model training, evaluation, and inference.
• Issue Discussion: Discuss potential issues encountered during experiments and their solutions.

2. How to Contribute¶

To add a file, simply create a new markdown file. That’s all you need to do. Images should be placed in the images folder.

3. How to Compile¶

We use myst to convert markdown files into a beautiful webpage/book.

If you need to compile this book, please install myst. Specifically, just run pip install mystmd (requires a node environment; if you don’t have a node environment, you will be prompted to install it the first time you use myst).

After that, run myst start, and you can access it at http://localhost:3000.

In addition to standard markdown, myst supports some additional usages, such as mathematical formulas, citationsVaswani et al., 2017, etc. For more details, refer to mystmd.

Through this README file, we hope to clearly convey the core content of the project, the contribution method, and the compilation method, helping more people to participate in the sharing of experimental experiences.