Can you explain in a step by step manner how we can implement this on our own model and dataset?

mit-han-lab / smoothquant

[ICML 2023] SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models

https://arxiv.org/abs/2211.10438

MIT License

1.1k stars 127 forks source link

Can you explain in a step by step manner how we can implement this on our own model and dataset? #49

Open shahaamirbader opened 11 months ago

shahaamirbader commented 11 months ago

For beginners like me it is very hard to understand the implementation of Smoothquant on a model of own choice as well as dataset. A small step by step guide or a video, explaining where exactly to make changes in the code, could help understand in the implementation of Smoothquant on our models.