Regarding Fine-Tuning Before Quantization of GGUF Models

#1
by Baharababah - opened

Hello, I’ve been using some of your quantized GGUF models and greatly appreciate your work in making them available to the community. I had a quick question regarding your quantization process:

Do you fine-tune the base model in any way prior to performing the quantization (e.g., for q4_K, q2_K, etc.), or are the models quantized directly from the original pretrained weights?

Understanding this would help me better evaluate the performance trade-offs and potential use cases for these models.

Thank you in advance for your time, and again, I appreciate the valuable contributions you're making!

Second State org

No, we have not done any finetuning before quantization. Thanks.

juntaoyuan changed discussion status to closed

Sign up or log in to comment