World's first gptq 4bit quant of `glm-4-9b` model. | |
Autogptq PR: https://github.com/AutoGPTQ/AutoGPTQ/pull/683 | |
Please note ChatGLM has tendency to switch from English to Chinese in mid-reply or in direct reply to English prompt. This issue happens in both native and quantized model and needs further investigation. |