In some cases RisuAI is no response

[KamiaoChan]

When utilizing the web UI's API (GGML) and employing a CPU for inferring with large models, it has been observed that generation requests taking around or more than 100 seconds are not received by the RISUAI panel. It appears that content exceeding the 100-second threshold is being treated as timeouts and subsequently discarded. Is there a rule for discarding based on timeouts? How can this issue be resolved?

[cool-delete]

Suggest this issue preferred fix