Quote:
Originally Posted by vyau
Hello:
I am running into a problem.
I have the latest Calibre on Linux
I am using Ollama with Qwen2.5 models.
On any PDF books, when I ask any kind of question, I would get a read timed out error (60).
I am guessing 60 is 60 seconds.
Is there anyway I can increase this timeout? I believe this maybe an http timeout error.
I am on a CPU only laptop, no Nvidia.
thanks
|
Hey vyan,
This message indicates that your request exceeded the 60-second generation time and was aborted.
I tested two small models locally on a Mac mini M2 machine: qwen2.5:latest (you must specify qwen2.5 on the configuration page to run successfully) and qwen2.5:0.5b. There were no issues.
I recommend testing with a smaller model first. It's possible that the generation time is too long due to a mismatch between your machine and the model size.
I will fix this issue and try to increase the maximum request time. Thank you for your valuable feedback!