Hello Mario,
I'm trying to test AutoTagger on my notebook with RAM 16.0 GB (15.6 GB usable) and SSD 933 GB running on Win11. I have Ollama installed and downloaded AI model.
NAME ID SIZE MODIFIED
llava:latest 8dd30f6b0cb1 4.7 GB 27 minutes ago
When running AutoTagger there is error. Is it possible somehow to bypass the error?
03.09 15:33:06+ 0 [17E94] 01 W> AutoTagger: Aborting because of error [600] '{"error":{"message":"model requires more system memory (6.1 GiB) than is available (5.8 GiB)","type":"api_error","param":null,"code":null}}
' 'V:\develop\IMatch5\src\IMEngine\IMEngineAIAutoTagger.cpp(369)'
03.09 15:33:06+ 0 [1B1D8] 01 W> UpdateQueue (AutoTagger): Service error 600 '{"error":{"message":"model requires more system memory (6.1 GiB) than is available (5.8 GiB)","type":"api_error","param":null,"code":null}}
' for file [3618] 'V:\develop\IMatch5\src\IMEngine\IMEngineUpdateQueueAutoTagger.cpp(233)'
Thanks in advance,
Dmitry
The message is clear to me: model requires more system memory (6.1 GiB) than is available (5.8 GiB).
When IMatch starts, memory utilization is already at 80%. Total: 15,943MB, available: 3,161MB.
Only 3.1 GB are available, which is very little, even for a 16 GB system.
Check which other applications and background services are running and close them when not needed.
Especially if you use Adobe software, which are all known memory hogs.
Note that if Ollama has to fall back to system memory (normal RAM), there is not enough memory on the graphic card to hold the model. And this will make everything 10 times slower, at least. For LlaVA you need 8 GB VRAM on the graphic card.
Thank you Mario, I've closed all running apps and finally AutoTagger produces the result. Issue closed.
Will play with the AI AutoTagger to see how it could be used.