Is there a step-by step set of instructions to get Ollama (local) to work with DT

DEVONthink Artificial Intelligence
1

I successfully installed Ollama with llama3.1 as the local model. Devonthink 4.2.2 sees it in settings

image
image1548×1178 106 KB

However, when I go to the chat inspector with a document open all of the buttons are greyed out.

I am not experienced with terminal commands but I did try to enable its web server using the terminal command “launchctl setenv OLLAMA_HOST “0.0.0.0” launchctl setenv OLLAMA_MODELS ~/.ollama/models”. It didn’t help.

I must be missing some steps somewhere, but nothing I can find on the web tells me what those are. Either step by step instructions or somebody pointing out what I missed or did wrong would be appreciated.

Thanks.

2

There is no need to run any Terminal commands (outside of downloading some specific models with ollama pull…).

When the Ollama app launches, its webserver automatically runs on localhost. Ollama’s setting to be exposed to the network is for connecting other devices.

All what buttons are grayed out?

3

Upper right part of my screen (with a document open)

image
image714×544 17.7 KB

4

There is no conversation to: clear, resend, copy, or save and no recent chats to choose from. Those buttons are only enabled if you’ve got a chat going. So it’s behaving as expected.

5

Thanks. This works. Should have seen this in the documentation, perhaps[ps, but missed it.

6

No worries!