So I finally got around to trying to set up a local AI model on my old desktop, following one of those popular guides. I’ve hit a wall where the installation script just keeps throwing a dependency error, even though I swear I have the right version. Has anyone else run into this with ollama and found a workaround that doesn’t involve starting completely from scratch?
That install grind with ollama on an aging PC sounds frustrating and almost funny at once
Often those dependency errors on ollama come from a mismatch between system libraries and what the installer expects when the CPU or glibc is older than the target you saw in the guide
Maybe the guide you followed assumes a recent OS or a container path and your rig is trying to do everything locally which can trip the script
This is the kind of problem that makes people shrug and just wipe it and start over which I get but it can be a stall not a dead end
Instead of chasing the fastest fix you might ask if keeping it locally is worth the effort and choose a lighter route or a different tool that fits the hardware
If you are testing the model for writing you might notice how ollama handles prompts even with rough setup and that tells you where to focus when you redo the install
A small nudge to think about the root cause could be to try a minimal environment and see if the error follows the package that may reveal whether the issue is the build or the host