I’ve also experienced it for a long time now. I’m not exactly sure what causes this, but I’ll try to look into it again.

Perhaps it’s issue #281?

It’s fixed in the nightly builds. If it’s a different problem, I think we need to open a separate ticket for it.

Yes, it is fixed in the nightly builds and will be included in the next release.

I had to rebuild this version with Java 17 instead of Java 11. Now it seems the build was successful, so it will be available on F-Droid soon 🥳

I’m glad you were able to fix this issue. I’m not sure what could have caused it, and I couldn’t reproduce it. Perhaps some setting caused it?

I’ve created a new release (v0.2.1) to fix the issues in the F-Droid building process(functionally, it’s the same as v0.2.0). Hopefully, this means the latest version will soon be available on F-Droid.

Did you manually re-login, or did Eternity tell you to do so? Just curious if I’ve missed anything 😅

It will take some time, unfortunately. The build has failed for some reason. I’ve observed this in my pipeline as well. I am still investigating the issue.

I think this issue is fixed in this release.

Yes, it’s going to be updated very soon

Yes, I’ve released it as beta so that Google Play users can also try the new changes before the release. These builds are similar to the nightly, but I have to build them manually so they are updated less frequently.

I’m not sure what causes this issue. Does this happen when you click on a post, or does it happen when you are browsing for posts?

Also, what can you see on the web UI for these posts?

From what I’ve seen, it’s definitely worth quantizing. I’ve used llama 3 8B (fp16) and llama 3 70B (q2_XS). The 70B version was way better, even with this quantization and it fits perfectly in 24 GB of VRAM. There’s also this comparison showing the quantization option and their benchmark scores:

Source

To run this particular model though, you would need about 45GB of RAM just for the q2_K quant according to Ollama. I think I could run this with my GPU and offload the rest of the layers to the CPU, but the performance wouldn’t be that great(e.g. less than 1 t/s).

Yes, you can find it here.

Are you using mistral 7B?

I also really like that model and their fine-tunes. If licensing is a concern, it’s definitely a great choice.

Mistral also has a new model, Mistral Nemo. I haven’t tried it myself, but I heard it’s quite good. It’s also licensed under Apache 2.0 as far as I know.

I haven’t tested it extensively, but open webui also has RAG functionality (chat with documents).

The UI it self is also kinda cool and it has other useful features like commands (for common prompts) and searching for stuff online (e.g. with searx). It works quite well with Ollama.

I have resolved this and other issues in the latest nightly build. I have also uploaded the fix to Google Play (it’s currently under review). If we do not encounter any further problems, I could release it soon.

I want to add kbin/mbin support in the future, but for now, Lemmy is my primary focus.

Thank you! I missed that. I will try to fix that too

Ahhh, your insurance is updated to Lemmy 0.19.5. The update will fix this issue.