1,747 karma · joined May 6, 2022
Olympia Neo Mini
Slightly larger:
Samsung Galaxy Xcover 5
Cyrus CM17 XA
None with 5G though.
Have a link?
It works just fine with OpenContacts.
This works fine with Open contacts from F-Droid.
IIUC Ubuntu and systemd however choose to dump the process manually for some reason and for that you need to have same permission as the target process.
That part sounds pretty cool if it works reliably.
However the brand-new Gemma 3n E2B and E4B models might fit with vision.
You absolutely can't, since you need to link to the system libGL.so and friends for gpu acceleration, libva.so for video acceleration, and so on.
And you really notice that the model is dumber on GPU, because OpenGL doesn't take accuracy that seriously.
[0] https://blog.tensorflow.org/2020/08/faster-mobile-gpu-infere...
CPU:
7.37 seconds to first token
35.55 tokens/second prefill speed
7.09 tokens/second decode speed
27.97 seconds to complete the answer
GPU:
1.96 seconds to first token
133.40 tokens/second prefill speed
7.95 tokens/second decode speed
14.80 seconds to complete the answer
Also:
> These models were evaluated at full precision (float32)
For 4B effective parameters that's 16 GB ram.
Hint: You have to set the Max tokens to 32000 for longer conversations. The slider makes it look like it's limited to 1024, just enter it manually.
Download the Edge Gallery apk from github: https://github.com/google-ai-edge/gallery/releases/tag/1.0.0
Download one of the .task files from huggingface: https://huggingface.co/collections/google/gemma-3n-preview-6...
Import the .task file in Edge Gallery with the + bottom right.
You can take pictures right from the app. The model is indeed pretty fast.
Ollama does, please try it.
Ollama is written in golang so of course they can not meaningfully contribute that back to llama.cpp.
You can do that on Android with NetGuard.