Version 1.8.8 is out and supports Gemma 3

[a-ghorbani]

Version 1.8.8 is out and now supports Gemma 3!

It may take 1–2 days to be available in stores. But, for Android, you can access it directly via the APK: https://github.com/a-ghorbani/pocketpal-ai/releases/download/v1.8.8/app-release.apk

You're welcome! 😊

---

Update:
iOS users can now join the TestFlight version here: https://testflight.apple.com/join/B3KE74MS

---

Update (March/15):
Version v1.8.9 for Android is released and fixes long text issues for Gemma 3.
You can try it now by joining the open testing (and soon, the full release).
https://play.google.com/store/apps/details?id=com.pocketpalai

[Disonantemus]

I can't upload benchmarks from APK downloaded from github, error:

[a-ghorbani]

This is expected. Only the official release from App/Play store will let you upload the benchmarks. This is to make sure the benchmarks are valid.

[i0799]

PocketPal中下载了一个Hugging face中的deepseek R1 1.5B大语言模型到本地，但是无法加载运行，是什么原因？旧版本的可以正常使用，最近更新到1.8.8版本后再下载大语言模型就无法加载运行。

[a-ghorbani]

请问您可以分享您的操作系统版本和设备型号吗？

[i0799]

Version 1.8.8 is out and now supports Gemma 3!

It may take 1–2 days to be available in stores. But, for Android, you can access it directly via the APK: https://github.com/a-ghorbani/pocketpal-ai/releases/download/v1.8.8/app-release.apk

You're welcome! 😊

---

Update:
iOS users can now join the TestFlight version here: https://testflight.apple.com/join/B3KE74MS

PocketPal中下载了一个Hugging face中的deepseek R1 1.5B大语言模型到本地，但是无法加载运行，是什么原因？旧版本的可以正常使用，最近更新到1.8.8版本后再下载大语言模型就无法加载运行。

[Disonantemus]

I downloaded the Hugging Face deepseek R1 1.5B language model in PocketPal, but I can't load and run it locally; I'm not sure what the issue is. It worked fine with an older version; I updated to version 1.8.8, and after that, I can't load the large language model anymore.

Something similar happened to me when:

I was using older Google play version, then I needed to uninstall it, to install newer version APK from GitHub.

But when I did start 1.8.8, all previously downloaded models were gone.

I didn't mind to much, because there were just like 4 small models, and I have not so slow connection.

I was worried only for the space storage, but didn't find any large files.

[MeowCata]

长上下文中gemma3出现问题

[a-ghorbani]

thanks for sharing. This is a known issue on Android: #240, ggml-org/llama.cpp#12357

[a-ghorbani]

This should be fixed now (version 1.8.9)

[MeowCata]

This should be fixed now (version 1.8.9)

Yep.

[Disonantemus]

长上下文中gemma3出现问题

I don't understand what says in you image, but I think it's related to sometimes model gets stuck repeating a small phrase again and again.

In my case, I did stop the answer, clear chat, and ask another thing without problem.