HNHacker News
TopNewBestAskShowJobs

nolist_policy

1,747 karma · joined May 6, 2022

submissionscomments
nolist_policy··on Bcachefs may be headed out of the kernel
Kent can appoint a suitable maintainer if he wishes. That's his job, not Linus'.
nolist_policy··on Bcachefs may be headed out of the kernel
Generally you need a maintainer for your subsystem who sends pull requests to Linus.
nolist_policy··on Verifiably Correct Lifting of Position-Independent x86-64 Binaries (2024)
Where is the source code?
nolist_policy··on Normalizing Flows Are Capable Generative Models
Gemma 3n E4B runs at 35tk/s prompt processing and 7-8 tk/s decode on my last last last gen flagship Android.
nolist_policy··on Normalizing Flows Are Capable Generative Models
You can bundle and use Gemma commercially[1].

[1] https://ai.google.dev/gemma/terms

nolist_policy··on Rust in the Linux kernel: part 2
Could you write interrupt handlers in it? You can't even do synchronous signal handling in Java.
nolist_policy··on Fairphone 6 is switching to a new design that's even more sustainable
Samsung Galaxy Xcover 4s

Olympia Neo Mini

Slightly larger:

Samsung Galaxy Xcover 5

Cyrus CM17 XA

None with 5G though.

nolist_policy··on Tell me about your favorite tree (a slow-web proposal)
> A few days ago someone posted an article linking luddites to the NASA moon landing program to AI-generated furry porn.

Have a link?

nolist_policy··on EU rules for durable, energy-efficient and repairable smartphones and tablets
Not true, no one stops you from publishing a app for Android 1.0 on the Play store. It's going to be hard though, you won't be able to request any permissions etc.
nolist_policy··on Samsung embeds IronSource spyware app on phones across WANA
> Without it, I can’t write to someone or call.

It works just fine with OpenContacts.

nolist_policy··on How easy is it for a developer to "sandbox" a program?
> In my phone "phone's app" I have the option to message on whatsapp next to each number, but when I click on it, whatsapp doesn't let me just message the number, it asks for permission to view all my contacts.

This works fine with Open contacts from F-Droid.

nolist_policy··on Slowing the flow of core-dump-related CVEs
The coredump is sparse in elf format.
nolist_policy··on Slowing the flow of core-dump-related CVEs
You're thinking to complicated. You can configure the coredump helper in a way that the kernel presents it with the coredump on stdin. So you drop privileges and self-sandbox at startup and only then start reading the coredump from stdin.

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.

nolist_policy··on GCP Outage
Google's local models as well (Gemini Nano/Gemma 3n)
nolist_policy··on OpenPOWER Foundation – Open-Source / Open Hardware PowerPC CPU ISA
No
nolist_policy··on Show HN: GnuRAMage – A Bash tool to sync files between HDDs and RAM disks
I have found that temporary files are flushed much less by btrfs than ext4. ext4 will do a disk write io for every created file, even if it's only a temporary file. For btrfs, new files that are deleted before write back will never hit the disk.
nolist_policy··on Shop with AI Mode, use AI to buy and try clothes on yourself virtually
> Just tap “track price” on any product listing and set the right size, color (or whatever options you prefer) and the amount you want to spend. Keep an eye out for a price drop notification and, if you’re ready to buy, just confirm the purchase details and tap “buy for me”.

That part sounds pretty cool if it works reliably.

nolist_policy··on What even is a small language model now?
Gemma 3 4B QAT int4 quantized from bartowsky should barely fit in a 4GB Raspberry Pi, but without the vision encoder.

However the brand-new Gemma 3n E2B and E4B models might fit with vision.

nolist_policy··on The Future of Flatpak
> Can't you just distribute static binaries? They are not that hard to compile.

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.

nolist_policy··on Gemma 3n preview: Mobile-first AI
It uses tflite in the background which can GPU accelerate with OpenGL ES 3.1 or OpenCL[0]. So it should work on pretty much any SOC.

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...

nolist_policy··on Gemma 3n preview: Mobile-first AI
Gemma-3n-E4B-it on my 2022 Galaxy Z Fold 4.

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

nolist_policy··on Gemma 3n preview: Mobile-first AI
Hmm, the Aider polyglot benchmark has been removed from the huggingface readme.

Also:

> These models were evaluated at full precision (float32)

For 4B effective parameters that's 16 GB ram.

nolist_policy··on Gemma 3n preview: Mobile-first AI
Okay from some first tries with story writing, gemma-3n-E4B-it seems to perform between plain Gemma 3 4B and 12B. It definitely retains the strong instruction following which is good.

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.

nolist_policy··on Gemma 3n preview: Mobile-first AI
You can try it on Android right now:

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.

nolist_policy··on Why Vibe When You Can Fly Machtiani Is Now Open Source
Quick question: Does Machtiani pull in whole files into the prompt/context or just specific parts?
nolist_policy··on Ollama's new engine for multimodal models
llama.cpp clearly does not support iSWA: https://github.com/ggml-org/llama.cpp/issues/12637

Ollama does, please try it.

nolist_policy··on Ollama's new engine for multimodal models
For one Ollama supports interleaved sliding window attention for Gemma 3 while llama.cpp doesn't.[0] iSWA reduces kv cache size to 1/6.

Ollama is written in golang so of course they can not meaningfully contribute that back to llama.cpp.

[0] https://github.com/ggml-org/llama.cpp/issues/12637

nolist_policy··on The recently lost file upload feature in the Nextcloud app for Android
Native code can work with files inside the app data folder. Then you can expose that folder with SAF in a separate Activity. Termux does this and it works really well.
nolist_policy··on The recently lost file upload feature in the Nextcloud app for Android
> End user should have the ability to block network connects to and from select networks, infrastructures, and applications. Example an application like ZoneAlarm or Open Snitch.

You can do that on Android with NetGuard.

nolist_policy··on The recently lost file upload feature in the Nextcloud app for Android
No, they would just use the SAF API.
← PreviousPage 6 of 25Next →