Viewing profile — scriptsmith

It's a reference to this story where an OpenAI model broke out of its sandbox during cyber benchmarking and hacked into HuggingFace, in order to obtain test solutions: https://news…

I made these to create styled text art for the MOTD of my university's HPC login node: https://github.com/ScriptSmith/tart https://tart.sh/

Is GitHub Copilot the best positioned platform for enterprise? They support Claude, GPT, Gemini, and now even open weight models. Larger orgs are paying at API rates anyway so it c…

Definitely took longer than I was expecting, then after two weeks I thought we'd never get it.

And now is this going to be a one-off, or routine with every new generation of models?

It's based on a Gemma 3n model, and yeah it's not the best. But if you have a use case that needs constrained JSON output for example, it's pretty neat. Maybe it would do better wi…

I've got some demos of what the new Prompt API in Chrome that uses a local model can do: https://adsm.dev/posts/prompt-api/#what-could-you-build-with... As OP says, it shines in co…

Chromium mostly does not support this, because it doesn't have the binary blob required to run the inference. However, it does still download the model weights and expose the Langu…

> My understanding is that you’ll have to explicitly agree to download alternate models in the future, per the specification I don't get that from the spec. Closest thing I could f…

Depends on where you get it. By default the flags will be enabled, but some packagers may choose to disable them. I haven't seen a major distro release chromium 148 yet. Weirdly th…

I wrote a more detailed blog post here: https://news.ycombinator.com/item?id=48028662

Chromium doesn't support this API because it needs a binary blob to run the inference, although in theory it may still be configured to download the weights: https://adsm.dev/posts…

It can only be called after the user has interacted with the page, but there's no dialogue from the browser https://developer.chrome.com/docs/ai/get-started#user-activa...

Author here. After trying out the Prompt API over the last week, I wrote up some details on the chromium internals, how to use the API, and made some toy demos. It's a 4 GB model t…

In my understanding, yes. I wrote a blog post about some of the internals here: https://news.ycombinator.com/item?id=48028662

Google has been trialling the Prompt API in chrome for the over a year, so before Gemma 4 existed. But they are indicating they'll move to Gemma 4: https://groups.google.com/a/chro…

It's based on Gemma 3n, and it's not the best. I find it works fine for simple classification, translation, interpretation of images & audio. It can write longer prose, but it's pr…

Those flags will exist already, but will default to enabled in 148. That other flag is for using a different open-source inference engine to the (from what I can tell) closed-sourc…

If Chrome has the #optimization-guide-on-device-model and #prompt-api-for-gemini-nano flags enabled, either because it's part of some Origin Trial / Early Stable Release or somethi…

Yes, you could turn it around to say that using Anthropic models in Cursor, Copilot, Junie, etc. is 'subsidising' Claude Code users.

The "First-class syntactic selection" reminds me of my most used shortcut(s) in Jetbrains IDEs: the Expand / Shrink Selection. Ctrl + W Ctrl + Shift + W https://www.jetbrains.com/h…

Yes, I've used the v3.2 3B-Instruct model in a Slack app. Specifically using vLLM, with a template: https://github.com/vllm-project/vllm/blob/main/examples/tool... Works as expecte…

To keep on-top of tabs in Firefox, I use 'Auto Tab Discard' [1] to discard tabs after a certain amount of inactivity. Then when I need to clean up my list of tabs, I click on any d…

Someone posted this in the thread, but I also can't log in to verify > Summary > CrowdStrike is aware of reports of crashes on Windows hosts related to the Falcon Sensor. > Details…