HenryNdubuaku

Thanks, noted!

So, its not a general language model, focused on tool call strictly for tiny edge-devices. There are solutions everywhere for high-capacity devices, Needle is for sub-$200 devices.

thanks!

Thanks, how can we help?

This is exactly why the confidence feature was introduced, the model knows when its wrong, we could hide that part and return a placeholder "sorry I only do function calls", would that be be…

Yes, we ablated Engrams rigorously and found that it returned world knowledge like FFN without without compute expenditure.

Random sentence is not a function call.

Users often stack a transcription model on top to get the voice prompt, then decode to actions. Think of Alexa and Siri.

Thanks, give it a splin!

This is cool!

haha, interesting analogy

1. Thanks for this! 2. We combine both on Cactus; rolling entropy method helps with per-token signal for early exit, while the probe runs on complete output. 3. Users hate this in production, drains b…

Smaller models in production can produce quite interesting outputs.

YOu are thinking along the right direction, we are going deeper into the signals.

no because we freeze the main weights and only train our added weights.

We are currently working on larger models atm! your suggestion is cool btw

Thats an interesting outlook, loosely similar.

anytime!

Correct, the study is verbose, we will compile into a neat shareable report and publish once we solve the pending caveats. Interesting username btw haha.

Interesting, did you add a commit and push tool?

Thanks, try needle and let us know!

We have enterprise users in production with similar workflows actually.

You can actually quickly test with the playground

Haha, great find, this is still an experimental checkpoint, much work is going into this now. These feedback help us improve, thanks!

I hope they don't get mad, cause we have a great working relationship with the Gemma team particularly. So, If they are uncomfortable, we'd do what they want.

Thanks!

Let us know how it goes!