Funny result from the web demo. I'm well aware that it's an extremely small and, well, stupid, model, but even so:
Query: HN
Result:
{ "function_calls": [ { "name": "lock_door", "arguments": { "door": "front door" } } ], "reasoning": "User wants to lock the door. No specific door mentioned, so use 'front door' as default.", "confidence": 0 }
I'd expect it to at least ignore (call no tools) for the queries that it doesn't understand. And it seems like it does do that, just not consistently.
The website says the model is for "tool calling, device use, and structured extraction". Your example just doesn't seem to be very relevant. FWIW, it did a pretty good job for tool calling when I tried it, and I think it could be pretty nice to have this running on locally and integrate with Home Assistant.
False positives are definitely relevant and worth measuring - natural language interfaces always have a discoverability problem, i.e., users not knowing what actions the system does and does not support. If the frontend of that system lacks the ability to reject unsupported commands, weird stuff happens.
Nonetheless, this is very cool work! If I can offer a small suggestion to the team at Cactus, it would be to evaluate your releases on some usability criteria (including false positives). Any serious integrator or adopter of these models would want to have that information available.
> "confidence": 0
OP and the linked page talk about the confidence score and using it as an action threshold, so it looks like an appropriate total response to me.
Right, but that's not the same thing as reporting a benchmark across a test set. It doesn't help me determine how well the model does across a decently-large sample size of commands. It doesn't tell me with what reliability the confidence will be below a given threshold when it should be, above that threshold when it should be, etc.
Thanks, noted!
This is extremely impressive if it works. But on the other hand, if the number of cases where it works as expected is lower than what we could catch with a (old-Siri-style) heuristics based approach, and the rest fails in unpredictable ways, I'd prefer the dumb old "I did not understand that" response.
I think the test above is about tool calling... That's how I read it. The issue here is known as "out of distribution detection" in the old-timey classification world.
I am not sure how a micro model will fundamentally solve it. Would love to understand what dannyw and team did there?
How did you draw an association between dannyw and Cactus? There are no 'Danny's on the list of GH contributors nor is there one named in the paper. Just curious.
Ah, my bad. I may have misread it.
"confidence": 0, so I guess you could threshold it
Only if the confidences are calibrated, and they probably aren't. Any idea how the confidence is derived?
no, this is the appropriate response to hearing the words "HN" :P
It seems to love locking doors. I was hoping this would dim the lights.
> I'm hungover
{ "function_calls": [ { "name": "lock_door", "arguments": { "door": "front door" } } ], "reasoning": "User wants to lock the door. 'hungover' implies a security door. No specific door named, so use 'front door' as default.", "confidence": 0 }
Cannot unlock, though:
Prompt: "Unlock my doors"
{ "function_calls": [], "reasoning": "No tool for unlocking or locking door exists. lock_door only performs lock/unlock actions, not unlock.", "confidence": 0.9492 }
I've got an identical output with the prompt "do not lock the door".
yeah I got the same, almost like its biased heavily towards that as the 0 ranking -- my prompt was just the word 'potato'
Was that the first message you sent it?
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 better or you prefer to see everything?
confidently stating that the confidence is right seems like a big claim
[dead]