Re:Llama 3.2: Revolutionizing edge AI and vision with open, ...
by Closi on Wednesday September 25, 2024 @05:46PM
(#41652200)
What about OpenAI Structured Outputs? This seems to do exactly this.
Re:Llama 3.2: Revolutionizing edge AI and vision with open, ...
by zackangelo on Wednesday September 25, 2024 @05:55PM
(#41652284)
I'm building this type of functionality on top of Llama models if you're interested: https://docs.mixlayer.com/examples/json-output
Re:Llama 3.2: Revolutionizing edge AI and vision with open, ...
by refulgentis on Wednesday September 25, 2024 @07:09PM
(#41652825)
I'm writing a Flutter AI client app, integrates with llama.cpp. I used a PoC of llama.cpp running in WASM, I'm desperate to signal the app is agnostic to AI provider, but it was horrifically slow, ended up backing out to WebMLC.
What are you doing underneath, here? If thats secret sauce, I'm curious what you're seeing in tokens/sec on ex. a phone vs. MacBook M-series.
Or are you deploying on servers?
Re:Llama 3.2: Revolutionizing edge AI and vision with open, ...
by refulgentis on Wednesday September 25, 2024 @05:55PM
(#41652278)
Correct, I think so too, seemed that update must be doing exactly this. tl;dr: in the context of Llama fn calling reliability, you don't need to reach for training, in fact, you'll do it and still have the same problem.
Re:Llama 3.2: Revolutionizing edge AI and vision with open, ...
by noahbp on Thursday September 26, 2024 @02:44AM
(#41655228)
What kind of damage is done by constraining token generation to valid JSON?
Re:Llama 3.2: Revolutionizing edge AI and vision with open, ...
by snovv_crash on Thursday September 26, 2024 @06:36AM
(#41656757)
Yeah, from my experience if you prompt something like:
respond in JSON in the following format: {"spam_score": X, "summary": "..."}
and _then_ you constrain the output to json, the quality of the output isn't affected.
In general, you'll do a ton of damage by constraining token generation to valid JSON - I've seen models as small as 800M handle JSON with that. It's ~impossible to train constraining into it with remotely the same reliability -- you have to erase a ton of conversational training that makes it say ex. "Sure! Here's the JSON you requested:"