Ten billion profiles/captures = one amp

[chicken]

I guess we are no longer talking about the new NAM parametric thing?

Did you find the code for it? I guess it uses some sort of model evolution (i.e. to fit topology better first).

I mean.. you don't have to. But it would be nice if you did.

Arguing with word definitions is like arguing with axioms. I posted definition google gave me from Oxford. You posted from ChatGPT. I used my set of axioms (definitions) and you used ChatGPT's. That is it to it.

 

[laxu]

Did you find the code for it? I guess it uses some sort of model evolution (i.e. to fit topology better first).

Doesn't seem to be in his Github or is a private repository.

 

[Lysander]

Doesn't seem to be in his Github or is a private repository.

Was going to mention the same. The blurb in his post is a bit... vague on details, but i was hoping to dive a bit deeper into the code. As far as i can tell, there's none published just yet.

 

[Orvillain]

Response by NAM designer?

The first publicly-available parametric Neural Amp Model

Today, I'm releasing ParametricOD, a plugin that uses NAM's parametric modeling capabilities to give you a model of my overdrive pedal that is accurate across the full range of the pedal's knobs and switches. Get it here: Users | Neural Amp Modeler. GUI for the plugin. The plugin is available in...

www.neuralampmodeler.com

Optimal experimental design - Wikipedia

en.wikipedia.org

That is the link that Steve referenced vis-a-vis, addressing the "too many static captures required" problem. I'm happy to report that the word guess features a grand sum of zero times.

 

[jay mitchell]

It is completely self evident that guessing and estimating (and the other things I mentioned) are not equivalent.

Only to a delusional mind. There are different types of guesses - e.g., "wild-ass," "blind," "educated," "informed" - but they are all still guesses. Mathematics as we know it wouldn't exist without methodologies for using guesses and refining them based on the results. Human learning is based on serial guesses with refinement based on the results.

Nobody involved in the ML space uses informal words like "guess" in any serious capacity.

Now that is a blind guess.

 

[chicken]

Estimation theory is a branch of statistics that deals with estimating the values of parameters based on measured empirical data that has a random component.

Guessing implies a level of randomness that machine learning and neural networks are not based on.

How much random is too much?

 

[Orvillain]

Mathematics as we know it wouldn't exist without methodologies for using guesses and refining them based on the results.

Ahem. Hypothesis.

<ducks>

 

[Orvillain]

How much random is too much?

No idea!

 

[chicken]

No idea!

I like how ChatGPT added formal words like "computation" to estimation definition. It's like "sales engineers", "cleaning manager". Everything is a computation. And you can't make a guess without computation. You can try to formally separate guess and estimation by degree/amount/quality of data used to make prediction. But then it's subjective. That's my educated guess.

 

[Orvillain]

And you can't make a guess without computation.

We've already established that you can.

I guess you have blue hair.

See. No computation required. I'm just guessing.

But if I want to estimate....

Okay... so your username is chicken. Probably means you have red hair.

 

[chicken]

right, so you insist on taking qualiity of prediction into account and retroactively renaming process based on it.

I'm not a chicken. I'm a human meatball. And if my train arrives 9 times late I will say "I guess it won't be on time today too". I don't know if there is anybody who will say "estimate". And 9 data points are ok for me.

 

[Orvillain]

right, so you insist on taking qualiity of prediction into account and retroactively renaming process based on it.

I'm not a chicken. I'm a human meatball. And if my train arrives 9 times late I will say "I guess it won't be on time today too". I don't know if there is anybody who will say "estimate". And 9 data points are ok for me.

Personally I'd say "I think it might be late today as well" but I wouldn't walk away from the platform and get a taxi instead, because taxi's are more expensive and in my neck of the woods, they're much slower than the train to get you to your final destination. Probably because they all rely on sat nav rather than black cab knowledge of the streets. But that's a whoooooolllleeeee other discussion!

In either case, I already acknowledged that these things are related. They're just not equivalent. And you've both agreed with me in your own ways. You're being much nicer about it, but in fairness, I didn't call you a bitch in a previous thread, giving you some sort of grievance against me... so I suppose that would explain the disparity there.

 

[chicken]

They are related because they are the same. Word estimation is just redressed guess to make it more grant friendly.

I mean the only reason I joined is because even if I knew everything in the field I wouldn't nag somebody for using non approved word on a guitar forum. If you first objected word "guess" with that ChatGPT screenshot I wouldn't dare to argue.

 

[Orvillain]

Word estimation is just redressed guess to make it more grant friendly.

That one is a bit too conspiracy minded for me, to be honest.


Bearing in mind that all this started with me saying:

Machine learning is not guessing.

and:

Listen. If you're writing a sonnet or some emo poetry to get into Byron's pants, then sure. You can get away with that kind of logic with a squint.

But if you want to be accurate and scientific, then words and their differences actually mean something.

which was clearly meant to be humourous in part, but actually hold a valid point .... well.... I feel pretty good about my conduct here.

I don't think yourself or Jay should think so highly of yourself, given your steadfast refusal to accept that different words sometimes mean slightly different things. I sort of expect it from Jay, given how he parades these guitar forums like a hallway prefect with a Granny Smith addiction.... but you? I don't know you that well. You seem alright in the main. Maybe have a sense of humour failure every now and then, but in the main, okay.

The fact is - I didn't "nag" @BenIfin at all. So your characterisation of that is misleading. You obviously have some other motivation. But perhaps there's I am just guessing. Or estimating. Or predicting. I don't know. Really hard to tell now.

 

[chicken]

this is how I saw it - complex topic, people trying to wrap their head around it, which is insanely good imo. all ML stuff must be accessible to general public since the manipulative danger they carry (it captures my amp well, maybe we don't need elections then <- that line of thinking). And instead of making it more accessible in my view you went with pointing miniscule difference (I still recognize it only formally). I also objected the form "Machine learning is not guessing." looked like instead of making technology accessible to fellow forum members your were showing off (or was it snobby? my English fails me here).

In short, I saw dismissive, snobby comments, not a friendly explanation and that triggered me.

 

[chicken]

refusal to accept that different words sometimes mean slightly different things

too generic. I only refuse to accept that "to estimate" is a little more than grant approved "to guess"

 

[Orvillain]

this is how I saw it - complex topic, people trying to wrap their head around it, which is insanely good imo. all ML stuff must be accessible to general public since the manipulative danger they carry (it captures my amp well, maybe we don't need elections then <- that line of thinking). And instead of making it more accessible in my view you went with pointing miniscule difference (I still recognize it only formally). I also objected the form "Machine learning is not guessing." looked like instead of making technology accessible to fellow forum members your were showing off (or was it snobby? my English fails me here).

In short, I saw dismissive, snobby comments, not a friendly explanation and that triggered me.

Probably a language barrier thing here then. My comments weren't at all snobby. They were just factual. You're putting an emotional slant on my posts which wasn't there.

I was trying to show @BenIfin that his comment about ML being guesswork was over simplification; and I stand by that 100%. I don't think that qualifies as showing off. As I'm ever reminded on discussion forums, I have very little to show-off about!!!

It isn't really my job to make technology accessible to fellow forum members. That isn't the purpose of my involvement in this thread.

The way I see it, you and Jay jumped down my throat for very banal and superficial reasons. I get why Jay would've done it. I didn't get why you did it. But now I do. You perceive snobbishness where there isn't any.

 

[yeky83]

And f.w.i.w .... based on my distant maths memory .... Gain, Bass, Mid, Treble, Presence, M-Vol = 6 knobs .... even with just 5 data points on each .... that works out to 7,776 data points / captures to accurately measure their real interactivity ........ do 10 increments of 1 over each of the 6 knobs and then its 60,466,176 data points / captures to accurately measure their real interactivity .... I love big numbers :)

Wrong maths. increments^knobs, not knobs^increments.

 

[Orvillain]

too generic. I only refuse to accept that "to estimate" is a little more than grant approved "to guess"

But you already did accept it when you offered your distinction from earlier.

Guessing - making a decision without sufficient information.
Estimating - making approximation based on incomplete information.

I can actually kind of get on board with that distinction. It falls in line with my understanding of the differences between the terms at least.

 

[chicken]

Then what was the purpose of your involvement? Just to point out what you think is incorrect? That totally ok. Especially if you didn't want to educate.