Skip to main content
Overfitting

Overfitting

Search complete. 8 mentions across 7 episodes found for "Overfitting".

Sep 9, 2026

Joel NiklausGUEST
66:05
But yeah, I, I don't know.
Joel NiklausGUEST
66:06
To me, it seems a bit over-engineered and a bit, uh, like overfitted [chuckles] in the end.
Joel NiklausGUEST
66:12
So yeah, I didn't do that.
Joel NiklausGUEST
66:14
So yeah, it, it looks like, um, there really isn't a, a good proxy that you can use, and you really need to train and evaluate each data set independently, it looks like, unfortunately.
Hustle and Grind

Hustle and Grind

Mr. Nuge

Sep 1 · 2 Mentions

Noah BloombergHOST
38:31
detent.
NugeGUEST
38:32
Like we're still going to be, we're getting all the parts overfit.
NugeGUEST
38:35
Like the lock face is going to be, uh, what's called where I wouldn't say overfit.
NugeGUEST
38:39
I mean, oversized.
NugeGUEST
38:41
So I'm going to be having to grind the lock faces and then tune everything.
JaydenHOST
22:15
If you overfit enough, an LLM is going to look amazing.
JaydenHOST
22:18
Like as if you overfit anything.
JaydenHOST
22:21
Overfit literally anything.
JaydenHOST
22:23
You backtest anything in a particular timeframe and you can put your sharp ratio through the roof.
JaydenHOST
22:29
But with that sort of thing, I always find it hard to believe that those that are sort of like, okay, well cool, we got LLMs trading our portfolio, it's like to go up against Jane Street, like what do you have that Jane Street does not have? What do you have that Citadel does not have? Like this is like the kind of engineering that a Jane Street will have.

Unknown podcast

The Quantum AI Digest: 19 August 2026 Daily Dispatch

Aug 19 · 1 Mention

speaker_0NARRATOR
9:45
Studies reveal that safety alignment in open-weight large language models can be easily bypassed, and so-called decoy hardening is proving unreliable.
speaker_0NARRATOR
9:55
Benchmarkpocalypse This is growing worry, as LLM agents are increasingly found to overfit or game performance benchmarks, leading to exaggerated claims that do not reflect real-world capabilities.
speaker_0NARRATOR
10:08
Workforce trends show that while AI is generating new job categories, games are often concentrated among larger firms.
speaker_0NARRATOR
10:16
Scholars warned that over-automation at the entry level could jeopardize the future talent pipeline, raising the risk of a tragedy of the commons for professional oversight of AI systems.
JJ ZachariasonGUEST
36:15
And so I do things like adding a, uh, you know, a, a program like a strength of schedule factor and, uh, you know, and, and age and a breakout factor and stuff like that.
JJ ZachariasonGUEST
36:25
So I'm testing all of these collegiate metrics, and then eventually I kinda bring them back down to only having a handful because very easy to overfit with modeling.
JJ ZachariasonGUEST
36:35
Overfit meaning you're taking Puka Nacua's rookie season and just associating all of his traits, assuming that future wide receivers should be more like Puka Nacua when the model should be realizing he's just more of an outlier.
Alfredo BrownHOST
36:50
Right.
JJ ZachariasonGUEST
36:50
Right?

Unknown podcast

The Inevitable: Understanding the 12 Technological Forces That Will Shape Our Future

Aug 16 · 1 Mention

speaker_0HOST
14:22
Right.
speaker_0HOST
14:23
Overfitting is a concept from machine learning, and it happens when a model learns the training data too well, to the point that it can't handle any new unexpected information.
speaker_1HOST
14:31
Yeah, it gets stuck.
speaker_0HOST
14:32
It's like having a personal chef who figures out I love pizza.
Tina CarsonHOST
12:21
And understanding why reveals one of the deepest tensions in quantitative finance.
Tina CarsonHOST
12:27
Overfitting occurs when a model captures noise rather than signal.
Tina CarsonHOST
12:30
Think of a student who memorizes answers to specific test questions rather than learning the underlying concepts.
Tina CarsonHOST
12:37
They might ace that particular test, but give them a slightly different set of questions and they fail.

We value your privacy

We use cookies to understand how you use our platform and to improve your experience. Click “Accept All” to consent, or “Decline non-essential” to opt out of non-essential cookies. Read our Privacy Policy.