← All IntelClip / AI Agents
The model capability gap is a data gap
From Emulated: The Data for Fully Autonomous Software Engineers and Companies — Joseph Wang · ≈2:26
“Models typically are only as good at as data is.”
“Model capability has never uh regressed whenever you introduce more high-quality data.”
What’s in it
- Frames model capability gaps as fundamentally data gaps
- Argues quality data additions never hurt model performance
- Sets up a look at what today's data landscape actually looks like
Clip transcript
>> So, like with everything in NML, uh the gap in models is usually a gap in data. Models typically are only as good at as data is. Um and to really highlight this point, right? Model capability has never uh regressed whenever you introduce more high-quality data. Um so, with that being said, what is the data gap then? What does data look like right now? And how is this influencing the model capability gap here?
Recommended reads
Comments
Sign in to comment.
Loading comments…