
MosaicML: Deep learning models for sale, all shapes and sizes
The Stack Overflow Podcast
00:00
The 7B MPT Model Is a Demo Track
We train a lot of models for contract, which means many of them don't see the light of day. So how do people know that we can actually train good models? Mm hmm. The 7B, you know, I'll say is the baby of the family. It definitely has some bigger batter siblings that are available for our customers but really it's a demo. A lot of folks do want to train from scratch. They want to have complete control over the pre-training data. And they're built for heavy duty fine tuning. We chose to use alibi in such a way that basically you can use as long of a context as you can fit on the GPU.
Transcript
Play full episode