3 ms·
It's a matter of ROI: When is it worth it to build something specialized?
by Garlef 1y ago
It's a matter of ROI: When is it worth it to build something specialized?
- ImJasonH 1y agoIs anybody working on making building specialized things easier and cheaper?
- -_- 1y agoYes! At https://RunRL.com https://RunRL.com we offer hosted RL fine-tuning, so all you need to provide is a dataset and reward function or environment.
- selim-now 1y agoyes! check out https://distillabs.ai/ https://distillabs.ai/ – follows a similar approach except the evaluation set is held out before the synthetic data generation, which I would argue makes it more robust (I'm affiliated)
- sigbottle 1y agoWell, one day it might be at the level of shell scripting. I don't think about "the tradeoffs of building a specialized shell script", I just do it because it's cheap and easy and solves a problem right then and there. I don't know how you would even begin to make this kind of same observation for ML models, but seems possible. The 2010s weren't exactly building out "trivial" models, but compared to the architectures and optimizations out now, yeah those models are toy by comparison.