4 ms·
Fine tuning seems to me to be dangerously close to a new snake oil of AI these days. The narrative goes, "look how awesome ChatGPT is, imagine how good it woul
by zmmmmm 3y ago
Fine tuning seems to me to be dangerously close to a new snake oil of AI these days.
The narrative goes, "look how awesome ChatGPT is, imagine how good it would be trained on just your company's documents".
Which 1000% misses the point. ChatGPT is because (a) it is trained on almost nothing short of the entire corpus of human language ever created. At > 1 trillion parameters, it can have ~1000 parameters for every human on the planet. Let that sink in. And then (b) because it has been subjected to an unknown but likely massive amount of human reinforcement feedback.
The idea that you can meaningfully impact the output of the model towards factual accuracy or logical correctness just by doing a small amount of fully automated training using a tiny corpus of company documents is seductive, but super far from robustly demonstrated as far as I'm aware. Yet this is the pitch being sold very often.
- janalsncm 3y agoWhat? Fine tuning has been a common technique for years now. Fine tuned BERT models were behind a lot of retrieval-based systems and they work well. A more recent example is stable diffusion fine tuned on specific subjects. Whether fine tuning can reduce hallucination is first of all a question which only pertains to decoders and which is highly dependent on how the model has been fine tuned.