3 ms·
The issue with all of the chat-optimised LLMs is that they can’t be incrementally updated. After the base training there are three seperate sets of additional
by jiggawatts 3y ago
The issue with all of the chat-optimised LLMs is that they can’t be incrementally updated.
After the base training there are three seperate sets of additional training to align the model, convince it to do question-response, and to improve the quality via feedback.
If you update the original base model, then all the tuning steps need to be repeated.
For a model the size of GPT-4 this is expensive and slow, which is why OpenAI hasn’t bothered.
- lhl 3y agoWhy would you only update the original base model and not simply continue fine-tuning? (This is a rhetorical question of course, and there are multiple empirical ways to confirm the majority of your answer is "hallucinated," not least of which is the fact that OpenAI keeps rolling out new versions and variants with different capabilities.)