Methods for conditioning
With the advent of large pre-trained language models like GPT, there has been growing interest in techniques to adapt these models for downstream tasks. As LLMs continue to develop, they will become even more effective and useful for a broader range of applications, and we can expect future advancements in fine-tuning and prompting techniques to help go even further in complex tasks that involve reasoning and tool use.
Several approaches have been proposed for conditioning. Here is a table summarizing the different techniques:
|
Stage |
Technique |
Examples |
|
Training |
Data curation |
Training on diverse data |
|
Objective function |
Careful design of training... |
