Optimizers API

The Optimizers are a key element in Synalinks, they updates the variables and backpropagate the rewards.

They are in charge of modifying and optimizing the variables (including the prompts) of each individual module composing a synalinks program.

graph LR
A[Training Data] -->|Provide x:DataModel| B[Program];
B -->|Generate y_pred:JsonDataModel| C[Reward];
A -->|Provide y_true:DataModel| C;
C -->|Compute reward:Float| D[Optimizer];
D -->|Update trainable_variable:Variable| B;

This reinforcement loop is what makes possible for the system to learn by repeatedly making predictions and refining its knowledge/methodology in order to maximize the reward.

Optimizers API overview

• Base Optimizer class
• RandomFewShot optimizer
• OPRO optimizer
• FewShotOPRO optimizer