Machine Learning Hyperparameters (original) (raw)

Rami Ismael

(March 11, 2025)

1 Introduction

This document presents the hyperparameters used in our machine learning model.

2 Hyperparameters

Table 1: Key Hyperparameters for Machine Learning Model

Parameter Symbol Value
Reward Scale Rssubscriptš‘…š‘ R_{s}italic_R start_POSTSUBSCRIPT italic_s end_POSTSUBSCRIPT 4
use_screen_explore Usā¢esubscriptš‘ˆš‘ š‘’U_{se}italic_U start_POSTSUBSCRIPT italic_s italic_e end_POSTSUBSCRIPT True
Explore Weight Ewsubscriptšøš‘¤E_{w}italic_E start_POSTSUBSCRIPT italic_w end_POSTSUBSCRIPT 3
Gamma Ī³š›¾\gammaitalic_Ī³ 0.998
Lambda Ī»šœ†\lambdaitalic_Ī» 0.95
n_step nssubscriptš‘›š‘ n_{s}italic_n start_POSTSUBSCRIPT italic_s end_POSTSUBSCRIPT 1024
num_env nesubscriptš‘›š‘’n_{e}italic_n start_POSTSUBSCRIPT italic_e end_POSTSUBSCRIPT 32
Value Function Coefficient Vfā¢csubscriptš‘‰š‘“š‘V_{fc}italic_V start_POSTSUBSCRIPT italic_f italic_c end_POSTSUBSCRIPT 0.5
Advantage Norm Ansubscriptš“š‘›A_{n}italic_A start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT minibatch-wise
Minibatch Size Mssubscriptš‘€š‘ M_{s}italic_M start_POSTSUBSCRIPT italic_s end_POSTSUBSCRIPT 4096
Clip Range Crsubscriptš¶š‘ŸC_{r}italic_C start_POSTSUBSCRIPT italic_r end_POSTSUBSCRIPT 0.2
Learning Rate Ī±š›¼\alphaitalic_Ī± 3.0e-4
Entropy Coefficient Ecsubscriptšøš‘E_{c}italic_E start_POSTSUBSCRIPT italic_c end_POSTSUBSCRIPT 0
Epochs EšøEitalic_E 3

3 Conclusion

The table above summarizes the key hyperparameters used in our experiments.