Hybrid Large-Scale Vision Encoder–Boosting Approaches for Predicting Big Five Personality Traits from Social Media Photos: Evidence from Iranian Samples

The experiments for this study, “Hybrid Large-Scale Vision Encoder–Boosting Approaches for Predicting Big Five Personality Traits from Social Media Photos: Evidence from Iranian Samples”, were implemented in a cloud-based environment using Google Colaboratory (Colab) with Jupyter notebooks. The runtime was Python 3.12.12 on Linux 6.6.105+. The CPU consisted of an Intel Xeon @ 2.20 GHz with 2 cores and 2 threads per core, and the system had approximately 12.67 GB of RAM and 113 GB of disk space. For GPU-accelerated experiments, an NVIDIA Tesla T4 GPU with 15 GB of memory and CUDA 12.4 was used. All required packages and dependencies were installed using pip within the Colab environment.The original Instagram photos and associated user IDs are not publicly shared due to confidentiality and ethical considerations. All user IDs were one-way hashed for anonymization, ensuring that the individuals represented in the dataset cannot be re-identified, while preserving the integrity of the data for analysis.