BLIP-3 fine-tuning with LoRA on VQAv2 benchmark; 70.1% accuracy (+9.8pp); HPC PERUN; NVIDIA H200; Bachelor thesis TUKE FEI 2026
machine-learning deep-learning hpc pytorch fine-tuning mutimodal large-language-models low-rank-adaptation blip3
-
Updated
Jul 1, 2026 - Python