FPGA-based acceleration of TinyLLaVA-Phi-2-SigLIP-3.1B inference on AMD Alveo U280 using Vitis HLS.
-
Updated
Nov 11, 2026 - HTML
FPGA-based acceleration of TinyLLaVA-Phi-2-SigLIP-3.1B inference on AMD Alveo U280 using Vitis HLS.
Scripts, and links to the trained models, from our paper: N. Gkalelis, V. Mezaris, "LLaVA-CKD: Bottom-Up Cascaded Knowledge Distillation for Vision-Language Models", IEEE ACCESS, 2026.
To associate your repository with the tinyllava topic, visit your repo's landing page and select "manage topics."