Missing Figure 2g and evidence of PXDesign-d outperforming hallucination based approaches

Hi team, thanks for releasing PXDesign and the great paper!

I was looking though your preprint and it seems that in section 3.2.2 you mention a figure 2g which seems to be missing.

"Runtime analysis (Figure 2g) shows that PXDesign-d delivers more successful designs
within 24h than any hallucination method, owing to its faster generation speed and higher pass rates."

I would be very interested in this figure and evidence that PXDesign-d has higher pass rates especially. I also am really curious about diversity. From the conclusion, it seems the team ran experiments showing higher structural diversity of PXDesign-d outputs compared to other hallucination approaches.

"Across extensive in silico benchmarks and wet-lab validation on seven biologically diverse protein targets, PXDesign-d demonstrates higher throughput, higher pass rates, and broader structural diversity than hallucination approaches, making it well-suited for large-scale exploratory campaigns."

I believe it would be very helpful for the design community if these figures and experiments were included. Seems like a lot of great work. I would appreciate it if or when these would be released.

Thanks team!