T180
A benchmark dataset in the field of crystal structure prediction containing 180 target structures, used to evaluate the element substitution and candidate ranking performance of template-based CSP methods.
Timeline (last 90 days)
Of the 2347 structures selected by the T180 benchmark, 469 (20%) had incorrect compositions in TCSP2.0.
Ablation experiments show that most of TCSP3.0's performance improvement on T180 and T800 comes from the correctness fixes (coefficient-preserving mapping and synchronous replacement).
On the T180 benchmark, TCSP3.0 achieves Top-1 and Top-20 match rates of 72.2% and 90.6%, respectively, whereas TCSP2.0 reaches only 28.3% and 45.6%.
On T180, TCSP3.0's Top-1 match rate (72.2%) outperforms PNcsp+ and the deep generative models CrystalFlow and MaskGXT (at most 63.9%).
All Facts (4)
Of the 2347 structures selected by the T180 benchmark, 469 (20%) had incorrect compositions in TCSP2.0.
50%UnverifiedAblation experiments show that most of TCSP3.0's performance improvement on T180 and T800 comes from the correctness fixes (coefficient-preserving mapping and synchronous replacement).
50%UnverifiedOn the T180 benchmark, TCSP3.0 achieves Top-1 and Top-20 match rates of 72.2% and 90.6%, respectively, whereas TCSP2.0 reaches only 28.3% and 45.6%.
50%UnverifiedOn T180, TCSP3.0's Top-1 match rate (72.2%) outperforms PNcsp+ and the deep generative models CrystalFlow and MaskGXT (at most 63.9%).
50%