The accuracy of DP Gen based on DPA-2 pre trained model is 0 for each iteration

[JiangXiaoMingSan]

According to the notebook on Bohrium:DP-Gen based on a DPA-2 pretrained model(https://bohrium.dp.tech/notebooks/62585747598)
When I marked TiTaAlCrFeNiCo according to the notebook, the accuracy of each iteration was 0, and the specific status is as follows
#stage id_stg. iter. accu. cand. fail. lvl_f_lo lvl_f_hi
Stage 0 --------------------
0 0 0 0.0000 1.0000 0.0000 0.0035 0.5000
0 1 1 0.0000 1.0000 0.0000 0.0127 0.5000
0 2 2 0.0000 1.0000 0.0000 0.0122 0.5000
0 3 3 0.0000 1.0000 0.0000 0.0139 0.5000
I reproduced the example of water in the notebook and obtained an iteration accuracy of 0, and the specific status is as follows
#stage id_stg. iter. accu. cand. fail. lvl_f_lo lvl_f_hi
Stage 0 --------------------
0 0 0 0.0000 1.0000 0.0000 0.0056 0.5000
Each iteration has downloadable lable_data, which can be found in the folder called 'prep-run-fp/output/labeled_data'.
But the accuracy of 0 confuses me. Is this a normal phenomenon?

[Jeremy1189]

I tested this on a TiVZrNb system, which shows the same problem, but I think it may not be normal.