Inconsistent learning curve between the suggested configure and presented logs

[voladorlu]

Hi, I tried the suggested config to run the code as follows. However, the evaluation results are different from the results reported in the logs folder. Does it use a different config to generate the logs?

"python main.py --dataset ali --gnn ngcf --dim 64 --lr 0.0001 --batch_size 1024 --gpu_id 0 --context_hops 3 --pool concat --ns mixgcf --K 1 --n_negs 64"

Take the "Recall@20" as an example, I got Recall@20=0.025 at 10th epoch evaluated on the test data, while the uploaded logs show that it can reach to 0.05. Do I miss something important to reproduce the learning curve shown in the logs? -:)

[huangtinglin]

Could you please provide the information about your running environment? Actually, I have just run the code and it can reproduce the performance.