Cannot get MRR@10 = 0.35+.
I have ran the experiment twice, one achieved 0.343 and another with 0.335.
I also noticed there is an issue with java version on the first run => this is probably due to compute canada changed their modules. I have modified the sample slurm script accordingly and did the second run. However the second run has a even lower MRR@10.
Cannot get MRR@10 = 0.35+. I have ran the experiment twice, one achieved 0.343 and another with 0.335.
I also noticed there is an issue with java version on the first run => this is probably due to compute canada changed their modules. I have modified the sample slurm script accordingly and did the second run. However the second run has a even lower MRR@10.
Reproduction logs: msmarco-psg-output.log [msmarco-psg-output-2.log](https://github.com/capreolus-ir/capreolus/files/7449055/msmarco-psg-output-2.log
Edit: Moved to this fork https://github.com/crystina-z/capreolus/blob/feature/eval%2Bptmaxp/docs/reproduction/MS_MARCO.md