search

terry-r123 / RNABenchmark

[NeurIPS 2024] BEACON: Benchmark for Comprehensive RNA Tasks and Language Models
Apache License 2.0
21 stars 1 forks source link

how to reproduce Table3's work #5

Open sherlockma11 opened 3 weeks ago

sherlockma11 commented 3 weeks ago

Thank you for bringing in such great work. I'm trying to understand how to reproduce Table3's work, I don't seem to see any relevant code. for example, if I want to reproduce MRL's R^2, which code do I use when I'm done training? Or I have to write one myself? Thanks!

sherlockma11 commented 3 weeks ago

Another Question, Which part of dataset is this table running, is just Val datasets?

terry-r123 commented 1 week ago

@sherlockma11 Thank you for your question.

  1. For the scripts of other models running other tasks, we will sort out the relevant scripts and update them afterwards. For now, you can refer to the script for ncRNA, and you can change the relevant parameters such as data path to MRL tasks.

  2. The table shows the results of test datasets. For the MRL task, the test dataset corresponds to Human setting and the val dataset corresponds to Random setting.