ML@Purdue x DagsHub Hackathon

ML@Purdue Logo DagsHub
days
hours
minutes
seconds

About

The ML@Purdue x DagsHub Hackathon is our club's first ever corporate event! Join us for three weeks of hacking new state-of-the-art solutions, MLOps workshops and replication studies guided by ML@Purdue and DagsHub support teams. Best teams win swag and cash prizes! We have two competitive tracks:

Guided

Build a state-of-the-art Machine Learning Model! Participate in a graph generation challenge with multimodal data! Obtain the best possible performance on our event leaderboard, and the three best-performing models on unseen data win prizes!

Advanced

Reimplement a state-of-the-art machine learning paper, run ablation studies, make it accessible to the entire machine learning community and publish it as a paper under your name! The three most successful projects also win prizes!


Specifics regarding delivarables are listed below!

Details

Prizes

Guided

  • 1st Place: $75 + SWAG
  • 2nd Place: SWAG
  • 3rd Place: SWAG

Advanced

  • 1st Place: $150 + SWAG
  • 2nd Place: $100 + SWAG
  • 3rd Place: SWAG

Locations & Timeline

October 14 (CL50 224)

Optional Team Finding Event @ 4pm

Kick-off Ceremony & Workshops @ 5-8pm

October 21 (BRNG 2280)

Data Engine & Office Hours @ 5-8pm

October 28 (BRNG 2280)

Competition Midnight & Office Hours @ 5-8pm

November 4 (BRNG 2280)

Submission Date & Office Hours @ 5-8pm

November 11 (CL50 224)

Closing Ceremony & Winners @ 5-8pm


Competition Tracks

Guided

  • Participate in a graph generation challenge!
  • Template for an end-to-end workflow with baseline performance will be provided.
  • Log data, experiments and notebooks to your DagsHub repository.
  • Model performance will be evaluated on a hidden test set.
  • Top 3 highest scores win prizes!

Advanced

  • Get to publish a paper under the Machine Learning Reproducibility Challenge!
  • Reproduce novel research from our approved list of papers.
  • Conduct ablation studies and write a reproducibility report documenting your findings
  • Your reproducibility report can get published to ReScience (platinum open-access peer-reviewed journal)!
  • 3 best papers win prizes!