ML Take-HomeSequence Prediction System
#RI-6a0e8ead
10 days ago
1
"Build sequence prediction system on ~17M rows of structured data without upfront compute provision"
What Actually Happened
Lack of upfront communication about compute requirements caused unnecessary work time and logistical issues
What I wish I knew
"Upfront communication about compute expectations would have helped prepare in advance"
Advice to Someone Facing This Now
If I could tell you one thing
"Clearly communicate compute requirements and consider providing pre-configured cloud setups or smaller-model baselines"
Regret Score
5/10
Significant downsides.
Complicated feelings about the outcome.
Complicated feelings about the outcome.
ML Take-Home Category Average
?%
of ML Take-Home decisions share a similar regret distribution. (Coming soon)
Impact
-
Financial and Psychosocial impact distributions for this subset.
Advice for Someone Facing This
Short, upvotable wisdom. The most helpful rises to the top.
No advice yet. Be the first to share wisdom.
React, question, or discuss this decision.