Human Preferences: Segment Returns
Beginner
The problem, from first principles
The reward model scores whole segments of behaviour, not single steps. A segment's return is the sum of the per-step predictions, and comparisons are made between segment returns.
From theory to code
Implement segment_return(rewards), returning the sum of the predicted rewards.
Constraints
- Returns a Python float.
Hints
Hint 1
Sum the array.
Keywords
research-papersreinforcement-learningalignmentreward-learning
Python
Run your code to see results
No results yet
Run your code to check it against the examples, or submit to run every test.