strategy
EvaluationStrategy ¶
Bases: ABC
Abstract strategy for different evaluation modes
Source code in src/recnexteval/evaluators/stream/strategy.py
6 7 8 9 10 11 12 | |
should_advance_window(algo_state_mgr, current_step, total_steps) abstractmethod ¶
Determine if should advance to next window
Source code in src/recnexteval/evaluators/stream/strategy.py
9 10 11 12 | |
SlidingWindowStrategy ¶
Bases: EvaluationStrategy
Strategy for sliding window evaluation
Source code in src/recnexteval/evaluators/stream/strategy.py
15 16 17 18 19 20 21 22 23 24 | |
should_advance_window(algo_state_mgr, current_step, total_steps) ¶
Advance only when all algorithms predicted
Source code in src/recnexteval/evaluators/stream/strategy.py
18 19 20 21 22 23 24 | |
SingleTimePointStrategy ¶
Bases: EvaluationStrategy
Strategy for sliding window evaluation
Source code in src/recnexteval/evaluators/stream/strategy.py
27 28 29 30 31 32 | |
should_advance_window(algo_state_mgr, current_step, total_steps) ¶
Advance only when all algorithms predicted
Source code in src/recnexteval/evaluators/stream/strategy.py
30 31 32 | |