maximum iteration count (default 100,000)
multiplier for calculating patience (default 1 := Wait lastupdate# * 1 after update)
threshold that iteration is marked as "improved" (default 99.5% = 0.995)
maximum-tolerant loss value. (default 0.0001)
multiplier used for count for validation. (default 1.0)
Validation checked whenever (validationFreq) * (#epoch for 1 training batch).
where #epoch for 1 iteration = round(1 / miniBatchFraction).
size of mini-batch. (default 10)
True if this trainer reuse previous temp data in disk. (default false)
True if want to collect all the data into the driver. (default false)
Persist level (default MEMORY_ONLY)
True if want to collect all the data into the driver.
True if want to collect all the data into the driver. (default false)
threshold that iteration is marked as "improved" (default 99.5% = 0.995)
maximum-tolerant loss value.
maximum-tolerant loss value. (default 0.0001)
maximum iteration count (default 100,000)
size of mini-batch.
size of mini-batch. (default 10)
True if this trainer reuse previous temp data in disk.
True if this trainer reuse previous temp data in disk. (default false)
Persist level (default MEMORY_ONLY)
multiplier used for count for validation.
multiplier used for count for validation. (default 1.0)
Validation checked whenever (validationFreq) * (#epoch for 1 training batch).
where #epoch for 1 iteration = round(1 / miniBatchFraction).
multiplier for calculating patience (default 1 := Wait lastupdate# * 1 after update)
Criteria: When to stop training
This case class defines when to stop training. Training stops if one of the following condition is satisfied.
- #Iteration ≥ maxIter
max(patience, bestIteration * patienceStep)
Validation is done for each
validationFreq
iterations, and whenever current/best loss ratio below improveThreshold, that iteration is marked as best iteration.maximum iteration count
(default 100,000)
multiplier for calculating patience
(default 1 := Wait lastupdate# * 1 after update)
threshold that iteration is marked as "improved"
(default 99.5% = 0.995)
maximum-tolerant loss value.
(default 0.0001)
multiplier used for count for validation.
(default 1.0)
Validation checked whenever (validationFreq) * (#epoch for 1 training batch). where #epoch for 1 iteration = round(1 / miniBatchFraction).size of mini-batch.
(default 10)
True if this trainer reuse previous temp data in disk.
(default false)
True if want to collect all the data into the driver.
(default false)
Persist level
(default MEMORY_ONLY)