Skip to main content
weave / ScoreLogger ScoreLogger manages scoring for a single prediction. Returned from EvaluationLogger.logPrediction(). Example 
const pred = await ev.logPrediction(example, output);
await pred.logScore("accuracy", 0.95);
await pred.logScore("relevance", 0.8);
await pred.finish(); // Finalizes the prediction

​
Table of contents

​
Constructors

​
Methods

​
Constructors

​
constructor

• new ScoreLogger(predMeta, evalLogger): ScoreLogger

​
Parameters

NameType
predMetaPredictAndScoreCallMetadata
evalLoggerEvaluationLogger

​
Returns

ScoreLogger

​
Defined in

evaluationLogger.ts:317

​
Methods

​
finish

▸ finish(): Promise<void> Finish the scoring process for the prediction. Finalizes the predict_and_score call with accumulated scores. Updates incremental aggregates and frees memory.

​
Returns

Promise<void>

​
Defined in

evaluationLogger.ts:410

​
logScore

▸ logScore(scorerName, score): Promise<void> Log a score for this prediction. Creates a scorer call as a child of predict_and_score.

​
Parameters

NameTypeDescription
scorerNamestringName of the scorer (e.g., “accuracy”, “f1_score”)
scoreanyThe score value

​
Returns

Promise<void>

​
Defined in

evaluationLogger.ts:332
⌘I