Answer Correctness

Exclusive to enterprise customers. Contact us to activate this feature.

Objective: The Answer Correctness Test checks whether the provided answer is correct based on the given context. This test ensures that the response accurately reflects the information found in the context. A higher score indicates a higher degree of correctness in the answer.

Required Arguments:

  • List of Chunks of Context: Sections of the context that contain the correct information against which the answer will be evaluated.

  • Response Received: The answer generated by the model that is being evaluated for correctness.

  • Threshold for Similarity Index: A parameter that determines the level of similarity at which the answer is considered correct.

Score Range: 0 (incorrect answer) to 1 (correct answer)

Interpretation: A higher score suggests that the answer closely matches the correct information provided in the context, indicating a higher level of accuracy. A lower score would indicate that the answer is not aligned with the context and may be incorrect.

Last updated