LLM Answer Match

Uses an LLM to check if the generated output answers a question correctly the same way as the expected output, even if their style is different.

POST /langevals/llm_answer_match/evaluate

Try it

Python

Offline Evaluation

import langwatch

df = langwatch.dataset.get_dataset("dataset-id").to_pandas()

evaluation = langwatch.evaluation.init("my-incredible-evaluation")

for index, row in evaluation.loop(df.iterrows()):
    # your execution code here       
    evaluation.run(
        "langevals/llm_answer_match",
        index=index,
        data={
            "input": row["input"],
            "output": output,
            "expected_output": row["expected_output"],
        },
        settings={}
    )

200

[
  {
    "status": "processed",
    "score": 123,
    "passed": true,
    "label": "<string>",
    "details": "<string>",
    "cost": {
      "currency": "<string>",
      "amount": 123
    },
    "raw_response": {},
    "error_type": "<string>",
    "traceback": [
      "<string>"
    ]
  }
]

Authorizations

X-Auth-Token · string · header · required

Body

application/json

data · object · required

Show child attributes

data. input · string

The input text to evaluate

data. output · string

The output text to evaluate

data. expected_output · string

Expected output for comparison

settings · object

Evaluator settings

Show child attributes

settings. model · string · default: openai/gpt-5

The model to use for evaluation

settings. max_tokens · number · default: 8192

Max tokens allowed for evaluation

settings. prompt · string · default: "Verify that the predicted answer matches the gold answer for the question. Style does not matter

Prompt for the comparison

Response

200

application/json

Successful evaluation

status · enum<string> · required

Available options: processed, skipped, error

score · number

Evaluation score

passed · boolean

Whether the evaluation passed

label · string

Evaluation label

details · string

Additional details about the evaluation

cost · object

Show child attributes

cost. currency · string · required

cost. amount · number · required

raw_response · object

Raw response from the evaluator

error_type · string

Type of error if status is 'error'

traceback · string[]

Error traceback if status is 'error'