API ReferenceBenchmarks
Cells
Get cell-level comparison data from a benchmark
GET
/api/v1/benchmarks/{executionId}/cellsReturns cell-level comparison data between extracted values and ground truth. Useful for identifying specific extraction errors.
Parameters
| Name | Type | In | Required | Description |
|---|---|---|---|---|
executionId | string | path | Yes | The benchmark execution ID |
columnId | string | query | No | Filter by column ID |
fileId | string | query | No | Filter by file ID |
onlyMismatches | boolean | query | No | Only return cells where extraction differs from ground truth(default: false) |
limit | integer | query | No | Results per page (1-500)(default: 100) |
offset | integer | query | No | Number of results to skip(default: 0) |
Request
const res = await fetch(
"https://app.limai.io/api/v1/benchmarks/exec_abc123/cells?onlyMismatches=true&limit=50",
{
headers: {
Authorization: "Bearer YOUR_API_TOKEN",
},
}
);
const { data, pagination, summary } = await res.json();Response
Response200
{
"deploymentId": "dep_001",
"data": [
{
"fileId": "file_001",
"fileName": "invoice_001.pdf",
"tableId": "tbl_001",
"tableName": "Invoices",
"columnId": "col_001",
"columnName": "Date",
"extractedValue": "01/15/2025",
"groundTruthValue": "2025-01-15",
"successRate": 0,
"measurementType": "EXACT_MATCH",
"isMismatch": true
}
],
"pagination": {
"total": 5,
"limit": 50,
"offset": 0
},
"summary": {
"totalCells": 150,
"mismatches": 5,
"mismatchRate": 0.033
}
}