---
title: "✨ Re-evaluate specific tasks from a previous run"
method: POST
path: "/api/inference-runs/re-evaluate/"
tags: ["Prompts"]
---

# ✨ Re-evaluate specific tasks from a previous run

`POST /api/inference-runs/re-evaluate/`

Create a new inference run targeting only specific tasks (re-evaluation).

Supports four modes:
- custom: provide explicit task_ids
- failed: re-run tasks with failed predictions from a source run
- incorrect: re-run tasks with score < 1.0 from a source run
- regressions: re-run tasks that regressed between two runs

## Request body

- ReEvaluateRequest
  - `mode` 'failed' | 'incorrect' | 'regressions' | 'custom', required — * `failed` - failed * `incorrect` - incorrect * `regressions` - regressions * `custom` - custom
  - `model_version_id` integer, required
  - `run_b_id` integer
  - `source_model_run_id` integer
  - `task_ids` integer[]

## Response `201`

- ReEvaluate
  - `mode` 'failed' | 'incorrect' | 'regressions' | 'custom', required — * `failed` - failed * `incorrect` - incorrect * `regressions` - regressions * `custom` - custom
  - `model_version_id` integer, required
  - `run_b_id` integer
  - `source_model_run_id` integer
  - `task_ids` integer[]

---

[API](https://skmtc.net/humansignal/apis/label-studio-api.md) · [All operations](https://skmtc.net/humansignal/apis/label-studio-api/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/humansignal/label-studio-api/revisions/1b113b8df950/schema)