{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreia5poxyxddkfybhzuyxasw6a6ndfkrvfvaa4vjjsbohlaraxqg7ke",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3menqoywriik2"
  },
  "path": "/t/fine-tuning-job-fails-after-3-retries-during-moderation-eval-refusals-v3-internal-error-gpt-4-1-mini-2025-04-14/1374026#post_1",
  "publishedAt": "2026-02-12T08:11:37.000Z",
  "site": "https://community.openai.com",
  "textContent": "Hi,\n\nI’m encountering an issue where a fine-tuning job completes training successfully but ultimately fails during the moderation evaluation phase after three retry attempts due to an internal error.\n\n### Job Information\n\n  * **Job ID:** `ftjob-OkPixpS21QjyQWCsIEJflUJb`\n\n  * **Training Method:** Supervised\n\n  * **Base Model:** `gpt-4.1-mini-2025-04-14`\n\n\n\n\n* * *\n\n### Timeline from Logs(TZ: Asia/Seoul)\n\n\n    15:31:07  Created fine-tuning job\n    15:31:07  Validating training and validation files\n    15:31:14  Files validated, moving job to queued state\n    15:31:17  Fine-tuning job started\n\n    16:18:34  Checkpoint created at step 873\n    16:18:34  Checkpoint created at step 1746\n    16:18:34  New fine-tuned model created\n    16:18:34  Evaluating model against our usage policies\n\n    16:29:37  Retrying moderation eval refusals_v3 (attempt 2/3) due to an internal error.\n    16:59:38  Retrying moderation eval refusals_v3 (attempt 3/3) due to an internal error.\n\n\n\nAfter the third retry attempt, the job status becomes **failed**.\n\n* * *\n\n### Observations\n\n  * The training phase completes successfully.\n\n  * A fine-tuned model is created before the moderation evaluation step.\n\n  * The failure occurs specifically during the `refusals_v3` moderation evaluation.\n\n  * The error message indicates an “internal error”, not a policy violation.\n\n\n\n\n* * *\n\n### Questions\n\n  1. Is this a known issue with moderation evaluation on `gpt-4.1-mini-2025-04-14`?\n\n  2. Does this indicate a problem in my training dataset format or content?\n\n  3. Is there a way to debug or bypass this moderation evaluation failure?\n\n  4. Should I retry the job, or is there a known mitigation?\n\n\n\n\nAny insight would be greatly appreciated.\n\nThank you!",
  "title": "Fine-tuning job fails after 3 retries during moderation eval refusals_v3 (internal error, gpt-4.1-mini-2025-04-14)"
}