{
  "type": "tweet",
  "id": "2115826392633880708",
  "url": "https://x.com/vanstriendaniel/status/2115826392633880708",
  "twitterUrl": "https://twitter.com/vanstriendaniel/status/2115826392633880708",
  "text": "I trained a specialist decision model for curating web-scale data. It's small (380M, fine-tuned from UltraBERT-380M), so it runs relatively cheaply at scale: 1M FineWeb pages took 19 minutes and ~$5 on @huggingface Jobs.\n\nIt thinks a quarter of those pages are product pages. 6% passed a simple \"educational, not spam\" filter, and 1 page in 1,020,000 got \"outstanding\".\n\nI've shared the model, the training data (with the source, labeller, and licence for every row), and the full recipe. \n\nI hope more people share how they prepare data for training open decision models! We've had some cool open model releases. Time for some more data releases! \n\nhttps://t.co/IdHCqnmnP1",
  "source": "",
  "retweetCount": 4,
  "replyCount": 4,
  "likeCount": 25,
  "quoteCount": 0,
  "viewCount": 1263,
  "createdAt": "Thu Oct 29 15:21:36 +0000 2026",
  "lang": "en",
  "bookmarkCount": 15,
  "isReply": false,
  "inReplyToId": null,
  "conversationId": "2115826392633880708",
  "displayTextRange": [
    0,
    278
  ],
  "inReplyToUserId": null,
  "inReplyToUsername": null,
  "author": {
    "type": "user",
    "userName": "vanstriendaniel",
    "url": "https://x.com/vanstriendaniel",
    "twitterUrl": "https://twitter.com/vanstriendaniel",
    "id": "2828117077",
    "name": "Daniel van Strien",
    "isVerified": false,
    "isBlueVerified": true,
    "verifiedType": null,
    "profilePicture": "https://pbs.twimg.com/profile_images/1274680904217251840/N_svCCtg_normal.jpg",
    "coverPicture": "https://pbs.twimg.com/profile_banners/2828117077/1787765117",
    "description": "",
    "location": "Scotland",
    "followers": 6712,
    "following": 1454,
    "status": "",
    "canDm": false,
    "canMediaTag": false,
    "createdAt": "Tue Sep 23 13:43:54 +0000 2014",
    "entities": {
      "description": {
        "urls": []
      },
      "url": {}
    },
    "fastFollowersCount": 0,
    "favouritesCount": 9016,
    "hasCustomTimelines": true,
    "isTranslator": false,
    "mediaCount": 848,
    "statusesCount": 5235,
    "withheldInCountries": [],
    "affiliatesHighlightedLabel": {},
    "possiblySensitive": false,
    "pinnedTweetIds": [
      "2092295830518562868"
    ],
    "profile_bio": {
      "description": "Machine Learning Librarian @huggingface 🤗 \nI like big datasets and small models.",
      "entities": {
        "description": {
          "user_mentions": [
            {
              "id_str": "",
              "indices": [
                27,
                39
              ],
              "name": "",
              "screen_name": "huggingface"
            }
          ]
        },
        "url": {
          "urls": [
            {
              "display_url": "danielvanstrien.xyz",
              "expanded_url": "https://danielvanstrien.xyz/",
              "indices": [
                0,
                23
              ],
              "url": "https://t.co/FnYe1w0XzI"
            }
          ]
        }
      }
    },
    "isAutomated": false,
    "automatedBy": null,
    "protected": false
  },
  "extendedEntities": {
    "media": [
      {
        "allow_download_status": {
          "allow_download": true
        },
        "display_url": "pic.x.com/TcRUVPzcqs",
        "expanded_url": "https://x.com/vanstriendaniel/status/2115826392633880708/photo/1",
        "ext_alt_text": "Diagram: a web page plus typed questions (topic: choice, format: choice, educational: score, spam: yes/no) go into a 380M decision model, which returns a probability for every option of each question.",
        "ext_master_playlist_only": [],
        "ext_media_availability": {
          "status": "Available"
        },
        "ext_playlists": [],
        "features": {
          "large": {
            "faces": []
          },
          "orig": {
            "faces": []
          }
        },
        "id_str": "2115825238499786752",
        "indices": [
          279,
          302
        ],
        "media_key": "3_2115825238499786752",
        "media_results": {
          "id": "QXBpTWVkaWFSZXN1bHRzOgwAAQoAAR1c69gjlgAACgACHVzs5NtWsIQAAA==",
          "result": {
            "__typename": "ApiMedia",
            "id": "QXBpTWVkaWE6DAABCgABHVzr2COWAAAKAAIdXOzk21awhAAA",
            "media_key": "3_2115825238499786752"
          }
        },
        "media_url_https": "https://pbs.twimg.com/media/HVzr2COWAAAulue.png",
        "original_info": {
          "focus_rects": [
            {
              "h": 756,
              "w": 1350,
              "x": 0,
              "y": 0
            },
            {
              "h": 1350,
              "w": 1350,
              "x": 0,
              "y": 0
            },
            {
              "h": 1350,
              "w": 1184,
              "x": 116,
              "y": 0
            },
            {
              "h": 1350,
              "w": 675,
              "x": 371,
              "y": 0
            },
            {
              "h": 1350,
              "w": 1350,
              "x": 0,
              "y": 0
            }
          ],
          "height": 1350,
          "width": 1350
        },
        "sizes": {
          "large": {
            "h": 1350,
            "w": 1350
          }
        },
        "type": "photo",
        "url": "https://t.co/TcRUVPzcqs"
      }
    ]
  },
  "card": null,
  "place": {},
  "entities": {
    "hashtags": [],
    "symbols": [],
    "urls": [
      {
        "display_url": "huggingface.co/davanstrien/ul…",
        "expanded_url": "https://huggingface.co/davanstrien/ultrabert-380m-curation",
        "indices": [
          650,
          673
        ],
        "url": "https://t.co/IdHCqnmnP1"
      }
    ],
    "user_mentions": [
      {
        "id_str": "778764142412984320",
        "indices": [
          202,
          214
        ],
        "name": "Hugging Face",
        "screen_name": "huggingface"
      }
    ]
  },
  "quoted_tweet": null,
  "retweeted_tweet": null,
  "isLimitedReply": false,
  "communityInfo": null,
  "article": null
}