🐦 Twitter Post Details

Viewing enriched Twitter post

@SergioPaniego

Earlier this month, Apple introduced Simple Self-Distillation: a fine-tuning method that improves models on coding tasks just by sampling from the model and training on its own outputs with plain cross-entropy and… it's already supported in TRL, built by @krasul. you can really feel the pace of development in the team 🐎 paper by @onloglogn, @richard_baihe, @UnderGroundJeg, Navdeep Jaitly, @trebolloc, @YizheZhangNLP at Apple 🍎 how it works: the model generates completions at a training-time temperature (T_train) with top_k/top_p truncation, then fine-tunes on them with plain cross-entropy. no labels or verifier needed you can try it right away with this ready-to-run example (Qwen3-4B on rStar-Coder): https://t.co/zizfISD6bq or benchmark a checkpoint with the eval script: https://t.co/mKlafTyKSe one neat insight from the paper: T_train and T_eval compose into an effective T_eff = T_train × T_eval, so a broad band of configs works well. even very noisy samples still help want to dig deeper? paper: https://t.co/aj1ZAcr8Mw trainer docs: https://t.co/TNVz93kZi9

Media 1
Media 2
Media 3
Media 4
Media 5

📊 Media Metadata

{
  "media": [
    {
      "url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2044424654287167544/media_0.jpg",
      "media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2044424654287167544/media_0.jpg",
      "type": "photo",
      "filename": "media_0.jpg"
    },
    {
      "url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2044424654287167544/media_1.jpg",
      "media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2044424654287167544/media_1.jpg",
      "type": "photo",
      "filename": "media_1.jpg"
    },
    {
      "url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2044424654287167544/media_2.jpg",
      "media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2044424654287167544/media_2.jpg",
      "type": "photo",
      "filename": "media_2.jpg"
    },
    {
      "url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2044424654287167544/media_3.jpg",
      "media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2044424654287167544/media_3.jpg",
      "type": "photo",
      "filename": "media_3.jpg"
    },
    {
      "url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2044424654287167544/media_4.jpg",
      "media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2044424654287167544/media_4.jpg",
      "type": "photo",
      "filename": "media_4.jpg"
    }
  ],
  "processed_at": "2026-04-15T17:50:33.677506",
  "pipeline_version": "2.0"
}

🔧 Raw API Response

{
  "type": "tweet",
  "id": "2044424654287167544",
  "url": "https://x.com/SergioPaniego/status/2044424654287167544",
  "twitterUrl": "https://twitter.com/SergioPaniego/status/2044424654287167544",
  "text": "Earlier this month, Apple introduced Simple Self-Distillation: a fine-tuning method that improves models on coding tasks just by sampling from the model and training on its own outputs with plain cross-entropy\n\nand… it's already supported in TRL, built by @krasul. you can really feel the pace of development in the team 🐎\n\npaper by @onloglogn, @richard_baihe,\n@UnderGroundJeg, Navdeep Jaitly, @trebolloc, @YizheZhangNLP\nat Apple 🍎\n\nhow it works: the model generates completions at a training-time temperature (T_train) with top_k/top_p truncation, then fine-tunes on them with plain cross-entropy. no labels or verifier needed\n\nyou can try it right away with this ready-to-run example (Qwen3-4B on rStar-Coder):\nhttps://t.co/zizfISD6bq\nor benchmark a checkpoint with the eval script:\nhttps://t.co/mKlafTyKSe\n\none neat insight from the paper: T_train and T_eval compose into an effective T_eff = T_train × T_eval, so a broad band of configs works well. even very noisy samples still help\n\nwant to dig deeper?\n\npaper: https://t.co/aj1ZAcr8Mw\ntrainer docs: https://t.co/TNVz93kZi9",
  "source": "Twitter for iPhone",
  "retweetCount": 17,
  "replyCount": 2,
  "likeCount": 90,
  "quoteCount": 1,
  "viewCount": 9045,
  "createdAt": "Wed Apr 15 14:36:35 +0000 2026",
  "lang": "en",
  "bookmarkCount": 73,
  "isReply": false,
  "inReplyToId": null,
  "conversationId": "2044424654287167544",
  "displayTextRange": [
    0,
    279
  ],
  "inReplyToUserId": null,
  "inReplyToUsername": null,
  "author": {
    "type": "user",
    "userName": "SergioPaniego",
    "url": "https://x.com/SergioPaniego",
    "twitterUrl": "https://twitter.com/SergioPaniego",
    "id": "330572456",
    "name": "Sergio Paniego",
    "isVerified": false,
    "isBlueVerified": true,
    "verifiedType": null,
    "profilePicture": "https://pbs.twimg.com/profile_images/1697156955603673088/uoaqJ13T_normal.jpg",
    "coverPicture": "https://pbs.twimg.com/profile_banners/330572456/1598456703",
    "description": "",
    "location": "Madrid, España",
    "followers": 3544,
    "following": 1662,
    "status": "",
    "canDm": true,
    "canMediaTag": false,
    "createdAt": "Wed Jul 06 20:36:22 +0000 2011",
    "entities": {
      "description": {
        "urls": []
      },
      "url": {}
    },
    "fastFollowersCount": 0,
    "favouritesCount": 13639,
    "hasCustomTimelines": true,
    "isTranslator": false,
    "mediaCount": 595,
    "statusesCount": 5149,
    "withheldInCountries": [],
    "affiliatesHighlightedLabel": {},
    "possiblySensitive": false,
    "pinnedTweetIds": [],
    "profile_bio": {
      "description": "Machine Learning Engineer @huggingface 🤗\nAI PhD. Technology enables us to be more human. 🏳️‍🌈",
      "entities": {
        "description": {
          "hashtags": [],
          "symbols": [],
          "urls": [],
          "user_mentions": [
            {
              "id_str": "0",
              "indices": [
                26,
                38
              ],
              "name": "",
              "screen_name": "huggingface"
            }
          ]
        },
        "url": {
          "urls": [
            {
              "display_url": "sergiopaniego.github.io",
              "expanded_url": "https://sergiopaniego.github.io/",
              "indices": [
                0,
                23
              ],
              "url": "https://t.co/0cLnQcCqKl"
            }
          ]
        }
      }
    },
    "isAutomated": false,
    "automatedBy": null
  },
  "extendedEntities": {
    "media": [
      {
        "allow_download_status": {
          "allow_download": true
        },
        "display_url": "pic.twitter.com/HLbOuxnIR4",
        "expanded_url": "https://twitter.com/SergioPaniego/status/2044424654287167544/photo/1",
        "ext_media_availability": {
          "status": "Available"
        },
        "features": {
          "large": {
            "faces": []
          },
          "orig": {
            "faces": []
          }
        },
        "id_str": "2044423178626256898",
        "indices": [
          280,
          303
        ],
        "media_key": "3_2044423178626256898",
        "media_results": {
          "id": "QXBpTWVkaWFSZXN1bHRzOgwAAQoAARxfQAyUl9ACCgACHF9BZCjWYDgAAA==",
          "result": {
            "__typename": "ApiMedia",
            "id": "QXBpTWVkaWE6DAABCgABHF9ADJSX0AIKAAIcX0FkKNZgOAAA",
            "media_key": "3_2044423178626256898"
          }
        },
        "media_url_https": "https://pbs.twimg.com/media/HF9ADJSX0AIRB3s.jpg",
        "original_info": {
          "focus_rects": [
            {
              "h": 1075,
              "w": 1920,
              "x": 0,
              "y": 0
            },
            {
              "h": 1080,
              "w": 1080,
              "x": 0,
              "y": 0
            },
            {
              "h": 1080,
              "w": 947,
              "x": 0,
              "y": 0
            },
            {
              "h": 1080,
              "w": 540,
              "x": 66,
              "y": 0
            },
            {
              "h": 1080,
              "w": 1920,
              "x": 0,
              "y": 0
            }
          ],
          "height": 1080,
          "width": 1920
        },
        "sizes": {
          "large": {
            "h": 1080,
            "w": 1920
          }
        },
        "type": "photo",
        "url": "https://t.co/HLbOuxnIR4"
      }
    ]
  },
  "card": null,
  "place": {},
  "entities": {
    "hashtags": [],
    "symbols": [],
    "urls": [
      {
        "display_url": "github.com/huggingface/tr…",
        "expanded_url": "https://github.com/huggingface/trl/blob/main/trl/experimental/ssd/ssd.py",
        "indices": [
          713,
          736
        ],
        "url": "https://t.co/zizfISD6bq"
      },
      {
        "display_url": "github.com/huggingface/tr…",
        "expanded_url": "https://github.com/huggingface/trl/blob/main/trl/experimental/ssd/ssd_eval.py",
        "indices": [
          785,
          808
        ],
        "url": "https://t.co/mKlafTyKSe"
      },
      {
        "display_url": "huggingface.co/papers/2604.01…",
        "expanded_url": "https://huggingface.co/papers/2604.01193",
        "indices": [
          1017,
          1040
        ],
        "url": "https://t.co/aj1ZAcr8Mw"
      },
      {
        "display_url": "huggingface.co/docs/trl/main/…",
        "expanded_url": "https://huggingface.co/docs/trl/main/en/ssd_trainer",
        "indices": [
          1055,
          1078
        ],
        "url": "https://t.co/TNVz93kZi9"
      }
    ],
    "user_mentions": [
      {
        "id_str": "8414042",
        "indices": [
          256,
          263
        ],
        "name": "Kashif Rasul",
        "screen_name": "krasul"
      },
      {
        "id_str": "877868973026455553",
        "indices": [
          333,
          343
        ],
        "name": "Ruixiang Zhang",
        "screen_name": "onloglogn"
      },
      {
        "id_str": "1102455266082082816",
        "indices": [
          345,
          359
        ],
        "name": "Richard He Bai",
        "screen_name": "richard_baihe"
      },
      {
        "id_str": "852388519410622464",
        "indices": [
          361,
          376
        ],
        "name": "Huangjie Zheng",
        "screen_name": "UnderGroundJeg"
      },
      {
        "id_str": "2027062806936535041",
        "indices": [
          394,
          404
        ],
        "name": "Ronan Collobert",
        "screen_name": "trebolloc"
      },
      {
        "id_str": "545985637",
        "indices": [
          406,
          420
        ],
        "name": "Yizhe Zhang @ ICLR 2026",
        "screen_name": "YizheZhangNLP"
      }
    ]
  },
  "quoted_tweet": null,
  "retweeted_tweet": null,
  "isLimitedReply": false,
  "communityInfo": null,
  "article": null
}