🐦 Twitter Post Details

Viewing enriched Twitter post

@omarsar0

New research from Yann LeCun and collaborators at NYU. It's a really good read for anyone working on efficient Transformer inference. The paper dissects two recurring phenomena in Transformer language models: massive activations (where a few tokens exhibit extreme outlier values, and attention sinks (where certain tokens attract disproportionate attention regardless of semantic relevance). They show the co-occurrence is largely an architectural artifact of pre-norm design, not a fundamental property. Massive activations function as implicit model parameters. Attention sinks modulate outputs locally. Why does it matter? These phenomena directly impact quantization, pruning, and KV-cache management. Understanding their root cause could enable better engineering decisions for efficient inference at scale. Paper: https://t.co/wfzeDpfu4x

Media 1
View on Twitter

📊 Media Metadata

{
  "media": [
    {
      "url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2030403147588604376/media_0.jpg",
      "media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2030403147588604376/media_0.jpg",
      "type": "photo",
      "filename": "media_0.jpg"
    }
  ],
  "processed_at": "2026-03-08T14:26:01.890620",
  "pipeline_version": "2.0"
}

🔧 Raw API Response

{
  "type": "tweet",
  "id": "2030403147588604376",
  "url": "https://x.com/omarsar0/status/2030403147588604376",
  "twitterUrl": "https://twitter.com/omarsar0/status/2030403147588604376",
  "text": "New research from Yann LeCun and collaborators at NYU.\n\nIt's a really good read for anyone working on efficient Transformer inference.\n\nThe paper dissects two recurring phenomena in Transformer language models: massive activations (where a few tokens exhibit extreme outlier values, and attention sinks (where certain tokens attract disproportionate attention regardless of semantic relevance).\n\nThey show the co-occurrence is largely an architectural artifact of pre-norm design, not a fundamental property.\n\nMassive activations function as implicit model parameters. Attention sinks modulate outputs locally.\n\nWhy does it matter?\n\nThese phenomena directly impact quantization, pruning, and KV-cache management.\n\nUnderstanding their root cause could enable better engineering decisions for efficient inference at scale.\n\nPaper: https://t.co/wfzeDpfu4x",
  "source": "Twitter for iPhone",
  "retweetCount": 37,
  "replyCount": 15,
  "likeCount": 243,
  "quoteCount": 1,
  "viewCount": 14896,
  "createdAt": "Sat Mar 07 22:00:08 +0000 2026",
  "lang": "en",
  "bookmarkCount": 253,
  "isReply": false,
  "inReplyToId": null,
  "conversationId": "2030403147588604376",
  "displayTextRange": [
    0,
    274
  ],
  "inReplyToUserId": null,
  "inReplyToUsername": null,
  "author": {
    "type": "user",
    "userName": "omarsar0",
    "url": "https://x.com/omarsar0",
    "twitterUrl": "https://twitter.com/omarsar0",
    "id": "3448284313",
    "name": "elvis",
    "isVerified": false,
    "isBlueVerified": true,
    "verifiedType": null,
    "profilePicture": "https://pbs.twimg.com/profile_images/939313677647282181/vZjFWtAn_normal.jpg",
    "coverPicture": "https://pbs.twimg.com/profile_banners/3448284313/1565974901",
    "description": "",
    "location": "DAIR.AI Academy",
    "followers": 292680,
    "following": 779,
    "status": "",
    "canDm": true,
    "canMediaTag": true,
    "createdAt": "Fri Sep 04 12:59:26 +0000 2015",
    "entities": {
      "description": {
        "urls": []
      },
      "url": {}
    },
    "fastFollowersCount": 0,
    "favouritesCount": 34982,
    "hasCustomTimelines": true,
    "isTranslator": true,
    "mediaCount": 4541,
    "statusesCount": 17422,
    "withheldInCountries": [],
    "affiliatesHighlightedLabel": {},
    "possiblySensitive": false,
    "pinnedTweetIds": [
      "2030403147588604376"
    ],
    "profile_bio": {
      "description": "Building @dair_ai • Prev: Meta AI, Elastic, PhD • New AI learning portal: https://t.co/1e8RZKs4uX",
      "entities": {
        "description": {
          "hashtags": [],
          "symbols": [],
          "urls": [
            {
              "display_url": "academy.dair.ai",
              "expanded_url": "https://academy.dair.ai/",
              "indices": [
                74,
                97
              ],
              "url": "https://t.co/1e8RZKs4uX"
            }
          ],
          "user_mentions": [
            {
              "id_str": "0",
              "indices": [
                9,
                17
              ],
              "name": "",
              "screen_name": "dair_ai"
            }
          ]
        },
        "url": {
          "urls": [
            {
              "display_url": "dair.ai",
              "expanded_url": "https://www.dair.ai/",
              "indices": [
                0,
                23
              ],
              "url": "https://t.co/XQto5ypSIk"
            }
          ]
        }
      }
    },
    "isAutomated": false,
    "automatedBy": null
  },
  "extendedEntities": {
    "media": [
      {
        "display_url": "pic.twitter.com/znwYUDUMU4",
        "expanded_url": "https://twitter.com/omarsar0/status/2030403147588604376/photo/1",
        "ext_media_availability": {
          "status": "Available"
        },
        "features": {
          "large": {
            "faces": []
          },
          "orig": {
            "faces": []
          }
        },
        "id_str": "2030403142828105728",
        "indices": [
          275,
          298
        ],
        "media_key": "3_2030403142828105728",
        "media_results": {
          "id": "QXBpTWVkaWFSZXN1bHRzOgwAAQoAARwtcOafGyAACgACHC1w57rakdgAAA==",
          "result": {
            "__typename": "ApiMedia",
            "id": "QXBpTWVkaWE6DAABCgABHC1w5p8bIAAKAAIcLXDnutqR2AAA",
            "media_key": "3_2030403142828105728"
          }
        },
        "media_url_https": "https://pbs.twimg.com/media/HC1w5p8bIAAK96g.jpg",
        "original_info": {
          "focus_rects": [
            {
              "h": 902,
              "w": 1610,
              "x": 0,
              "y": 0
            },
            {
              "h": 1610,
              "w": 1610,
              "x": 0,
              "y": 0
            },
            {
              "h": 1784,
              "w": 1565,
              "x": 0,
              "y": 0
            },
            {
              "h": 1784,
              "w": 892,
              "x": 133,
              "y": 0
            },
            {
              "h": 1784,
              "w": 1610,
              "x": 0,
              "y": 0
            }
          ],
          "height": 1784,
          "width": 1610
        },
        "sizes": {
          "large": {
            "h": 1784,
            "w": 1610
          }
        },
        "type": "photo",
        "url": "https://t.co/znwYUDUMU4"
      }
    ]
  },
  "card": null,
  "place": {},
  "entities": {
    "hashtags": [],
    "symbols": [],
    "urls": [
      {
        "display_url": "arxiv.org/abs/2603.05498",
        "expanded_url": "https://arxiv.org/abs/2603.05498",
        "indices": [
          829,
          852
        ],
        "url": "https://t.co/wfzeDpfu4x"
      }
    ],
    "user_mentions": []
  },
  "quoted_tweet": null,
  "retweeted_tweet": null,
  "isLimitedReply": false,
  "article": null
}