Viewing enriched Twitter post
Can regularization based JEPA (e.g. SIGReg) scale and compete with SOTA foundation models (DINO)? Here is the answer: yes and with 10x less data. VISReg (slight variation of SIGReg) competes with DINOv2-LVD142M while only training on inet22k. Try it out: https://t.co/vBhrNAmFq6 https://t.co/XERFZEAE8t
{
"media": [
{
"url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2072088917348630648/media_0.jpg",
"media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2072088917348630648/media_0.jpg",
"type": "photo",
"filename": "media_0.jpg"
},
{
"url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2072088917348630648/media_1.jpg",
"media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2072088917348630648/media_1.jpg",
"type": "photo",
"filename": "media_1.jpg"
}
],
"processed_at": "2026-06-30T23:03:05.803849",
"pipeline_version": "2.0"
}{
"type": "tweet",
"id": "2072088917348630648",
"url": "https://x.com/randall_balestr/status/2072088917348630648",
"twitterUrl": "https://twitter.com/randall_balestr/status/2072088917348630648",
"text": "Can regularization based JEPA (e.g. SIGReg) scale and compete with SOTA foundation models (DINO)? Here is the answer: yes and with 10x less data.\nVISReg (slight variation of SIGReg) competes with DINOv2-LVD142M while only training on inet22k.\nTry it out: https://t.co/vBhrNAmFq6 https://t.co/XERFZEAE8t",
"source": "Twitter for iPhone",
"retweetCount": 2,
"replyCount": 0,
"likeCount": 6,
"quoteCount": 0,
"viewCount": 229,
"createdAt": "Tue Jun 30 22:44:30 +0000 2026",
"lang": "en",
"bookmarkCount": 2,
"isReply": false,
"inReplyToId": null,
"conversationId": "2072088917348630648",
"displayTextRange": [
0,
278
],
"inReplyToUserId": null,
"inReplyToUsername": null,
"author": {
"type": "user",
"userName": "randall_balestr",
"url": "https://x.com/randall_balestr",
"twitterUrl": "https://twitter.com/randall_balestr",
"id": "1246070462679040000",
"name": "Randall Balestriero",
"isVerified": false,
"isBlueVerified": true,
"verifiedType": null,
"profilePicture": "https://pbs.twimg.com/profile_images/1327365059685871619/4VtwVXV2_normal.jpg",
"coverPicture": "https://pbs.twimg.com/profile_banners/1246070462679040000/1723917931",
"description": "",
"location": "USA",
"followers": 6209,
"following": 237,
"status": "",
"canDm": true,
"canMediaTag": true,
"createdAt": "Fri Apr 03 13:46:59 +0000 2020",
"entities": {
"description": {
"urls": []
},
"url": {}
},
"fastFollowersCount": 0,
"favouritesCount": 487,
"hasCustomTimelines": true,
"isTranslator": false,
"mediaCount": 214,
"statusesCount": 695,
"withheldInCountries": [],
"affiliatesHighlightedLabel": {},
"possiblySensitive": false,
"pinnedTweetIds": [
"2060026304209022984"
],
"profile_bio": {
"description": "AI Researcher: From theory to practice (and back)\nPostdoc @MetaAI with @ylecun\nPhD @RiceUniversity with @rbaraniuk\nMasters @ENS_Ulm @Paris_Sorbonne",
"entities": {
"description": {
"user_mentions": [
{
"id_str": "",
"indices": [
58,
65
],
"name": "",
"screen_name": "MetaAI"
},
{
"id_str": "",
"indices": [
71,
78
],
"name": "",
"screen_name": "ylecun"
},
{
"id_str": "",
"indices": [
83,
98
],
"name": "",
"screen_name": "RiceUniversity"
},
{
"id_str": "",
"indices": [
104,
114
],
"name": "",
"screen_name": "rbaraniuk"
},
{
"id_str": "",
"indices": [
123,
131
],
"name": "",
"screen_name": "ENS_Ulm"
},
{
"id_str": "",
"indices": [
132,
147
],
"name": "",
"screen_name": "Paris_Sorbonne"
}
]
},
"url": {
"urls": [
{
"display_url": "randallbalestriero.github.io",
"expanded_url": "https://randallbalestriero.github.io",
"indices": [
0,
23
],
"url": "https://t.co/vb5e6gCvqF"
}
]
}
}
},
"isAutomated": false,
"automatedBy": null
},
"extendedEntities": {
"media": [
{
"allow_download_status": {
"allow_download": true
},
"display_url": "pic.twitter.com/XERFZEAE8t",
"expanded_url": "https://twitter.com/randall_balestr/status/2072088917348630648/photo/1",
"ext_master_playlist_only": [],
"ext_media_availability": {
"status": "Available"
},
"ext_playlists": [],
"features": {
"large": {
"faces": []
},
"orig": {
"faces": []
}
},
"id_str": "2072088408730599424",
"indices": [
279,
302
],
"media_key": "3_2072088408730599424",
"media_results": {
"id": "QXBpTWVkaWFSZXN1bHRzOgwAAQoAARzBiW3pl4AACgACHMGJ5FWWwHgAAA==",
"result": {
"__typename": "ApiMedia",
"id": "QXBpTWVkaWE6DAABCgABHMGJbemXgAAKAAIcwYnkVZbAeAAA",
"media_key": "3_2072088408730599424"
}
},
"media_url_https": "https://pbs.twimg.com/media/HMGJbemXgAARWrX.png",
"original_info": {
"focus_rects": [
{
"h": 158,
"w": 282,
"x": 0,
"y": 0
},
{
"h": 158,
"w": 158,
"x": 0,
"y": 0
},
{
"h": 158,
"w": 139,
"x": 0,
"y": 0
},
{
"h": 158,
"w": 79,
"x": 26,
"y": 0
},
{
"h": 158,
"w": 374,
"x": 0,
"y": 0
}
],
"height": 158,
"width": 374
},
"sizes": {
"large": {
"h": 158,
"w": 374
}
},
"type": "photo",
"url": "https://t.co/XERFZEAE8t"
}
]
},
"card": null,
"place": {},
"entities": {
"urls": [
{
"display_url": "huggingface.co/BooBooWu/visreg",
"expanded_url": "https://huggingface.co/BooBooWu/visreg",
"indices": [
255,
278
],
"url": "https://t.co/vBhrNAmFq6"
}
]
},
"quoted_tweet": {
"type": "tweet",
"id": "2070866534462087626",
"url": "https://x.com/HaiyuWu1/status/2070866534462087626",
"twitterUrl": "https://twitter.com/HaiyuWu1/status/2070866534462087626",
"text": "Working on world model or SSL? You definitely need to try our new work: VISReg!\n\nWhat does it achieve?\n💪 Strong collapse prevention: High gradient when embedding collapse\n⚡ Friendly to scale training: Linear complexity to scaling factors\n🧩 Easy to train: Similar to LeJEPA, it is a heuristic-free method\n🏆 Best OOD performance: Achieving the best accuracy on 6 OOD datasets\n📉 Data efficiency: Achieving a similar OOD average accuracy to DINOv2 with 90% less data\n🧬 Robust to low-quality datasets: It is robust to long-tailed and sparse datasets\n\nOur results also indicate that SIGReg type methods can scale up, filling in the missing piece in @ylecun's great talk https://t.co/P9TXmk3fFa.\n\nA big thanks to my co-author @randall_balestr and my manager @DrMorganLevine. Also, huge gratitude to @ylecun for connecting us to make this project happen! 🤝\n\n#SelfSupervisedLearning #JEPA #WorldModel",
"source": "Twitter for iPhone",
"retweetCount": 3,
"replyCount": 3,
"likeCount": 30,
"quoteCount": 3,
"viewCount": 7190,
"createdAt": "Sat Jun 27 13:47:11 +0000 2026",
"lang": "en",
"bookmarkCount": 17,
"isReply": false,
"inReplyToId": null,
"conversationId": "2070866534462087626",
"displayTextRange": [
0,
276
],
"inReplyToUserId": null,
"inReplyToUsername": null,
"author": {
"type": "user",
"userName": "HaiyuWu1",
"url": "https://x.com/HaiyuWu1",
"twitterUrl": "https://twitter.com/HaiyuWu1",
"id": "1512086648930148353",
"name": "Haiyu Wu",
"isVerified": false,
"isBlueVerified": true,
"verifiedType": null,
"profilePicture": "https://pbs.twimg.com/profile_images/1833875070005002241/xu88w6rH_normal.jpg",
"coverPicture": "https://pbs.twimg.com/profile_banners/1512086648930148353/1761024508",
"description": "",
"location": "San Diego, CA",
"followers": 81,
"following": 69,
"status": "",
"canDm": false,
"canMediaTag": true,
"createdAt": "Thu Apr 07 15:15:41 +0000 2022",
"entities": {
"description": {
"urls": []
},
"url": {}
},
"fastFollowersCount": 0,
"favouritesCount": 140,
"hasCustomTimelines": true,
"isTranslator": false,
"mediaCount": 5,
"statusesCount": 61,
"withheldInCountries": [],
"affiliatesHighlightedLabel": {},
"possiblySensitive": false,
"pinnedTweetIds": [
"2070866534462087626"
],
"profile_bio": {
"description": "Research interests: World model | Multi-model. \nI am unflashy but I glow. I am rooted but I flow.",
"entities": {
"description": {},
"url": {
"urls": [
{
"display_url": "haiyuwu.github.io",
"expanded_url": "https://haiyuwu.github.io/",
"indices": [
0,
23
],
"url": "https://t.co/VdJ2thamHZ"
}
]
}
}
},
"isAutomated": false,
"automatedBy": null
},
"extendedEntities": {
"media": [
{
"additional_media_info": {
"monetizable": false
},
"allow_download_status": {
"allow_download": true
},
"display_url": "pic.twitter.com/PJLekazWyp",
"expanded_url": "https://twitter.com/HaiyuWu1/status/2070866534462087626/video/1",
"ext_master_playlist_only": [],
"ext_media_availability": {
"status": "Available"
},
"ext_playlists": [],
"id_str": "2070866482800861184",
"indices": [
277,
300
],
"media_key": "13_2070866482800861184",
"media_results": {
"id": "QXBpTWVkaWFSZXN1bHRzOgwABAoAARy9Mhgcm6AAAAA=",
"result": {
"__typename": "ApiMedia",
"id": "QXBpTWVkaWE6DAAECgABHL0yGByboAAAAA==",
"media_key": "13_2070866482800861184"
}
},
"media_url_https": "https://pbs.twimg.com/amplify_video_thumb/2070866482800861184/img/s-Pi2z_hWvSWbf7v.jpg",
"original_info": {
"focus_rects": [],
"height": 1080,
"width": 1920
},
"sizes": {
"large": {
"h": 1080,
"w": 1920
}
},
"type": "video",
"url": "https://t.co/PJLekazWyp",
"video_info": {
"aspect_ratio": [
16,
9
],
"duration_millis": 36548,
"variants": [
{
"content_type": "application/x-mpegURL",
"url": "https://video.twimg.com/amplify_video/2070866482800861184/pl/vGG85lov132KMmBd.m3u8?tag=28&v=cfc"
},
{
"bitrate": 256000,
"content_type": "video/mp4",
"url": "https://video.twimg.com/amplify_video/2070866482800861184/vid/avc1/480x270/02AfzsadsERuRDGK.mp4?tag=28"
},
{
"bitrate": 832000,
"content_type": "video/mp4",
"url": "https://video.twimg.com/amplify_video/2070866482800861184/vid/avc1/640x360/683WXZY-7m1eBt_z.mp4?tag=28"
},
{
"bitrate": 2176000,
"content_type": "video/mp4",
"url": "https://video.twimg.com/amplify_video/2070866482800861184/vid/avc1/1280x720/E_Ak7xaZ17uh3gG_.mp4?tag=28"
},
{
"bitrate": 10368000,
"content_type": "video/mp4",
"url": "https://video.twimg.com/amplify_video/2070866482800861184/vid/avc1/1920x1080/OkjpIWBf9uFyGipk.mp4?tag=28"
}
]
}
}
]
},
"card": null,
"place": {},
"entities": {
"hashtags": [
{
"indices": [
850,
873
],
"text": "SelfSupervisedLearning"
},
{
"indices": [
874,
879
],
"text": "JEPA"
},
{
"indices": [
880,
891
],
"text": "WorldModel"
}
],
"symbols": [],
"timestamps": [],
"urls": [
{
"display_url": "youtube.com/watch?v=72Xj8k…",
"expanded_url": "https://www.youtube.com/watch?v=72Xj8k5WQX4",
"indices": [
664,
687
],
"url": "https://t.co/P9TXmk3fFa"
}
],
"user_mentions": [
{
"id_str": "48008938",
"indices": [
643,
650
],
"name": "Yann LeCun",
"screen_name": "ylecun"
},
{
"id_str": "1246070462679040000",
"indices": [
719,
735
],
"name": "Randall Balestriero",
"screen_name": "randall_balestr"
},
{
"id_str": "1233871671095853058",
"indices": [
751,
766
],
"name": "Morgan Levine",
"screen_name": "DrMorganLevine"
},
{
"id_str": "48008938",
"indices": [
792,
799
],
"name": "Yann LeCun",
"screen_name": "ylecun"
}
]
},
"quoted_tweet": null,
"retweeted_tweet": null,
"isLimitedReply": false,
"communityInfo": null,
"article": null
},
"retweeted_tweet": null,
"isLimitedReply": false,
"communityInfo": null,
"article": null
}