Viewing enriched Twitter post
New NanoGPT Speedrun WR at 86.1 (-0.7s), by replacing partitioned hyperconnections with a simple idea: feed the exact same context vector into the last 3 attn layers, so late stage attn doesn't get polluted by prediction MLPs. Opinion: AI research agents are handicapped until they have a mech-interp toolkit. Many sub-3min architecture improvements came from analyzing weights. https://t.co/o9WeUF7PHl
{
"media": [
{
"url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2030465730718908884/media_0.jpg",
"media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2030465730718908884/media_0.jpg",
"type": "photo",
"filename": "media_0.jpg"
},
{
"url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2030465730718908884/media_1.jpg",
"media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2030465730718908884/media_1.jpg",
"type": "photo",
"filename": "media_1.jpg"
}
],
"processed_at": "2026-03-08T14:29:25.960703",
"pipeline_version": "2.0"
}{
"type": "tweet",
"id": "2030465730718908884",
"url": "https://x.com/classiclarryd/status/2030465730718908884",
"twitterUrl": "https://twitter.com/classiclarryd/status/2030465730718908884",
"text": "New NanoGPT Speedrun WR at 86.1 (-0.7s), by replacing partitioned hyperconnections with a simple idea: feed the exact same context vector into the last 3 attn layers, so late stage attn doesn't get polluted by prediction MLPs. Opinion: AI research agents are handicapped until they have a mech-interp toolkit. Many sub-3min architecture improvements came from analyzing weights. https://t.co/o9WeUF7PHl",
"source": "Twitter for iPhone",
"retweetCount": 13,
"replyCount": 5,
"likeCount": 244,
"quoteCount": 0,
"viewCount": 15928,
"createdAt": "Sun Mar 08 02:08:49 +0000 2026",
"lang": "en",
"bookmarkCount": 126,
"isReply": false,
"inReplyToId": null,
"conversationId": "2030465730718908884",
"displayTextRange": [
0,
276
],
"inReplyToUserId": null,
"inReplyToUsername": null,
"author": {
"type": "user",
"userName": "classiclarryd",
"url": "https://x.com/classiclarryd",
"twitterUrl": "https://twitter.com/classiclarryd",
"id": "1791834924317560832",
"name": "Larry Dial",
"isVerified": false,
"isBlueVerified": true,
"verifiedType": null,
"profilePicture": "https://pbs.twimg.com/profile_images/1959245933268672512/oAdbl4gC_normal.jpg",
"coverPicture": "https://pbs.twimg.com/profile_banners/1791834924317560832/1755955844",
"description": "",
"location": "",
"followers": 1170,
"following": 27,
"status": "",
"canDm": true,
"canMediaTag": true,
"createdAt": "Sat May 18 14:15:22 +0000 2024",
"entities": {
"description": {
"urls": []
},
"url": {}
},
"fastFollowersCount": 0,
"favouritesCount": 17,
"hasCustomTimelines": true,
"isTranslator": false,
"mediaCount": 24,
"statusesCount": 82,
"withheldInCountries": [],
"affiliatesHighlightedLabel": {},
"possiblySensitive": false,
"pinnedTweetIds": [],
"profile_bio": {
"description": "Technical Staff at Open Athena, working on Marin",
"entities": {
"description": {
"hashtags": [],
"symbols": [],
"urls": [],
"user_mentions": []
},
"url": {
"urls": [
{
"display_url": "larrydial.com",
"expanded_url": "http://larrydial.com",
"indices": [
0,
23
],
"url": "https://t.co/YF5dlTkrgU"
}
]
}
}
},
"isAutomated": false,
"automatedBy": null
},
"extendedEntities": {
"media": [
{
"allow_download_status": {
"allow_download": true
},
"display_url": "pic.twitter.com/Zlx6VmKzF1",
"expanded_url": "https://twitter.com/classiclarryd/status/2030465730718908884/photo/1",
"ext_media_availability": {
"status": "Available"
},
"features": {
"large": {
"faces": []
},
"orig": {
"faces": []
}
},
"id_str": "2030463021269807104",
"indices": [
277,
300
],
"media_key": "3_2030463021269807104",
"media_results": {
"id": "QXBpTWVkaWFSZXN1bHRzOgwAAQoAARwtp1woWnAACgACHC2p0wAawdQAAA==",
"result": {
"__typename": "ApiMedia",
"id": "QXBpTWVkaWE6DAABCgABHC2nXChacAAKAAIcLanTABrB1AAA",
"media_key": "3_2030463021269807104"
}
},
"media_url_https": "https://pbs.twimg.com/media/HC2nXChacAAM-EU.jpg",
"original_info": {
"focus_rects": [
{
"h": 410,
"w": 732,
"x": 0,
"y": 0
},
{
"h": 410,
"w": 410,
"x": 132,
"y": 0
},
{
"h": 410,
"w": 360,
"x": 157,
"y": 0
},
{
"h": 410,
"w": 205,
"x": 235,
"y": 0
},
{
"h": 410,
"w": 1226,
"x": 0,
"y": 0
}
],
"height": 410,
"width": 1226
},
"sizes": {
"large": {
"h": 410,
"w": 1226
}
},
"type": "photo",
"url": "https://t.co/Zlx6VmKzF1"
}
]
},
"card": null,
"place": {},
"entities": {
"hashtags": [],
"symbols": [],
"urls": [
{
"display_url": "github.com/KellerJordan/m…",
"expanded_url": "https://github.com/KellerJordan/modded-nanogpt/pull/241",
"indices": [
379,
402
],
"url": "https://t.co/o9WeUF7PHl"
}
],
"user_mentions": []
},
"quoted_tweet": null,
"retweeted_tweet": null,
"isLimitedReply": false,
"article": null
}