Viewing enriched Twitter post
Instead of forcing models to hold everything in an active context window, we can use hypernetworks to instantly compile documents and tasks directly into the model's weights. A step towards giving language models durable memory and fast adaptation. Blog: https://t.co/iHoifpsLMu
{
"score": 0.42,
"score_components": {
"author": 0.09,
"engagement": 0.0,
"quality": 0.12000000000000002,
"source": 0.135,
"nlp": 0.05,
"recency": 0.025
},
"scored_at": "2026-03-01T12:19:35.525684",
"import_source": "api_import",
"source_tagged_at": "2026-03-01T12:19:35.525699",
"enriched": true,
"enriched_at": "2026-03-01T12:19:35.525702"
}{
"type": "tweet",
"id": "2027240562898976770",
"url": "https://x.com/hardmaru/status/2027240562898976770",
"twitterUrl": "https://twitter.com/hardmaru/status/2027240562898976770",
"text": "Instead of forcing models to hold everything in an active context window, we can use hypernetworks to instantly compile documents and tasks directly into the model's weights. A step towards giving language models durable memory and fast adaptation.\n\nBlog: https://t.co/iHoifpsLMu",
"source": "Twitter for iPhone",
"retweetCount": 229,
"replyCount": 61,
"likeCount": 2505,
"quoteCount": 36,
"viewCount": 290558,
"createdAt": "Fri Feb 27 04:33:09 +0000 2026",
"lang": "en",
"bookmarkCount": 1994,
"isReply": false,
"inReplyToId": null,
"conversationId": "2027240562898976770",
"displayTextRange": [
0,
279
],
"inReplyToUserId": null,
"inReplyToUsername": null,
"author": {
"type": "user",
"userName": "hardmaru",
"url": "https://x.com/hardmaru",
"twitterUrl": "https://twitter.com/hardmaru",
"id": "2895499182",
"name": "hardmaru",
"isVerified": false,
"isBlueVerified": true,
"verifiedType": null,
"profilePicture": "https://pbs.twimg.com/profile_images/1678402467078234113/XN5Oy2UP_normal.jpg",
"coverPicture": "https://pbs.twimg.com/profile_banners/2895499182/1429351923",
"description": "",
"location": "Minato-ku, Tokyo",
"followers": 387697,
"following": 1813,
"status": "",
"canDm": false,
"canMediaTag": true,
"createdAt": "Mon Nov 10 11:05:07 +0000 2014",
"entities": {
"description": {
"urls": []
},
"url": {}
},
"fastFollowersCount": 0,
"favouritesCount": 143443,
"hasCustomTimelines": true,
"isTranslator": false,
"mediaCount": 4454,
"statusesCount": 25772,
"withheldInCountries": [],
"affiliatesHighlightedLabel": {},
"possiblySensitive": false,
"pinnedTweetIds": [
"1990204623471395284"
],
"profile_bio": {
"description": "Co-Founder and CEO @SakanaAILabs 🎏",
"entities": {
"description": {
"hashtags": [],
"symbols": [],
"urls": [],
"user_mentions": [
{
"id_str": "0",
"indices": [
19,
32
],
"name": "",
"screen_name": "SakanaAILabs"
}
]
},
"url": {
"urls": [
{
"display_url": "sakana.ai",
"expanded_url": "https://sakana.ai/",
"indices": [
0,
23
],
"url": "https://t.co/cVQF43vIdu"
}
]
}
}
},
"isAutomated": false,
"automatedBy": null
},
"extendedEntities": {},
"card": {
"binding_values": [
{
"key": "vanity_url",
"value": {
"scribe_key": "vanity_url",
"string_value": "pub.sakana.ai"
}
},
{
"key": "domain",
"value": {
"string_value": "pub.sakana.ai"
}
},
{
"key": "title",
"value": {
"string_value": "Instant LLM Updates with Doc-to-LoRA and Text-to-LoRA"
}
},
{
"key": "card_url",
"value": {
"scribe_key": "card_url",
"string_value": "https://t.co/e2b8G9LJKe"
}
}
],
"card_platform": {
"platform": {
"audience": {
"name": "production"
},
"device": {
"name": "iPhone",
"version": "13"
}
}
},
"name": "summary",
"url": "https://t.co/e2b8G9LJKe",
"user_refs_results": []
},
"place": {},
"entities": {
"hashtags": [],
"symbols": [],
"urls": [
{
"display_url": "pub.sakana.ai/doc-to-lora/",
"expanded_url": "https://pub.sakana.ai/doc-to-lora/",
"indices": [
256,
279
],
"url": "https://t.co/iHoifpsLMu"
}
],
"user_mentions": []
},
"quoted_tweet": {
"type": "tweet",
"id": "2027240298666209535",
"url": "https://x.com/SakanaAILabs/status/2027240298666209535",
"twitterUrl": "https://twitter.com/SakanaAILabs/status/2027240298666209535",
"text": "We’re excited to introduce Doc-to-LoRA and Text-to-LoRA, two related research exploring how to make LLM customization faster and more accessible.\n\nhttps://t.co/ApVzVsBuv1\n\nBy training a Hypernetwork to generate LoRA adapters on the fly, these methods allow models to instantly internalize new information or adapt to new tasks.\n\nBiological systems naturally rely on two key cognitive abilities: durable long-term memory to store facts, and rapid adaptation to handle new tasks given limited sensory cues. While modern LLMs are highly capable, they still lack this flexibility. Traditionally, adding long-term memory or adapting an LLM to a specific downstream task requires an expensive and time-consuming model update, such as fine-tuning or context distillation, or relies on memory-intensive long prompts.\n\nTo bypass these limitations, our work focuses on the concept of cost amortization. We pay the meta-training cost once to train a hypernetwork capable of producing tasks or document specific LoRAs on demand. This turns what used to be a heavy engineering pipeline into a single, inexpensive forward pass. Instead of performing per-task optimization, the hypernetwork meta-learns update rules to instantly modify an LLM given a new task description or a long document.\n\nIn our experiments, Text-to-LoRA successfully specializes models to unseen tasks using just a natural language description. Building on this, Doc-to-LoRA is able to internalize factual documents. On a needle-in-a-haystack task, Doc-to-LoRA achieves near-perfect accuracy on instances five times longer than the base model's context window. It can even generalize to transfer visual information from a vision-language model into a text-only LLM, allowing it to classify images purely through internalized weights.\n\nImportantly, both methods run with sub-second latency, enabling rapid experimentation while avoiding the overhead of traditional model updates. This approach is a step towards lowering the technical barriers of model customization, allowing end-users to specialize foundation models via simple text inputs. We have released our code and papers for the community to explore.\n\nDoc-to-LoRA\nPaper: https://t.co/87xEEpf0GN\nCode: https://t.co/zBfQi2L9LW\n\nText-to-LoRA\nPaper: https://t.co/emLRZ4Vdvo\nCode: https://t.co/b9mrdoWWRB",
"source": "Twitter for iPhone",
"retweetCount": 346,
"replyCount": 70,
"likeCount": 2103,
"quoteCount": 102,
"viewCount": 550844,
"createdAt": "Fri Feb 27 04:32:06 +0000 2026",
"lang": "en",
"bookmarkCount": 2020,
"isReply": false,
"inReplyToId": null,
"conversationId": "2027240298666209535",
"displayTextRange": [
0,
276
],
"inReplyToUserId": null,
"inReplyToUsername": null,
"author": {
"type": "user",
"userName": "SakanaAILabs",
"url": "https://x.com/SakanaAILabs",
"twitterUrl": "https://twitter.com/SakanaAILabs",
"id": "218811492",
"name": "Sakana AI",
"isVerified": false,
"isBlueVerified": true,
"verifiedType": "Business",
"profilePicture": "https://pbs.twimg.com/profile_images/1885939209388929024/dtnrOdGp_normal.jpg",
"coverPicture": "https://pbs.twimg.com/profile_banners/218811492/1686643464",
"description": "",
"location": "Tokyo, Japan",
"followers": 62135,
"following": 0,
"status": "",
"canDm": false,
"canMediaTag": true,
"createdAt": "Tue Nov 23 10:20:07 +0000 2010",
"entities": {
"description": {
"urls": []
},
"url": {}
},
"fastFollowersCount": 0,
"favouritesCount": 1,
"hasCustomTimelines": true,
"isTranslator": false,
"mediaCount": 317,
"statusesCount": 899,
"withheldInCountries": [],
"affiliatesHighlightedLabel": {},
"possiblySensitive": false,
"pinnedTweetIds": [
"1823178623513239992"
],
"profile_bio": {
"description": "Sakana AI is an AI R&D company based in Tokyo. We want to develop AI solutions for Japan’s needs, and democratize AI in Japan. https://t.co/OWL1EpmWtn",
"entities": {
"description": {
"hashtags": [],
"symbols": [],
"urls": [
{
"display_url": "sakana.ai/careers",
"expanded_url": "http://sakana.ai/careers",
"indices": [
127,
150
],
"url": "https://t.co/OWL1EpmWtn"
}
],
"user_mentions": []
},
"url": {
"urls": [
{
"display_url": "sakana.ai",
"expanded_url": "https://sakana.ai/",
"indices": [
0,
23
],
"url": "https://t.co/1m2lSgnfB2"
}
]
}
}
},
"isAutomated": false,
"automatedBy": null
},
"extendedEntities": {
"media": [
{
"display_url": "pic.twitter.com/gId3J6hgEr",
"expanded_url": "https://twitter.com/SakanaAILabs/status/2027240298666209535/photo/1",
"ext_media_availability": {
"status": "Available"
},
"id_str": "2027240151857176577",
"indices": [
277,
300
],
"media_key": "16_2027240151857176577",
"media_results": {
"id": "QXBpTWVkaWFSZXN1bHRzOgwAAgoAARwiNC1l2yABCgACHCI0T5RbMP8AAA==",
"result": {
"__typename": "ApiMedia",
"id": "QXBpTWVkaWE6DAACCgABHCI0LWXbIAEKAAIcIjRPlFsw/wAA",
"media_key": "16_2027240151857176577"
}
},
"media_url_https": "https://pbs.twimg.com/tweet_video_thumb/HCI0LWXbIAEKKAD.jpg",
"original_info": {
"focus_rects": [],
"height": 480,
"width": 1080
},
"sizes": {
"large": {
"h": 480,
"w": 1080
}
},
"type": "animated_gif",
"url": "https://t.co/gId3J6hgEr",
"video_info": {
"aspect_ratio": [
9,
4
],
"variants": [
{
"bitrate": 0,
"content_type": "video/mp4",
"url": "https://video.twimg.com/tweet_video/HCI0LWXbIAEKKAD.mp4"
}
]
}
}
]
},
"card": null,
"place": {},
"entities": {
"hashtags": [],
"symbols": [],
"urls": [
{
"display_url": "pub.sakana.ai/doc-to-lora/",
"expanded_url": "https://pub.sakana.ai/doc-to-lora/",
"indices": [
147,
170
],
"url": "https://t.co/ApVzVsBuv1"
},
{
"display_url": "arxiv.org/abs/2602.15902",
"expanded_url": "https://arxiv.org/abs/2602.15902",
"indices": [
2186,
2209
],
"url": "https://t.co/87xEEpf0GN"
},
{
"display_url": "github.com/SakanaAI/Doc-t…",
"expanded_url": "https://github.com/SakanaAI/Doc-to-LoRA",
"indices": [
2216,
2239
],
"url": "https://t.co/zBfQi2L9LW"
},
{
"display_url": "arxiv.org/abs/2506.06105",
"expanded_url": "https://arxiv.org/abs/2506.06105",
"indices": [
2261,
2284
],
"url": "https://t.co/emLRZ4Vdvo"
},
{
"display_url": "github.com/SakanaAI/Text-…",
"expanded_url": "https://github.com/SakanaAI/Text-to-LoRA",
"indices": [
2291,
2314
],
"url": "https://t.co/b9mrdoWWRB"
}
],
"user_mentions": []
},
"quoted_tweet": null,
"retweeted_tweet": null,
"isLimitedReply": false,
"article": null
},
"retweeted_tweet": null,
"isLimitedReply": false,
"article": null
}