Viewing enriched Twitter post
Search-R1: Training LLMs to Reason and Leverage Search Engines with RL This paper tackles search-augmented reasoning by teaching LLMs to query a search engine multiple times—while they reason—using reinforcement learning. Read on for more: • Multi-turn retrieval – The LLM can… https://t.co/pRMWzpthIa
{
"score": 1.0,
"scored_at": "2025-08-09T13:47:19.523685",
"import_source": "unknown_source",
"links_checked": true,
"checked_at": "2025-08-10T10:33:21.941080",
"original_structure": "had_data_only",
"media": [
{
"id": "1900550990249832448",
"type": "photo",
"url": "https://t.co/pRMWzpthIa",
"media_url": null,
"media_url_https": "https://pbs.twimg.com/media/GmAdDltbsAAN0Q_.jpg",
"display_url": "pic.x.com/pRMWzpthIa",
"expanded_url": "https://x.com/omarsar0/status/1900550994116960391/photo/1"
}
]
}{
"tweet": {
"bookmark_count": 228,
"bookmarked": false,
"created_at": "Fri Mar 14 14:14:02 +0000 2025",
"conversation_id_str": "1900550994116960391",
"display_text_range": [
0,
280
],
"entities": {
"hashtags": [],
"media": [
{
"display_url": "pic.x.com/pRMWzpthIa",
"expanded_url": "https://x.com/omarsar0/status/1900550994116960391/photo/1",
"id_str": "1900550990249832448",
"indices": [
281,
304
],
"media_key": "3_1900550990249832448",
"media_url_https": "https://pbs.twimg.com/media/GmAdDltbsAAN0Q_.jpg",
"type": "photo",
"url": "https://t.co/pRMWzpthIa",
"ext_media_availability": {
"status": "Available"
},
"features": {
"large": {
"faces": [
{
"x": 135,
"y": 593,
"h": 55,
"w": 55
},
{
"x": 660,
"y": 594,
"h": 56,
"w": 56
},
{
"x": 578,
"y": 130,
"h": 57,
"w": 57
},
{
"x": 453,
"y": 148,
"h": 69,
"w": 69
},
{
"x": 76,
"y": 124,
"h": 74,
"w": 74
}
]
},
"medium": {
"faces": [
{
"x": 135,
"y": 593,
"h": 55,
"w": 55
},
{
"x": 660,
"y": 594,
"h": 56,
"w": 56
},
{
"x": 578,
"y": 130,
"h": 57,
"w": 57
},
{
"x": 453,
"y": 148,
"h": 69,
"w": 69
},
{
"x": 76,
"y": 124,
"h": 74,
"w": 74
}
]
},
"small": {
"faces": [
{
"x": 100,
"y": 440,
"h": 40,
"w": 40
},
{
"x": 490,
"y": 441,
"h": 41,
"w": 41
},
{
"x": 429,
"y": 96,
"h": 42,
"w": 42
},
{
"x": 336,
"y": 109,
"h": 51,
"w": 51
},
{
"x": 56,
"y": 92,
"h": 54,
"w": 54
}
]
},
"orig": {
"faces": [
{
"x": 135,
"y": 593,
"h": 55,
"w": 55
},
{
"x": 660,
"y": 594,
"h": 56,
"w": 56
},
{
"x": 578,
"y": 130,
"h": 57,
"w": 57
},
{
"x": 453,
"y": 148,
"h": 69,
"w": 69
},
{
"x": 76,
"y": 124,
"h": 74,
"w": 74
}
]
}
},
"sizes": {
"large": {
"h": 783,
"w": 915,
"resize": "fit"
},
"medium": {
"h": 783,
"w": 915,
"resize": "fit"
},
"small": {
"h": 582,
"w": 680,
"resize": "fit"
},
"thumb": {
"h": 150,
"w": 150,
"resize": "crop"
}
},
"original_info": {
"height": 783,
"width": 915,
"focus_rects": [
{
"x": 0,
"y": 0,
"w": 915,
"h": 512
},
{
"x": 43,
"y": 0,
"w": 783,
"h": 783
},
{
"x": 91,
"y": 0,
"w": 687,
"h": 783
},
{
"x": 238,
"y": 0,
"w": 392,
"h": 783
},
{
"x": 0,
"y": 0,
"w": 915,
"h": 783
}
]
},
"media_results": {
"result": {
"media_key": "3_1900550990249832448"
}
}
}
],
"symbols": [],
"timestamps": [],
"urls": [],
"user_mentions": []
},
"extended_entities": {
"media": [
{
"display_url": "pic.x.com/pRMWzpthIa",
"expanded_url": "https://x.com/omarsar0/status/1900550994116960391/photo/1",
"id_str": "1900550990249832448",
"indices": [
281,
304
],
"media_key": "3_1900550990249832448",
"media_url_https": "https://pbs.twimg.com/media/GmAdDltbsAAN0Q_.jpg",
"type": "photo",
"url": "https://t.co/pRMWzpthIa",
"ext_media_availability": {
"status": "Available"
},
"features": {
"large": {
"faces": [
{
"x": 135,
"y": 593,
"h": 55,
"w": 55
},
{
"x": 660,
"y": 594,
"h": 56,
"w": 56
},
{
"x": 578,
"y": 130,
"h": 57,
"w": 57
},
{
"x": 453,
"y": 148,
"h": 69,
"w": 69
},
{
"x": 76,
"y": 124,
"h": 74,
"w": 74
}
]
},
"medium": {
"faces": [
{
"x": 135,
"y": 593,
"h": 55,
"w": 55
},
{
"x": 660,
"y": 594,
"h": 56,
"w": 56
},
{
"x": 578,
"y": 130,
"h": 57,
"w": 57
},
{
"x": 453,
"y": 148,
"h": 69,
"w": 69
},
{
"x": 76,
"y": 124,
"h": 74,
"w": 74
}
]
},
"small": {
"faces": [
{
"x": 100,
"y": 440,
"h": 40,
"w": 40
},
{
"x": 490,
"y": 441,
"h": 41,
"w": 41
},
{
"x": 429,
"y": 96,
"h": 42,
"w": 42
},
{
"x": 336,
"y": 109,
"h": 51,
"w": 51
},
{
"x": 56,
"y": 92,
"h": 54,
"w": 54
}
]
},
"orig": {
"faces": [
{
"x": 135,
"y": 593,
"h": 55,
"w": 55
},
{
"x": 660,
"y": 594,
"h": 56,
"w": 56
},
{
"x": 578,
"y": 130,
"h": 57,
"w": 57
},
{
"x": 453,
"y": 148,
"h": 69,
"w": 69
},
{
"x": 76,
"y": 124,
"h": 74,
"w": 74
}
]
}
},
"sizes": {
"large": {
"h": 783,
"w": 915,
"resize": "fit"
},
"medium": {
"h": 783,
"w": 915,
"resize": "fit"
},
"small": {
"h": 582,
"w": 680,
"resize": "fit"
},
"thumb": {
"h": 150,
"w": 150,
"resize": "crop"
}
},
"original_info": {
"height": 783,
"width": 915,
"focus_rects": [
{
"x": 0,
"y": 0,
"w": 915,
"h": 512
},
{
"x": 43,
"y": 0,
"w": 783,
"h": 783
},
{
"x": 91,
"y": 0,
"w": 687,
"h": 783
},
{
"x": 238,
"y": 0,
"w": 392,
"h": 783
},
{
"x": 0,
"y": 0,
"w": 915,
"h": 783
}
]
},
"media_results": {
"result": {
"media_key": "3_1900550990249832448"
}
}
}
]
},
"favorite_count": 233,
"favorited": false,
"full_text": "Search-R1: Training LLMs to Reason and Leverage Search Engines with RL\n\nThis paper tackles search-augmented reasoning by teaching LLMs to query a search engine multiple times—while they reason—using reinforcement learning.\n\nRead on for more:\n\n• Multi-turn retrieval – The LLM can… https://t.co/pRMWzpthIa",
"is_quote_status": false,
"lang": "en",
"possibly_sensitive": false,
"possibly_sensitive_editable": true,
"quote_count": 4,
"reply_count": 5,
"retweet_count": 41,
"retweeted": false,
"user_id_str": "3448284313",
"id_str": "1900550994116960391",
"note_tweet": {
"is_expandable": true,
"note_tweet_results": {
"result": {
"id": "Tm90ZVR3ZWV0OjE5MDA1NTA5OTM5MjM5ODEzMTI=",
"text": "Search-R1: Training LLMs to Reason and Leverage Search Engines with RL\n\nThis paper tackles search-augmented reasoning by teaching LLMs to query a search engine multiple times—while they reason—using reinforcement learning.\n\nRead on for more:\n\n• Multi-turn retrieval – The LLM can interleave text generation with repeated calls to a search engine, refining queries each step. This differs from simple one-shot retrieval-augmented generation (RAG).\n\n• Fully RL-based training – Unlike prior “tool-using” approaches that need large supervised datasets, the authors rely on outcome rewards only. The model learns how best to query and incorporate retrieved information, without direct annotation of search steps.\n\n• Retrieved token masking – To stabilize training, the authors ensure only model-generated tokens are optimized in the policy gradient, preventing the search engine’s returned text from skewing the RL updates.\n\n• Impressive gains – Across seven QA benchmarks (NQ, TriviaQA, PopQA, HotpotQA, etc.), Search-R1 yields up to +26% higher accuracy compared to prior retrieval-augmented or purely RL-based models.\n\n• Flexible across architectures – The framework works on both “base” and “instruct” variants of Qwen and LLaMA, showing its general applicability.\n\nThis looks extremely useful and applicable to building effective deep research agents. The code and model checkpoints were also made available.",
"entity_set": {
"hashtags": [],
"symbols": [],
"urls": [],
"user_mentions": []
}
}
}
}
},
"user": {
"__typename": "User",
"id": "VXNlcjozNDQ4Mjg0MzEz",
"rest_id": "3448284313",
"affiliates_highlighted_label": {},
"has_graduated_access": true,
"is_blue_verified": true,
"profile_image_shape": "Circle",
"legacy": {
"can_dm": true,
"can_media_tag": true,
"created_at": "Fri Sep 04 12:59:26 +0000 2015",
"default_profile": false,
"default_profile_image": false,
"description": "Building with AI agents @dair_ai • Prev: Meta AI, Galactica LLM, Elastic, PaperswithCode, PhD • I share insights on how to build with LLMs & AI Agents ⬇️",
"entities": {
"description": {
"urls": []
},
"url": {
"urls": [
{
"display_url": "dair-ai.thinkific.com",
"expanded_url": "https://dair-ai.thinkific.com/",
"url": "https://t.co/JBU5beHQNs",
"indices": [
0,
23
]
}
]
}
},
"fast_followers_count": 0,
"favourites_count": 30980,
"followers_count": 251285,
"friends_count": 619,
"has_custom_timelines": true,
"is_translator": false,
"listed_count": 4202,
"location": "",
"media_count": 3523,
"name": "elvis",
"normal_followers_count": 251285,
"pinned_tweet_ids_str": [
"1938256196626153624"
],
"possibly_sensitive": false,
"profile_banner_url": "https://pbs.twimg.com/profile_banners/3448284313/1565974901",
"profile_image_url_https": "https://pbs.twimg.com/profile_images/939313677647282181/vZjFWtAn_normal.jpg",
"profile_interstitial_type": "",
"screen_name": "omarsar0",
"statuses_count": 14745,
"translator_type": "regular",
"url": "https://t.co/JBU5beHQNs",
"verified": false,
"want_retweets": false,
"withheld_in_countries": []
},
"professional": {
"rest_id": "1460724937073770505",
"professional_type": "Creator",
"category": [
{
"id": 713,
"name": "Science & Technology",
"icon_name": ""
}
]
},
"tipjar_settings": {
"is_enabled": false,
"cash_app_handle": "£elvisomarsaravia"
}
},
"views": "29880"
}