[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.] #  @aigleeson Louis Gleeson Louis Gleeson posts on X about $googl, open ai, accuracy, builder the most. They currently have XXXXXX followers and XXX posts still getting attention that total XXXXXX engagements in the last XX hours. ### Engagements: XXXXXX [#](/creator/twitter::1904718477296062464/interactions)  - X Week XXXXXXX -XX% - X Month XXXXXXX -XX% - X Months XXXXXXXXX +308% ### Mentions: XX [#](/creator/twitter::1904718477296062464/posts_active)  - X Week XX -XX% - X Month XX -XX% - X Months XXX +1,000% ### Followers: XXXXXX [#](/creator/twitter::1904718477296062464/followers)  - X Week XXXXXX +0.76% - X Month XXXXXX +2.70% - X Months XXXXXX +412% ### CreatorRank: XXXXXXX [#](/creator/twitter::1904718477296062464/influencer_rank)  ### Social Influence [#](/creator/twitter::1904718477296062464/influence) --- **Social category influence** [technology brands](/list/technology-brands) XX% [stocks](/list/stocks) X% **Social topic influence** [$googl](/topic/$googl) #3193, [open ai](/topic/open-ai) #2426, [accuracy](/topic/accuracy) #550, [builder](/topic/builder) 2%, [faster](/topic/faster) #795, [plain](/topic/plain) 1%, [llms](/topic/llms) 1%, [all the](/topic/all-the) 1%, [dawn](/topic/dawn) 1%, [artificial](/topic/artificial) X% **Top accounts mentioned or mentioned by** [@tstless](/creator/undefined) [@laurencekuek](/creator/undefined) [@thedevonwayne](/creator/undefined) [@andres1turing](/creator/undefined) [@talkingmusicz](/creator/undefined) [@otioai](/creator/undefined) [@jakepaul](/creator/undefined) [@mikedkirby](/creator/undefined) [@medeenatee](/creator/undefined) [@ansarullahanas_](/creator/undefined) [@willfitinpocket](/creator/undefined) [@thegoodgru](/creator/undefined) [@extended_brain](/creator/undefined) [@vbridgesruiz](/creator/undefined) [@semihal9000](/creator/undefined) [@agiguardian](/creator/undefined) [@j_scott_art](/creator/undefined) [@bo_brand](/creator/undefined) [@f28x5](/creator/undefined) [@n_d_nite](/creator/undefined) **Top assets mentioned** [Alphabet Inc Class A (GOOGL)](/topic/$googl) ### Top Social Posts [#](/creator/twitter::1904718477296062464/posts) --- Top posts by engagements in the last XX hours "I finally understand how large language models actually work After reading the 2025 textbook Foundations of LLMs It blew my mind and cleared up years of confusion Heres everything i learned (in plain english):" [X Link](https://x.com/aigleeson/status/1950852155008770555) [@aigleeson](/creator/x/aigleeson) 2025-07-31T09:32Z 13.3K followers, 1.1M engagements "RIP N8n and Zapier. OpenAI is launching Agent Builder at DevDay (October 6): a visual canvas for building agentic workflows with MCP integrations ChatKit widgets and tool connections. No code. Just drag drop and deploy agents. Here's what changed:" [X Link](https://x.com/aigleeson/status/1975150407744991708) [@aigleeson](/creator/x/aigleeson) 2025-10-06T10:45Z 13.3K followers, 12K engagements "Agent Builder is OpenAI's answer to "we have the best models but the worst UX for building with them." Visual workflow editor. Native MCP support. Pre-built connectors. Deploy directly to ChatGPT or API. This is what Cursor did for coding but for agent development" [X Link](https://x.com/aigleeson/status/1975150424933175383) [@aigleeson](/creator/x/aigleeson) 2025-10-06T10:45Z 13.3K followers, XXX engagements "@LaurenceKuek A view on those platforms are worth a lot more to an advertiser" [X Link](https://x.com/aigleeson/status/1979303077573545989) [@aigleeson](/creator/x/aigleeson) 2025-10-17T21:46Z 13.3K followers, XX engagements "holy sh*t OpenAI Google & Anthropic just dropped their internal AI playbooks for free how to build AI agents how to scale infra how to prompt better than XX% of devs Get all the guides below with links for free (no login required)" [X Link](https://x.com/aigleeson/status/1947944075426152762) [@aigleeson](/creator/x/aigleeson) 2025-07-23T08:57Z 13.4K followers, 625.5K engagements "Gemma is Google DeepMinds open sibling to Gemini the same architecture scaled down for accessibility. The wild stat: 250M downloads 85000 community variations This things becoming the Linux of LLMs" [X Link](https://x.com/aigleeson/status/1976954851751084184) [@aigleeson](/creator/x/aigleeson) 2025-10-11T10:15Z 13.4K followers, 3644 engagements "Full guide: Fine-tune Gemma with QLoRA in Colab Convert for MediaPipe or Transformers.js Run it client-side Publish on Hugging Face Docs + demo code here:" [X Link](https://x.com/aigleeson/status/1976954926497718437) [@aigleeson](/creator/x/aigleeson) 2025-10-11T10:15Z 13.4K followers, 3026 engagements "Holy shit AGI finally has a number. 🤯 For the first time we can actually measure how close we are to real Artificial General Intelligence thanks to a new paper from Yoshua Bengio Dawn Song Max Tegmark Eric Schmidt and others. For years everyones been throwing around the term AGI like its some mystical milestone. Something vague. Something out there in the future. But this paper just pinned it down with a definition that finally makes sense. They describe AGI as: An AI that can match the cognitive versatility and proficiency of a well-educated adult. No hype. No human-level buzzwords. Just a" [X Link](https://x.com/aigleeson/status/1980212988004200607) [@aigleeson](/creator/x/aigleeson) 2025-10-20T10:02Z 13.4K followers, 107.2K engagements "Whats powerful here is that this framework lets us track AGI like a scoreboard. Instead of arguing are we close to AGI we can now ask: How much cognitive ground remains before parity Next frontier: memory + multimodal reasoning" [X Link](https://x.com/aigleeson/status/1980213090206797998) [@aigleeson](/creator/x/aigleeson) 2025-10-20T10:02Z 13.4K followers, 1830 engagements "🚨 Google just did it again. They built a voice model that doesnt transcribe speech it understands it. Its called Speech-to-Retrieval (S2R) and its about to make voice search feel telepathic. Heres how it works (and why its a bigger deal than you think)" [X Link](https://x.com/aigleeson/status/1977681172181467638) [@aigleeson](/creator/x/aigleeson) 2025-10-13T10:21Z 13.3K followers, 57.8K engagements "The biggest labor shift in internet history just went unnoticed. AI now writes the majority of articles online according to a new study from Graphite. Nov 2022: ChatGPT launches Nov 2023: XX% AI-written Nov 2024: AI surpasses humans Researchers verified this across 65000 URLs XXXX% detection accuracy. But then came the twist: by mid-2024 AI content creation stopped growing. Why Because it doesnt perform. Companies flooded the web with AI-written posts saw no traffic and slammed the brakes. The result: an internet full of machine-generated content that humans no longer click. This isnt the" [X Link](https://x.com/aigleeson/status/1979526381915746716) [@aigleeson](/creator/x/aigleeson) 2025-10-18T12:34Z 13.4K followers, 9174 engagements "Action generalization is wild. Train on 2500 hours of video but only XXX hours have paired actions. The model still learns XX% of full action-conditioning accuracy. It extrapolates: train on Overworld actions only generalizes to Nether/End dimensions it only saw in unlabeled videos" [X Link](https://x.com/aigleeson/status/1974430453580075411) [@aigleeson](/creator/x/aigleeson) 2025-10-04T11:04Z 13.4K followers, XX engagements "Imagine: Your own fine-tuned language model running on your laptop phone or watch. No API keys. No subscriptions. No middlemen. Thats where AI is going local personal sovereign" [X Link](https://x.com/aigleeson/status/1976954914564981140) [@aigleeson](/creator/x/aigleeson) 2025-10-11T10:15Z 13.4K followers, 2911 engagements "Heres what 4-bit pretraining actually looks like under the hood 👇 Every GEMM (matrix multiply) runs in NVFP4. Only the final 1015% of layers stay in BF16 for stability. Gradients get stochastic rounding for unbiased updates. This figure says it all a full 4-bit pipeline running end-to-end" [X Link](https://x.com/aigleeson/status/1978405001182740800) [@aigleeson](/creator/x/aigleeson) 2025-10-15T10:18Z 13.4K followers, XXX engagements "They even compared it to MXFP4 the older 4-bit format. NVFP4 reaches the same loss XX% faster. MXFP4 needs 1.36T tokens to catch up with NVFP4s 1T. Thats a one-third reduction in training time. 4-bit isnt just possible. Its now efficient. Read the full paper here:" [X Link](https://x.com/aigleeson/status/1978405017502863647) [@aigleeson](/creator/x/aigleeson) 2025-10-15T10:18Z 13.4K followers, XXX engagements "Holy shit. Stanford just killed prompt engineering. It's called "Verbalized Sampling" and it proves aligned AI models aren't broken we've just been prompting them wrong this entire time. Here's the problem: Post-training alignment causes mode collapse. Ask ChatGPT " tell me a joke about coffee" X times and you'll get the SAME joke. Every. Single. Time. Everyone blamed the algorithms. Turns out it's deeper than that. The real culprit 'Typicality bias' in human preference data. Annotators systematically favor familiar conventional responses. This bias gets baked into reward models and aligned" [X Link](https://x.com/aigleeson/status/1979128281304899823) [@aigleeson](/creator/x/aigleeson) 2025-10-17T10:12Z 13.4K followers, 6083 engagements "After X years of using ChatGPT I can say that it is the technology that has revolutionized my life the most along with the Internet. So here are XX prompts that have transformed my day-to-day life and that could do the same for you:" [X Link](https://x.com/aigleeson/status/1955627856358502642) [@aigleeson](/creator/x/aigleeson) 2025-08-13T13:49Z 13.4K followers, 1M engagements "Holy shit. Google just launched Gemini XXX Computer Use and it quietly kills the prompt engineer era. It can use browsers spreadsheets even IDEs no scripts no hacks. Heres how it works (and why this changes everything):" [X Link](https://x.com/aigleeson/status/1976232713813360857) [@aigleeson](/creator/x/aigleeson) 2025-10-09T10:26Z 13.4K followers, 52.5K engagements "Safety is the big story though. Google trained safety into the model itself: - A per-step external safety service checks every action - System instructions can force user confirmation before high-risk tasks - No bypassing CAPTCHAs no hacking OS-level control Its cautious by design but still agentic" [X Link](https://x.com/aigleeson/status/1976232774643351624) [@aigleeson](/creator/x/aigleeson) 2025-10-09T10:26Z 13.4K followers, 2008 engagements "OMG you can now fine-tune your own AI model and run it entirely on your device. No servers. No API keys. No data leaks. Gemma X 270M just made local LLMs real and it only takes about an hour to build your own. Heres how:" [X Link](https://x.com/aigleeson/status/1976954839558217792) [@aigleeson](/creator/x/aigleeson) 2025-10-11T10:15Z 13.4K followers, 79.1K engagements "Step 1: Fine-tune Use QLoRA (Quantized Low-Rank Adaptation). You only update a few adapter weights instead of retraining billions. That means you can fine-tune Gemma X 270M on a free Colab T4 GPU in minutes" [X Link](https://x.com/aigleeson/status/1976954869455241549) [@aigleeson](/creator/x/aigleeson) 2025-10-11T10:15Z 13.4K followers, 3545 engagements "The paper also distinguishes between speed and intelligence. GPT-5 performs simple tasks quickly but raw speed smarter thinking it just means faster prediction. Humans trade speed for reflection. AIs dont yet" [X Link](https://x.com/aigleeson/status/1980213078072680600) [@aigleeson](/creator/x/aigleeson) 2025-10-20T10:02Z 13.4K followers, 1795 engagements "🚨 Google just shocked the world. They built an AI that can see cancer at the genetic level. Its called DeepSomatic and it identifies the DNA mutations driving tumors with record-breaking accuracy. Heres how it works (and why it changes everything):" [X Link](https://x.com/aigleeson/status/1980264788036059177) [@aigleeson](/creator/x/aigleeson) 2025-10-20T13:28Z 13.4K followers, 3672 engagements
[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]
@aigleeson Louis Gleeson
Louis Gleeson posts on X about $googl, open ai, accuracy, builder the most. They currently have XXXXXX followers and XXX posts still getting attention that total XXXXXX engagements in the last XX hours.
Engagements: XXXXXX #
- X Week XXXXXXX -XX%
- X Month XXXXXXX -XX%
- X Months XXXXXXXXX +308%
Mentions: XX #
- X Week XX -XX%
- X Month XX -XX%
- X Months XXX +1,000%
Followers: XXXXXX #
- X Week XXXXXX +0.76%
- X Month XXXXXX +2.70%
- X Months XXXXXX +412%
CreatorRank: XXXXXXX #
Social Influence #
Social category influence technology brands XX% stocks X%
Social topic influence $googl #3193, open ai #2426, accuracy #550, builder 2%, faster #795, plain 1%, llms 1%, all the 1%, dawn 1%, artificial X%
Top accounts mentioned or mentioned by @tstless @laurencekuek @thedevonwayne @andres1turing @talkingmusicz @otioai @jakepaul @mikedkirby @medeenatee @ansarullahanas_ @willfitinpocket @thegoodgru @extended_brain @vbridgesruiz @semihal9000 @agiguardian @j_scott_art @bo_brand @f28x5 @n_d_nite
Top assets mentioned Alphabet Inc Class A (GOOGL)
Top Social Posts #
Top posts by engagements in the last XX hours
"I finally understand how large language models actually work After reading the 2025 textbook Foundations of LLMs It blew my mind and cleared up years of confusion Heres everything i learned (in plain english):"
X Link @aigleeson 2025-07-31T09:32Z 13.3K followers, 1.1M engagements
"RIP N8n and Zapier. OpenAI is launching Agent Builder at DevDay (October 6): a visual canvas for building agentic workflows with MCP integrations ChatKit widgets and tool connections. No code. Just drag drop and deploy agents. Here's what changed:"
X Link @aigleeson 2025-10-06T10:45Z 13.3K followers, 12K engagements
"Agent Builder is OpenAI's answer to "we have the best models but the worst UX for building with them." Visual workflow editor. Native MCP support. Pre-built connectors. Deploy directly to ChatGPT or API. This is what Cursor did for coding but for agent development"
X Link @aigleeson 2025-10-06T10:45Z 13.3K followers, XXX engagements
"@LaurenceKuek A view on those platforms are worth a lot more to an advertiser"
X Link @aigleeson 2025-10-17T21:46Z 13.3K followers, XX engagements
"holy sh*t OpenAI Google & Anthropic just dropped their internal AI playbooks for free how to build AI agents how to scale infra how to prompt better than XX% of devs Get all the guides below with links for free (no login required)"
X Link @aigleeson 2025-07-23T08:57Z 13.4K followers, 625.5K engagements
"Gemma is Google DeepMinds open sibling to Gemini the same architecture scaled down for accessibility. The wild stat: 250M downloads 85000 community variations This things becoming the Linux of LLMs"
X Link @aigleeson 2025-10-11T10:15Z 13.4K followers, 3644 engagements
"Full guide: Fine-tune Gemma with QLoRA in Colab Convert for MediaPipe or Transformers.js Run it client-side Publish on Hugging Face Docs + demo code here:"
X Link @aigleeson 2025-10-11T10:15Z 13.4K followers, 3026 engagements
"Holy shit AGI finally has a number. 🤯 For the first time we can actually measure how close we are to real Artificial General Intelligence thanks to a new paper from Yoshua Bengio Dawn Song Max Tegmark Eric Schmidt and others. For years everyones been throwing around the term AGI like its some mystical milestone. Something vague. Something out there in the future. But this paper just pinned it down with a definition that finally makes sense. They describe AGI as: An AI that can match the cognitive versatility and proficiency of a well-educated adult. No hype. No human-level buzzwords. Just a"
X Link @aigleeson 2025-10-20T10:02Z 13.4K followers, 107.2K engagements
"Whats powerful here is that this framework lets us track AGI like a scoreboard. Instead of arguing are we close to AGI we can now ask: How much cognitive ground remains before parity Next frontier: memory + multimodal reasoning"
X Link @aigleeson 2025-10-20T10:02Z 13.4K followers, 1830 engagements
"🚨 Google just did it again. They built a voice model that doesnt transcribe speech it understands it. Its called Speech-to-Retrieval (S2R) and its about to make voice search feel telepathic. Heres how it works (and why its a bigger deal than you think)"
X Link @aigleeson 2025-10-13T10:21Z 13.3K followers, 57.8K engagements
"The biggest labor shift in internet history just went unnoticed. AI now writes the majority of articles online according to a new study from Graphite. Nov 2022: ChatGPT launches Nov 2023: XX% AI-written Nov 2024: AI surpasses humans Researchers verified this across 65000 URLs XXXX% detection accuracy. But then came the twist: by mid-2024 AI content creation stopped growing. Why Because it doesnt perform. Companies flooded the web with AI-written posts saw no traffic and slammed the brakes. The result: an internet full of machine-generated content that humans no longer click. This isnt the"
X Link @aigleeson 2025-10-18T12:34Z 13.4K followers, 9174 engagements
"Action generalization is wild. Train on 2500 hours of video but only XXX hours have paired actions. The model still learns XX% of full action-conditioning accuracy. It extrapolates: train on Overworld actions only generalizes to Nether/End dimensions it only saw in unlabeled videos"
X Link @aigleeson 2025-10-04T11:04Z 13.4K followers, XX engagements
"Imagine: Your own fine-tuned language model running on your laptop phone or watch. No API keys. No subscriptions. No middlemen. Thats where AI is going local personal sovereign"
X Link @aigleeson 2025-10-11T10:15Z 13.4K followers, 2911 engagements
"Heres what 4-bit pretraining actually looks like under the hood 👇 Every GEMM (matrix multiply) runs in NVFP4. Only the final 1015% of layers stay in BF16 for stability. Gradients get stochastic rounding for unbiased updates. This figure says it all a full 4-bit pipeline running end-to-end"
X Link @aigleeson 2025-10-15T10:18Z 13.4K followers, XXX engagements
"They even compared it to MXFP4 the older 4-bit format. NVFP4 reaches the same loss XX% faster. MXFP4 needs 1.36T tokens to catch up with NVFP4s 1T. Thats a one-third reduction in training time. 4-bit isnt just possible. Its now efficient. Read the full paper here:"
X Link @aigleeson 2025-10-15T10:18Z 13.4K followers, XXX engagements
"Holy shit. Stanford just killed prompt engineering. It's called "Verbalized Sampling" and it proves aligned AI models aren't broken we've just been prompting them wrong this entire time. Here's the problem: Post-training alignment causes mode collapse. Ask ChatGPT " tell me a joke about coffee" X times and you'll get the SAME joke. Every. Single. Time. Everyone blamed the algorithms. Turns out it's deeper than that. The real culprit 'Typicality bias' in human preference data. Annotators systematically favor familiar conventional responses. This bias gets baked into reward models and aligned"
X Link @aigleeson 2025-10-17T10:12Z 13.4K followers, 6083 engagements
"After X years of using ChatGPT I can say that it is the technology that has revolutionized my life the most along with the Internet. So here are XX prompts that have transformed my day-to-day life and that could do the same for you:"
X Link @aigleeson 2025-08-13T13:49Z 13.4K followers, 1M engagements
"Holy shit. Google just launched Gemini XXX Computer Use and it quietly kills the prompt engineer era. It can use browsers spreadsheets even IDEs no scripts no hacks. Heres how it works (and why this changes everything):"
X Link @aigleeson 2025-10-09T10:26Z 13.4K followers, 52.5K engagements
"Safety is the big story though. Google trained safety into the model itself: - A per-step external safety service checks every action - System instructions can force user confirmation before high-risk tasks - No bypassing CAPTCHAs no hacking OS-level control Its cautious by design but still agentic"
X Link @aigleeson 2025-10-09T10:26Z 13.4K followers, 2008 engagements
"OMG you can now fine-tune your own AI model and run it entirely on your device. No servers. No API keys. No data leaks. Gemma X 270M just made local LLMs real and it only takes about an hour to build your own. Heres how:"
X Link @aigleeson 2025-10-11T10:15Z 13.4K followers, 79.1K engagements
"Step 1: Fine-tune Use QLoRA (Quantized Low-Rank Adaptation). You only update a few adapter weights instead of retraining billions. That means you can fine-tune Gemma X 270M on a free Colab T4 GPU in minutes"
X Link @aigleeson 2025-10-11T10:15Z 13.4K followers, 3545 engagements
"The paper also distinguishes between speed and intelligence. GPT-5 performs simple tasks quickly but raw speed smarter thinking it just means faster prediction. Humans trade speed for reflection. AIs dont yet"
X Link @aigleeson 2025-10-20T10:02Z 13.4K followers, 1795 engagements
"🚨 Google just shocked the world. They built an AI that can see cancer at the genetic level. Its called DeepSomatic and it identifies the DNA mutations driving tumors with record-breaking accuracy. Heres how it works (and why it changes everything):"
X Link @aigleeson 2025-10-20T13:28Z 13.4K followers, 3672 engagements