Mervin Praison (@mervinpraison) 's Twitter Profile
Mervin Praison

@mervinpraison

AI mer.vin

ID: 87552534

linkhttps://mer.vin calendar_today04-11-2009 22:35:05

457 Tweet

4,4K Followers

2,2K Following

Groq Inc (@groqinc) 's Twitter Profile Photo

Unlock the power of Serverless RAG (Retrieval-Augmented Generation) applications with this step-by-step tutorial powered by Groq!

Mervin Praison (@mervinpraison) 's Twitter Profile Photo

Solar Pro LLM Released: Did it Pass the AI Tests? ๐Ÿ’ก Best-in-class intelligence on a single GPU โš™๏ธ 22B parameters ๐Ÿš€ Created using Depth Upscaling ๐Ÿฆพ On Par with LLaMA 3.1 (70B) ๐ŸŒ Open-source for public use ๐Ÿ”‘ Available via, Huggingface, API and AWS Sub:

Mervin Praison (@mervinpraison) 's Twitter Profile Photo

OpenAI's Strawberry o1: World's best AI released! Test Results โš™๏ธ Reasoning-focused AI ๐Ÿš€ OpenAI's o1 Outperforms GPT-4o ๐Ÿ“ˆ PhD-level task accuracy ๐Ÿง  Excels at complex problems ๐Ÿง‘โ€๐Ÿ’ป Specialised for coding ๐Ÿ’ก Self-fact-checking Sub: youtube.com/watch?v=43PtzKโ€ฆ OpenAI OpenAI Developers

Tom Yeh (@proftomyeh) 's Twitter Profile Photo

How does OpenAI train the Strawberry๐Ÿ“ (o1) model to spend more time thinking? I read the report. The report is mostly about ๐˜ธ๐˜ฉ๐˜ข๐˜ต impressive benchmark results they got. But in term of the ๐˜ฉ๐˜ฐ๐˜ธ, the report only offers one sentence: "Through reinforcement learning, o1

Mervin Praison (@mervinpraison) 's Twitter Profile Photo

AI News 0:00 - ๐Ÿš€ Introduction 0:37 - ๐Ÿ—ฃ๏ธ Empathetic Voice Interface (EVI 2) Hume 1:46 - ๐Ÿฆ™ Llama Omni 8B AI at Meta 2:24 - ๐Ÿ“„ Jina AI reader - HTML to markdown LM 2:33 - ๐ŸŸ Fish Speech: open-source TTS Fish Audio 3:01 - ๐Ÿ“ท OCR2 Philipp Schmid 3:23 - ๐Ÿ‘๏ธ Mistral AI Pixtral 12B

Mervin Praison (@mervinpraison) 's Twitter Profile Photo

OpenAI o1 in ChatLLM & more features Abacus.AI bit.ly/abacusai-chatlโ€ฆ ๐Ÿ–ฅ๏ธ Multi-model AI interface ๐ŸŽจ AI image generation ๐Ÿ’ป Code creation and testing ๐Ÿ“Š AI-powered presentations ๐ŸŒ URL scraping and analysis ๐Ÿ“ธ Screenshot interpretation ๐Ÿ“ Custom knowledge integration ๐ŸŽฎ UI

Mervin Praison (@mervinpraison) 's Twitter Profile Photo

Google NotebookLM: ANYONE Can Create Podcast in 5 mins using AI ๐Ÿ“„ Upload and summarise documents โ“ AI-powered Q&A on uploaded content ๐ŸŽ™๏ธ AI-generated podcasts with human voices ๐Ÿ”— Google Drive integration ๐Ÿ“š Virtual research assistant capabilities ๐Ÿ—‚๏ธ Organize and synthesize

Whired (@whi__red) 's Twitter Profile Photo

unexployed WIP alแป‹a Mervin Praison is a machine for simple, beginner-friendly YouTube videos that are always bleeding edge He also supplies a ton of his matching code (for free!) that you can clone and play around with

Mervin Praison (@mervinpraison) 's Twitter Profile Photo

AI News 0:39 - ๐Ÿ’ผ Microsoft AI assistant for work 1:22 - ๐Ÿ‘“ Snapchat AI spectacles 2:00 - ๐ŸŽจ Snapchat's AI image and video generation 2:16 - ๐ŸŽฌ Runway Gen 3 Alpha video generation 2:32 - ๐Ÿค– Pudu Robotics 2:44 - ๐Ÿ’ƒ Disney's research on robot movement 3:04 - ๐ŸŒ 1X

Mervin Praison (@mervinpraison) 's Twitter Profile Photo

Llama 3.2 Released by AI at Meta ๐Ÿ“Š Model Sizes: 1B, 3B, 11B, 90B ๐Ÿ“ฑ On-device: 1B, 3B | Multimodal: 11B, 90B ๐Ÿ“ 1B and 3B models support 128K Context Length ๐Ÿ“ˆ 90B model outperforms GPT-4 in some benchmarks. ๐ŸŒ Available on cloud providers and model APIs. ๐Ÿ”„ 15% fewer tokens