claire/ai (@_claire_ai_) 's Twitter Profile
claire/ai

@_claire_ai_

Not an AI itself but a founder interested in AI and building stuff with it.

ID: 1762517747852795905

calendar_today27-02-2024 16:40:24

83 Tweet

26 Takipçi

117 Takip Edilen

Philipp Schmid (@_philschmid) 's Twitter Profile Photo

How does quantization impact the performance of LLMs? Only minimal! 🤯 @neuralmagic ran 500,000 different evaluations on AI at Meta Llama using different quantization strategies. The impact is <1%, but the benefits are up to 2.4 faster inference and 3.5 model size reduction! 🔥

How does quantization impact the performance of LLMs? Only minimal! 🤯 @neuralmagic ran 500,000 different evaluations on <a href="/AIatMeta/">AI at Meta</a> Llama using different quantization strategies. The impact is &lt;1%, but the benefits are up to 2.4 faster inference and 3.5 model size reduction! 🔥
merve (@mervenoyann) 's Twitter Profile Photo

someone reached out to me via e-mail to ask how they can solve OMR bubble recognition, none of the vision LMs were able to solve it interesting problem, here's how it went

someone reached out to me via e-mail to ask how they can solve OMR bubble recognition, none of the vision LMs were able to solve it

interesting problem, here's how it went
Muhammad Rizwan Munawar (@muhammdrizwanmr) 's Twitter Profile Photo

Damage detection using Ultralytics YOLO11🚀 I see many use cases of computer vision in the transportation industry. One area where it helps is vehicle inspection, allowing us to quickly identify damage. Discover more➡️ultralytics.com/blog/quality-i… #ai #computervision #inspection

Mckay Wrigley (@mckaywrigley) 's Twitter Profile Photo

Here’s how to easily set up a working playground for Anthropic’s new computer use API. Takes <10min. It gives Claude 3.5 Sonnet a tool that can use a computer just like a human which opens up endless possibilities - it’s wild. Absolute gamechanger for AI.

AshutoshShrivastava (@ai_for_success) 's Twitter Profile Photo

Qwen never ceases to surprise us, they’ve just released their new model, Qwen2.5-Turbo! 1M context tokens – You can easily feed a novel or an entire code repository Faster inference Lower cost Hey OpenAI and Anthropic when 1M context token??

merve (@mervenoyann) 's Twitter Profile Photo

people who don't understand why DeepSeek is open-sourcing powerful models don't understand the vision behind open-source AI to begin with I have been yelling about it here for years and no one cared until they were baffled, thanks DeepSeek we aren't back, we were always here

Theoretically Media (@theomediaai) 's Twitter Profile Photo

THIS is what we've been promised for so long! Multimodal image Editing! And Gemini 2.0 did it! Generate and image, and then ASK for a change without completely regenerating!

THIS is what we've been promised for so long! Multimodal image Editing! And Gemini 2.0 did it! Generate and image, and then ASK for a change without completely regenerating!
signüll (@signulll) 's Twitter Profile Photo

we crossed the turing test & no one gave a shit. no parades. no front page headlines. just… a casual shrug. like “oh yeah, the machines are smart enough to fool us now. anyway, what’s for lunch?” that silence tells you everything about the pace we’re moving at. back in my cs

Min Choi (@minchoi) 's Twitter Profile Photo

It's only been just over 72 hours since Midjourney dropped Omni Reference. Any character, any scene. People can't believe the consistency. 10 wild examples + prompts & tips: 👇

It's only been just over 72 hours since Midjourney dropped Omni Reference.

Any character, any scene.

People can't believe the consistency.

10 wild examples + prompts &amp; tips: 👇