Profile avatar
ankits0052.bsky.social
AI Research Enthusiast | Multimedia Analysis | LLMs Associate director at Accenture. PhD in LTI CMU, prev: Google, Bosch, merl, ARM Looking for the next breakthrough that will lead to AGI - understanding why LLMs actually work ankitshah009.github.io
46 posts 753 followers 1,366 following
Prolific Poster
Active Commenter

guides.library.ucsb.edu/az/databases...

How to prompt the O1 model?

Deepseek R1 cost 🤯🤯🤯

Amazing art work

Can we just admit that we NEED immigrants for our economy to run, from agriculture to high tech? We’re a goddamn nation of immigrants, aren’t we? Let’s be proud of our immigrant heritage!

Figure credits - www.reddit.com/r/OpenAI/com...

Grok is the best integrated ChatGPT is the fastest Claude is the smartest Perplexity is best for web search Gemini handles the largest files

This guy is 62 But his epigenetic age is 37 It is possible to reverse your age and get fitter as well. Dave Pascoe’s routine looks complex, but it’s based on simple principles. If you’re over 30, this breakdown of his biohacking system could be valuable to your health:

www.wsj.com/tech/ai/elon...

Devin costs 500$ per month as compared to 20$ of Cursor www.youtube.com/watch?v=oU3H...

Great idea with mixture of experts coding! Sources: Meta's internal coding tool, Metamate, incorporates OpenAI's GPT-4 alongside Llama to make the tool more useful for developers and other Meta staff (Kali Hays/Fortune) fortune.com/2024/12/03/m... www.techmeme.com/241203/p34#a...

Anthropic's Computer Use API meets speech LLM to create an autonomous Voice AI Agent that lets you control a computer with your voice. replit.com/@jerhadf/Hum...

My mailbox is literally flooded with discussions of whether “Deep learning is hitting a wall”, almost none of which cite my much-ridiculed-at-the-time 2022 paper called “Deep learning is hitting a wall”. nautil.us/deep-learnin...

First open-weights for @OpenAI-o1-like reasoning model! QwQ from the Qwen team is a 32B model that beats O1 mini and competes w/ O1 preview and is available under Apache 2.0 on @huggingface! 🤯 Has notable limitations including language mixing, recursive reasoning loops, and safety considerations

techcrunch.com/2024/11/27/a...

www.microsoft.com/en-us/resear...

Linkup connects LLMs with premium content sources (legally) techcrunch.com/2024/11/28/l...

The most important life skill: Speed. Make decisions fast. Get to the point. Write shorter emails. Make sh*t happen. Be resourceful. If you do this you’ll live a life of freedom. Without urgency you’ll learn what it’s like to be financially and mentally broke.

Wow is this real or generated?

blog.ezyang.com/2019/05/pyto... PyTorch Internals.

How to run AI models locally? -> Go to 🤗 @huggingface -> Grab the GGUF model link -> Drop it into 👋 Jan Hub That's all there is to it.

go.bsky.app/RVyCym

Did you know there’s a site to search for starter packs of people to follow? Built by @mubashariqbal.com blueskydirectory.com/starter-pack...

Speech/ASR folks -- Whisper hallucinations (techcrunch.com/2024/10/26/o...) look a lot like the "faithfulness hallucinations" we have with LLMs -- can we just reuse the tech we already have for mitigating them in transformer-based LLMs? E.g. arxiv.org/abs/2410.18860 or arxiv.org/abs/2410.15999

AI Adoption in 2024: 74% of Companies Struggle to Achieve and Scale Value. www.bcg.com/press/24octo...