Debasis

ISSCC 2025 conference statistics #ISSCC25
Happening in San Francisco

Marketcaps of the top semiconductor companies based on Visual Capitalist report

#NVIDIA with $3.4T marketcap followed by #Broadcom ($1.1T) and #TSMC ($1T).. semiconductor market is heating up.. semiconductor market is poised to be $1T by 2028 on the wake of #AI and #AGI

Jensen Huang with NVL72 Blackwell chips on a wafer-like substrate

NVIDIA CEO with Captain America pose at CES 2025. How does NVL72 AI server would look like on a fictional superlarge wafer? This is what is demoed. #NVIDIA #NVL72 #AI #CES25

www.youtube.com/live/k82RwXq...

10 Breakthrough Technologies 2025 What will really matter in the long run? That’s the question we tackle each year as we compile this annual list.

10 Breakthrough Technologies 2025 - 1. Vera Rubin telescope, 2. GenAI search, 3. small LM, 4. Cattle emission reduction tech, 5. Robotaxis, 6. clean Jet fuel, 7. fast-learning robot, 8. HIV-prevention med, 9. Green steel, & 10. Stem-cell therapies
www.technologyreview.com/2025/01/03/1...

#GenAI #AI

Oil-painting by Chris diDomizio (2021) @ Jimmy Carter Presidential Library

Jimmy Carter - 39th President of United States

OpenAI announces new o3 model OpenAI announced a new o3 model on Friday. In a tweet ahead of its livestream, the company confirmed previous reports that it is skipping o2 due to copyright concerns, and is releasing early evals for the o3 model. It appears to be a successor to its o1…

OpenAI announces new o3 model

Introducing Phi-4: Microsoft’s Newest Small Language Model Specializing in Complex Reasoning | Microsoft Community Hub Today we are introducing Phi-4, our 14B parameter state-of-the-art small language model (SLM) that excels at complex reasoning in areas such as math, in...

Phi-4: Microsoft's Newest Small Language Model Specializing in Complex Reasoning
https://techcommunity.microsoft.com/blog/aiplatformblog/introducing-phi-4-microsoft%E2%80%99s-newest-small-language-model-specializing-in-comple/4357090

https://news.ycombinator.com/item?id=42405323

Richard Ho

OpenAI hardware engineer Rixhard Ho @ #neurips 2024: balanced system needed for #GenAI by optimizing computer, memory, network and storage for better performance per watt and TCO

LLM Models

Well, great news is that she and team from UW and AI2 are making the models completely open for others to use.

How much does 7B parameter generative #AI model cost to train with 5 trillion tokens? About $350K - accepted to Prof. Hannah Hajishirzi, Univ Washington during her talk at #neurips 2024

Whole semiconductor market was supposed to be a trillion dollar by 2030. It seems it changed now. What is the projected semiconductor market by 2030 when we expect the #AI #server market ro by $830

It was wrong to single out one group. She said she is going to remove the information if she presents the same material again.

MIT Prof Rosalind Picard at #NeurIPS 2024 about most and least trusted professions in USA, and where #AI researchers might be.

Next in #AI, are: Agentic AI, reasoning, understanding, and self-awareness - according to Ilya Sutskever, during his Test of Time Award at #neurips 2024 for 2014 paper 'Sequence to Sequence Learning with Neural Networks'

We had a wonderful Keynote Talk by Lidong Zhou this morning.
A Match Made in Silicon: The Co-Evolution of Systems and AI
West Exhibition Hall C, B3
https://buff.ly/49HcwwY

According to LSTM inventor Sepp Hochreiter, #LLM scaling reached the saturating point. Now it is time for industrialization of #AI
#Neurips

Reading about scaling laws recently I came by the interesting point:
Focus on a balance between models' size and performance is more important that aiming for larger models

Tsinghua University and ModelBest Inc propose the idea of “capacity density” to measure how efficiently a model uses its size

Neurips 2024 statistics

Record submission at #NeurIPS 2024. Total submission 15670; accepted papers: 4038...

#Neurips 2024 is packed

AWS re:Invent 2024 came with 𝟴𝟮𝟬 𝘀𝗲𝘀𝘀𝗶𝗼𝗻𝘀 𝗳𝗼𝗿 𝗚𝗲𝗻𝗔𝗜/𝗠𝗟 🤖 - that's an overwhelming 34% out of the 2378 total sessions! ✨

Regarding the announcements, we've created a small summary for the (opinionated) top 10 only considering GenAI & ML.

Lisa Su Is TIME's 2024 CEO of the Year The CEO engineered a remarkable turnaround for AMD

Congratulations, AMD CEO Lisa Su, on being named TIME’s CEO of the year!

time.com/7200909/ceo-...

#HPC #AI @time-magazine.bsky.social

Tuesday, December 10th, on NeurIPS
The first day of NeurIPS 2024 features three Expo sessions, Affinity workshops, and Tutorials. This day will include a wide variety of industry participants. See the schedule for presenters, topics, times and locations:

Overview of paper browser. A cluster for reinforcement learning is selected.

Paper Browser: only papers assigned to "physical models - physics" are shown.

Paper Browser: Filtered by author "Hoover" and detail is shown

Paper Brower: ZOOOOM in

🎺 Here comes the official 2024 NeurIPS paper browser:
- browse all NeurIPS papers in a visual way
- select clusters of interest and get cluster summary
- ZOOOOM in
- filter by human assigned keywords
- filter by substring (authors, titles)

neurips2024.vizhub.ai

#neurips by IBM Research Cambridge

AI roadmap

#AI technology development roadmap from Deloitte expert. Expected #AGI to be in 2030 or beyond.

A summer research opportunity for undergrads (& Chicago-area high school students) focusing on rigorous, applied, interdisciplinary data science research and rooted in a cohort community datascience.uchicago.edu/education/su... -- with @kylechard.bsky.social and others from @globuslabs.bsky.social

SC'24 recap The premiere annual conference of the high-performance computing community, SC24, was held in Atlanta last week, and it attracted a reco...

I spent the Thanksgiving break typing up my notes from #SC24 which I've posted online. 30% more words than my notes from SC23 (sorry!). Feedback is welcome!

https://buff.ly/41fBhho

#HPC

GenAI model comparison with human IQ.

OpenAI o1 model (Level 2 model) a
Surpassed the average human IQ of 100 - according to trackingAI.org
#AI #GenAI #OpenAI

About AI Scientist HPE at SC2024

AI Scientist: should be able to generate & test novel hypotheses grounded in scientific knowledge. It's
discovery process is composed to subtasks addressed by specialized AI agents. These agents would acquire & refine knowledge, drive experiments & simulations.
- HPE @ SC24
#GenAI #AIScientist

Good overview of LLM app architecture from @andreessenhorowitz.bsky.social. More than a year old, but still relevant.

a16z.com/emerging-arc...

#llm #ai #product #rag #aiagent