Stay ahead of the AI curve, courtesy of Graphcore's research team. Each month they sift through the latest papers and pick the most important and interesting work in the world of Artificial Intelligence. https://lnkd.in/eedXymGV
About us
Graphcore has created a new processor, the Intelligence Processing Unit (IPU), specifically designed for artificial intelligence. The IPU’s unique architecture means developers can run current machine learning models orders of magnitude faster. More importantly, it lets AI researchers undertake entirely new types of work, not possible using current technologies, to drive the next great breakthroughs in general machine intelligence. Our next generation 3D Wafer-on-Wafer Bow IPU systems are helping AI innovators worldwide to build better, more innovative AI solutions, whether their focus is on language and vision, exploring graph neural networks and LSTMs or creating something entirely new. We believe our IPU technology will become the worldwide standard for artificial intelligence compute. The performance of Graphcore’s IPU is going to be transformative across all industries and sectors whether you are a medical researcher, roboticist or building autonomous cars. Our team is at the forefront of the artificial intelligence revolution, enabling innovators from all industries and sectors to expand human potential with technology. What we do, really makes a difference. We're always interested in hearing from exceptional people to join our team.
- Website
-
http://www.graphcore.ai
External link for Graphcore
- Industry
- Semiconductor Manufacturing
- Company size
- 501-1,000 employees
- Headquarters
- Bristol, Bristol
- Type
- Privately Held
- Founded
- 2016
- Specialties
- artificial intelligence, deep learning, supercomputing, graph neural networks, NLP, and computer vision
Locations
Employees at Graphcore
Updates
-
Join Graphcore's Artemiy Bulavin on July 2nd for the inaugural LLVM Project Meetup in Bristol. Come for the compiler chat - stay for the drinks and food. https://lu.ma/34rv0grf
-
Graphcore reposted this
The Research and Applied AI Summit (RAAIS) returns to London on June 13. Excited to hear from this year’s incredible speaker lineup—always a highlight of the AI calendar. Looking forward to the ideas, the energy, and the conversations >> raais.co
At The Research and Applied AI Summit (RAAIS) last year, Sylvain Viguier Director of AI of Graphcore discusses how hardware improvements are key to making AI more accessible and sustainable.
-
How low can you go? Graphcore's Alex Titterton explains how lower-precision number formats are enabling more efficient AI models, examines the quantization techniques that make low-precision AI possible, and asks whether this approach is suitable for more than just LLMs. Alex was speaking at the recent Bristol MLOps Meetup. https://lnkd.in/ekTcyCTY
Lower precision AI models explained - Alex Titterton: Bristol MLOps Meetup
https://www.youtube.com/
-
The AI world is still processing the impact of DeepSeek V3 and its claimed efficiency gains. In his deep-dive talk - given at the recent MLOps Bristol Meetup - Graphcore's Alexandre Payot examines the techniques that made DeepSeek V3 possible, and how they can be applied to other Mixture of Experts models. https://lnkd.in/ek8x9472
DeepSeek and Mixture of Experts model efficiency explained: Alexandre Payot at Bristol MLOps Meetup
https://www.youtube.com/
-
There's still time to sign up for this week's (March 6th) Bristol MLOps meetup. We'll be talking DeepSeek, LLM efficiency, and using graphs to produce naturalistic LLM chat systems. https://lu.ma/y3k6cr06
-
We couldn't do what we do at Graphcore without our many suppliers of products and services. Ensuring that our partners are paid promptly helps their businesses thrive too. Congratulations to Graphcore's Accounts Payable team on receiving a Silver Award in the Fair Payment Code Awards - ensuring that at least 95% of all invoices are paid within 60 days and at least 95% of small business invoices within 30 days. Graphcore is also signed-up to the UK Small Business Commissioner's Fair Payment Code. https://lnkd.in/eWqMDPNd
-
Join us for the next Bristol MLOps meetup on March 6th. We'll be talking DeepSeek, LLM efficiency, and using graphs to produce naturalistic LLM chat systems. Sign up via the link. https://lu.ma/y3k6cr06