FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning FlashAttention-2 is 2x faster than previous version. Release blog: https://
crfm.stanford.edu/2023/07/17/fla
sh2.html
… Paper: https://
tridao.me/publications/f
lash2/flash2.pdf
…
AI HARDWARE
-
FlashAttention-2: 2x Faster Attention with Better Parallelism
By
–
-
Meta and Qualcomm Collaborate to Run Llama 2 on Smartphone
By
–
Meta and Qualcomm collaborate to run Llama 2 on smartphone https://actuia.com/actualite/meta-et-qualcomm-collaborent-pour-executer-llama-2-sur-smartphone/
… #AI #artificialintelligence
@Meta @Qualcomm -
BittWare Offers Multiple Paths for Groq AI Inference Solutions
By
–
@BittWareInc offers several paths to get @GroqInc #AI #inference, whether you are in #development or ready for deployment. For more information: http://
bittware.com/products/groq/ -
EvoJAX Accelerates Artificial Evolution Experiments on GPUs
By
–
Happy to see EvoJAX being used for this work too, to accelerate artificial evolution experiments on GPUs!
-
MIT Develops Robotic Touch Sensor Skin for Robots
By
–
MIT researchers give robots a "human touch" with sensor skin, a robotic touch sensor fabrication method: https://
bit.ly/43wr5yC -
Groq Hiring Senior Manager System QA Test Strategy
By
–
@GroqInc is #hiring a Senior Manager of System QA & Test. Help defy gravity and reshape the landscape of #AI by working on an overall system QA test strategy as well as managing testing, resource planning, and schedules. https://
groq.com/careers/?gh_ji
d=5575906003
…
#hiringtoptalent #hiringnow #SQA -
Tenstorrent focuses on customer onboarding and model performance
By
–
“
@tenstorrent
’s primary mission [today] is onboarding more customers and compiling more models performantly.” @jimkxa in @ojoyoshida . Read the full article here –> https://
tenstorrent.com/research/the-o
jo-yoshi-report-jim-kellers-journey-from-cpus-to-ceo/
… -
Groq Welcomes Alex Fediw to Enable Predictable AI Performance
By
–
Groq solutions produce predictable, repeatable performance metrics at scale. We're thrilled to have Alex Fediw on our team to help enable developers to look ahead, identify & adjust components of their model, instead of making trade-offs between HW and SW. #hiring #hiringnow pic.twitter.com/70IuZrzQ12
— Groq Inc (@GroqInc) 18 juillet 2023Groq solutions produce predictable, repeatable performance metrics at scale. We're thrilled to have Alex Fediw on our team to help enable developers to look ahead, identify & adjust components of their model, instead of making trade-offs between HW and SW. #hiring #hiringnow
-
India launches AIRAWAT AI supercomputer for ecosystem development
By
–
Say hello to India’s AI supercomputer AIRAWAT! It is a worthwhile development to promote India’s strong AI ecosystem, abundant data, vibrant digital economy, and talented workforce.
— IndiaAI (@OfficialINDIAai) 18 juillet 2023
Want to read more about the capabilities of AIRAWAT? Visit: https://t.co/FlYYJjmD7n pic.twitter.com/YQFS6HazVGSay hello to India’s AI supercomputer AIRAWAT! It is a worthwhile development to promote India’s strong AI ecosystem, abundant data, vibrant digital economy, and talented workforce. Want to read more about the capabilities of AIRAWAT? Visit: https://
rb.gy/21ys5 -
Tenstorrent Cambridge Event: RISC-V AI Hardware Innovation
By
–
Cambridge, come learn more about what @tenstorrent is building with #riscv. Sign up here –> https://
bit.ly/TTCambridgeCon
vo
…