“ It's a natural move … and based on my interactions with them this is clearly a personal relationship between the two, and not just a business one … it's clear that they both want to see the company succeed” Thanks @IanCutress https://
morethanmoore.substack.com/p/exclusive-al
ignment-for-tenstorrent
…
@tenstorrent
-
Tenstorrent Leadership Alignment Signals Strong Personal Commitment
By
–
-
Reversible Architectures Enable Arbitrarily Deep Residual Neural Networks
By
–
#TenstorrentTop10: each week Tenstorrent will be highlighting a paper that has inspired our product development. This week: Reversible Architectures for Arbitrarily Deep Residual Neural Networks. #NeuralNetworks #research #tenstorrent https://
tenstorrent.com/research/tenst
orrent-top-10-reversible-architectures-for-arbitrarily-deep-residual-neural-networks/
… -
Understanding Knowledge: Information Transformation and Relationships
By
–
"You can look at something, identify and modify it in your head, move it around… what we call understanding … is the sum of the relationship to all of that information. Knowledge is the layer above information, which is transformative." – @jimkxa
-
Synthetic Gradients Inspire Tenstorrent Neural Network Research
By
–
#TenstorrentTop10: each week Tenstorrent will be highlighting a paper that has inspired our product development. This week's paper is Decoupled Neural Interfaces using Synthetic Gradients. #neuralnetwork #syntheticgradients # #research #tenstorrent https://
tenstorrent.com/research/tenst
orrent-top-10-decoupled-neural-interfaces-using-synthetic-gradients/
… -
Tenstorrent Unveils Ascalon High Performance RISC-V AI Processor
By
–
"Introducing Tenstorrent's high performance #riscv processor Ascalon, the disruptive high-performance processor for AI and server with the best performance and power efficiency." -Chief CPU Architect Wei-han Lien introduces Ascalon at the RISC-V Summit
-
Scalable Chip Architecture for Trillion Parameter ML Training
By
–
“Their design … has made it so that 1 core looks like 1 chip looks like a whole network. ML training with trillion parameter models requires a lot of compute, you need an architecture that operates the same with 1 chip or 1000 chips.” Thanks @IanCutress
-
Software Stack Enables ML Training and Inference Across Frameworks
By
–
"Software is one of the hardest obstacles in cracking the ML training and inference market. Our software stack is capable of running both training and inference from all the different popular machine learning frameworks and architecture design." @TMLS_TO
-
Tenstorrent’s High-Performance RISC-V Processor for Computation
By
–
@Tenstorrent
's Wei-han Lien talking about High-performance @RISC_V Processor for Computation Acceleration and Server. Full talk to come, thanks for having us #RISCVSummit! #riscv #highperformance #hardware #compute #tenstorrent -
Join TMLS Toronto Conference on Machine Learning
By
–
Come join us at @TMLS_TO 's Toronto Conference next week. Register here –> https://
torontomachinelearning.com #computing #machinelearning #tenstorrent #toronto -
Belief in Problem-Solving Powers Innovation and Breakthrough
By
–
"If you think something is unsolvable it will not get solved. Solving problems is partly about believing you can solve everything and sometimes that means you have to push through the fundamentals." – @jimkxa at @TWSemicon