sid retweeted
Xilinx is now part of @AMD, creating the industry’s high-performance and adaptive computing leader. You can still find our updates on our Xilinx social media channels for the time being. We will update you here if that changes. More info on the deal here: https://teapaw.org/t.co/E6IaUKcrE3
sid retweeted
Xilinx Vivado 2021.1, a 30GB download, contains an 816MB file called "Third_Party_Software_End_User_License_Agreement.txt that contains nothing but 18M lines of open source licenses...
sid retweeted
New blog post: "A from-scratch tour of Bitcoin in Python" karpathy.github.io/2021/06/2… I've tinkered with Bitcoin last ~2 mo for fun, in this tour we create, digitally sign, and broadcast a Bitcoin transaction in pure Python, from scratch, and with zero dependencies. Final tx in hex:
sid retweeted
A Gentle Intro to Transformer language models and how eccox.io makes them more transparent
youtube.com/watch?v=rHrItfNe…
My talk at @PydataKhobar is now live! Thanks to the organizers.
Colab: colab.research.google.com/gi…
sid retweeted
Hey twitterettes. I need to list new HDLs for a presentation. Can you tell me which ones I missed? Got bluespec, chisel, clash, hardcaml, lava, (n)migen, myhdl, quokka, silice, spinalhdl, tl-verilog
sid retweeted
Came across two incredible posts by @aerinykim:
The intuition behind Shannon’s Entropy
towardsdatascience.com/the-i…
Perplexity Intuition (and its derivation)
towardsdatascience.com/perpl…
Intuitive explanations. Thanks for your wonderful work, Aerin!
sid retweeted
How GPT3 works. A visual thread.
A trained language model generates text.
We can optionally pass it some text as input, which influences its output.
The output is generated from what the model "learned" during its training period where it scanned vast amounts of text.
1/n
sid retweeted
This is mind blowing.
With GPT-3, I built a layout generator where you just describe any layout you want, and it generates the JSX code for you.
W H A T
sid retweeted
You can now log on to ISCA's Whova app to access the recorded videos of keynotes and talks for the workshop. whova.com/portal/webapp/aisc…
sid retweeted
I'm excited to present the latest in @risc_v manycore technology. This is Plenticore: A 4097-core RISC-V SoClet Architecture for Ultra-inefficient Floating-point Computing. The full paper can be read here. github.com/olofk/serv/raw/do…
Hey @hotchipsorg. Got room for an extra talk?
sid retweeted
So excited to share our work on RL for Chip Placement, which enables chip optimization in < 6 hrs, whereas baselines require manual efforts and can take weeks. Joint work with my co-TL @annadgoldie, @JeffDean (cont.)
Paper: arxiv.org/abs/2004.10746
Blog: ai.googleblog.com/2020/04/ch…
sid retweeted
Done at last. We (@vivsze, Yu-Hsin Chen and @TienJuYang) finished our book on "Efficient Processing of Deep Neural Networks" and it is now available for pre-order from @MorganClaypool. The book is scheduled for publication in May 2020. Find out more at morganclaypoolpublishers.com…
Please check @GregDavill OrangeCrab - he is missing only 40 pcs. to start next batch production! And on offer if powerful 25F and 128 Mbytes (1Gbit) DDR3L Memory groupgets.com/campaigns/710-…













