Chip Placement with Deep Reinforcement Learning (Paper Explained)

  Рет қаралды 11,705

Yannic Kilcher

Yannic Kilcher

Күн бұрын

Пікірлер: 24
@aleks1980
@aleks1980 4 жыл бұрын
Love how heavily you have been uploading, thank you for all the work, really great to learn from you.
@roman6575
@roman6575 4 жыл бұрын
In case, anybody is interested, wire routing congestion occurs on a system-on-chip when a lot of wires (or metal lines) are routed in a narrow space.
@brunomartel4639
@brunomartel4639 4 жыл бұрын
what's wrong? some kind of interference in the flow? emf?
@bentsherman
@bentsherman 3 жыл бұрын
@@brunomartel4639 yeah if two wires go in parallel with each other for enough distance they will interfere with each other, they call it crosstalk
@mohammadjavadnouri2623
@mohammadjavadnouri2623 4 жыл бұрын
Hi, does it have and code implementation or dataset? I could not find any in their source paper.
@herp_derpingson
@herp_derpingson 4 жыл бұрын
Can you do a dedicated video on Graph Conv? So far I havent seen any consensus about the best way to do graph conv.
@YannicKilcher
@YannicKilcher 4 жыл бұрын
probably because there is no consensus :D I might give it a try
@UncoveredTruths
@UncoveredTruths 4 жыл бұрын
the graph stuff is somewhere between a meme and cutting edge
@cunningham.s_law
@cunningham.s_law 4 жыл бұрын
You should make livestream where you try to reproduce paper.
@sahiluppal3093
@sahiluppal3093 4 жыл бұрын
We are entering the era of Fine tuning.
@mamalamini3960
@mamalamini3960 3 жыл бұрын
Does anyone know if there are simulators for this chip placement anywhere ?
@ribbonwing
@ribbonwing 3 жыл бұрын
"How amazing!" *Laughs* "Yeah humanity is doomed..." XD
@doublevgreen
@doublevgreen 11 ай бұрын
thanks for the explanation this is very interesting
@sphereron
@sphereron 4 жыл бұрын
Hey Yannic, could you cover Network Deconvolution? The paper seems like it's going to be a foundational method in a couple years replacing batch norm
@YannicKilcher
@YannicKilcher 4 жыл бұрын
I have a hard time believing single papers that make claims like this. The problem isn't that ND might be better, but that it would have to be substantially better across the field for people to switch over
@Johnathanaa7
@Johnathanaa7 4 жыл бұрын
cranking these out
@thirugnanamv9977
@thirugnanamv9977 4 жыл бұрын
Hi Yannic great. can you please explain more deep reinforcement learning papers on real world problems. Pleaaaseeeeeeee.
@sitioprueba2855
@sitioprueba2855 2 жыл бұрын
thanks for such great content!
@bhuvaneshs.k638
@bhuvaneshs.k638 4 жыл бұрын
I wanted more information on converting netlist to graph representation (actual implementation procedure)
@rohanmishra2415
@rohanmishra2415 4 ай бұрын
hello, I am new to this field can you help me with this query if you could clear it ?
@yumendy6502
@yumendy6502 3 жыл бұрын
Thank you!
@pooglechen3251
@pooglechen3251 4 жыл бұрын
Can we get something similar for ... SimCity :)
@YannicKilcher
@YannicKilcher 4 жыл бұрын
Nice idea 😁 cities would look amazing
@wahabfiles6260
@wahabfiles6260 4 жыл бұрын
Singularity would mean the computer even decide which components to use as well and not just arranging the "given" components. Its way far far from the singularity.
Google’s Chip Designing AI
13:18
Asianometry
Рет қаралды 192 М.
Can You Find Hulk's True Love? Real vs Fake Girlfriend Challenge | Roblox 3D
00:24
Reinforcement Learning with Augmented Data (Paper Explained)
22:15
Yannic Kilcher
Рет қаралды 7 М.
The moment we stopped understanding AI [AlexNet]
17:38
Welch Labs
Рет қаралды 1,3 МЛН
Concept Learning with Energy-Based Models (Paper Explained)
39:29
Yannic Kilcher
Рет қаралды 32 М.
Reinforcement Learning, by the Book
18:19
Mutual Information
Рет қаралды 106 М.
The Impact Of Machine Learning On Chip Design
17:03
Semiconductor Engineering
Рет қаралды 4,5 М.
An introduction to Reinforcement Learning
16:27
Arxiv Insights
Рет қаралды 661 М.
Reinforcement Learning Series: Overview of Methods
21:37
Steve Brunton
Рет қаралды 102 М.
Has Generative AI Already Peaked? - Computerphile
12:48
Computerphile
Рет қаралды 1 МЛН