Yes, and 1 bit as well.
BRN Discussion Ongoing
- Thread starter TechGirl
- Start date
FrederikSchack
Regular
Yes, I saw that superficially some time ago, but it's still only a proposal from BrainChips perspective. I can also design a test that shows that I'm better than you and you can do the reverse
thelittleshort
Top Bloke
Fullmoonfever
Top 20
Well that was a nice day.
Just catching up.
Amazing what a sanctioned acceptable sensitive Ann can do
Wondering on the 8 bit and if ties in with the suggested industry standard some of the heavy hitters have put together via a white paper or am I on the wrong track?
Would it not be easier & beneficial to have the lower bit flexibility on a client needs basis (still an Akida point of difference in the mkt) whilst also "towing the line" so to speak when it comes to what is being pushed as the standard to come?
Obviously, they will have their own interests in mind when submitting this sort of proposal as a standard however, I guess if you want to be part of the club, need to make it as compatible and interoperable as poss.
Apparently 8 bit benefits transformers networks.
developer.nvidia.com
Discuss (1)
+4
Like
Tags: Arm, featured, News, Transformers
AI processing requires full-stack innovation across hardware and software platforms to address the growing computational demands of neural networks. A key area to drive efficiency is using lower precision number formats to improve computational efficiency, reduce memory usage, and optimize for interconnect bandwidth.
To realize these benefits, the industry has moved from 32-bit precisions to 16-bit, and now even 8-bit precision formats. Transformer networks, which are one of the most important innovations in AI, benefit from an 8-bit floating point precision in particular. We believe that having a common interchange format will enable rapid advancements and the interoperability of both hardware and software platforms to advance computing.
NVIDIA, Arm, and Intel have jointly authored a whitepaper, FP8 Formats for Deep Learning, describing an 8-bit floating point (FP8) specification. It provides a common format that accelerates AI development by optimizing memory usage and works for both AI training and inference. This FP8 specification has two variants, E5M2 and E4M3.
This format is natively implemented in the NVIDIA Hopper architecture and has shown excellent results in initial testing. It will immediately benefit from the work being done by the broader ecosystem, including the AI frameworks, in implementing it for developers.
E5M2 uses five bits for the exponent and two bits for the mantissa and is a truncated IEEE FP16 format. In circumstances where more precision is required at the expense of some numerical range, the E4M3 format makes a few adjustments to extend the range representable with a four-bit exponent and a three-bit mantissa.
The new format saves additional computational cycles since it uses just eight bits. It can be used for both AI training and inference without requiring any re-casting between precisions. Furthermore, by minimizing deviations from existing floating point formats, it enables the greatest latitude for future AI innovation while still adhering to current conventions.
In MLPerf Inference v2.1, the AI industry’s leading benchmark, NVIDIA Hopper leveraged this new FP8 format to deliver a 4.5x speedup on the BERT high-accuracy model, gaining throughput without compromising on accuracy.
By adopting an interchangeable format that maintains accuracy, AI models will operate consistently and performantly across all hardware platforms, and help advance the state of the art of AI.
Standards bodies and the industry as a whole are encouraged to build platforms that can efficiently adopt the new standard. This will help accelerate AI development and deployment by providing a universal, interchangeable precision.
Just catching up.
Amazing what a sanctioned acceptable sensitive Ann can do
Wondering on the 8 bit and if ties in with the suggested industry standard some of the heavy hitters have put together via a white paper or am I on the wrong track?
Would it not be easier & beneficial to have the lower bit flexibility on a client needs basis (still an Akida point of difference in the mkt) whilst also "towing the line" so to speak when it comes to what is being pushed as the standard to come?
Obviously, they will have their own interests in mind when submitting this sort of proposal as a standard however, I guess if you want to be part of the club, need to make it as compatible and interoperable as poss.
Apparently 8 bit benefits transformers networks.
NVIDIA, Arm, and Intel Publish FP8 Specification for Standardization as an Interchange Format for AI | NVIDIA Technical Blog
NVIDIA, Arm, and Intel jointly propose a FP8 format for standardization as an interchange format for AI and improve computational efficiency.
NVIDIA, Arm, and Intel Publish FP8 Specification for Standardization as an Interchange Format for AI
By Shar NarasimhanDiscuss (1)
+4
Like
Tags: Arm, featured, News, Transformers
AI processing requires full-stack innovation across hardware and software platforms to address the growing computational demands of neural networks. A key area to drive efficiency is using lower precision number formats to improve computational efficiency, reduce memory usage, and optimize for interconnect bandwidth.
To realize these benefits, the industry has moved from 32-bit precisions to 16-bit, and now even 8-bit precision formats. Transformer networks, which are one of the most important innovations in AI, benefit from an 8-bit floating point precision in particular. We believe that having a common interchange format will enable rapid advancements and the interoperability of both hardware and software platforms to advance computing.
NVIDIA, Arm, and Intel have jointly authored a whitepaper, FP8 Formats for Deep Learning, describing an 8-bit floating point (FP8) specification. It provides a common format that accelerates AI development by optimizing memory usage and works for both AI training and inference. This FP8 specification has two variants, E5M2 and E4M3.
This format is natively implemented in the NVIDIA Hopper architecture and has shown excellent results in initial testing. It will immediately benefit from the work being done by the broader ecosystem, including the AI frameworks, in implementing it for developers.
Compatibility and flexibility
FP8 minimizes deviations from existing IEEE 754 floating point formats with a good balance between hardware and software to leverage existing implementations, accelerate adoption, and improve developer productivity.E5M2 uses five bits for the exponent and two bits for the mantissa and is a truncated IEEE FP16 format. In circumstances where more precision is required at the expense of some numerical range, the E4M3 format makes a few adjustments to extend the range representable with a four-bit exponent and a three-bit mantissa.
The new format saves additional computational cycles since it uses just eight bits. It can be used for both AI training and inference without requiring any re-casting between precisions. Furthermore, by minimizing deviations from existing floating point formats, it enables the greatest latitude for future AI innovation while still adhering to current conventions.
High-accuracy training and inference
Testing the proposed FP8 format shows comparable accuracy to 16-bit precisions across a wide array of use cases, architectures, and networks. Results on transformers, computer vision, and GAN networks all show that FP8 training accuracy is similar to 16-bit precisions while delivering significant speedups. For more information about accuracy studies, see the FP8 Formats for Deep Learning whitepaper.In MLPerf Inference v2.1, the AI industry’s leading benchmark, NVIDIA Hopper leveraged this new FP8 format to deliver a 4.5x speedup on the BERT high-accuracy model, gaining throughput without compromising on accuracy.
Moving towards standardization
NVIDIA, Arm, and Intel have published this specification in an open, license-free format to encourage broad industry adoption. They will also submit this proposal to IEEE.By adopting an interchangeable format that maintains accuracy, AI models will operate consistently and performantly across all hardware platforms, and help advance the state of the art of AI.
Standards bodies and the industry as a whole are encouraged to build platforms that can efficiently adopt the new standard. This will help accelerate AI development and deployment by providing a universal, interchangeable precision.
Space Cadet
Regular
More money in America
Does it still say trusted by Valeo? I couldn't find it anymoreHas anyone seen new website of BRN it looks stunning.
Home
Unlock the power of AI with BrainChip. Enhance data processing, Edge apps and neural networks at the speed of tomorrow. Explore now!brainchip.com
ndefries
Regular
Has anyone tried to guess the password. It's not Akida2
Fullmoonfever
Top 20
Home page slider of partners.
Which could actually go a bit slower for mine.
Emotion3D, Ai Labs, MB, Prophesee, Valeo etc all there
Fullmoonfever
Top 20
Space Cadet
Regular
equanimous
Norse clairvoyant shapeshifter goddess
If you have been in this forum long enough you would have seen there have been a ton of independent bench marking with Akida in it.Yes, I saw that superficially some time ago, but it's still only a proposal from BrainChips perspective. I can also design a test that shows that I'm better than you and you can do the reverse
Your welcome to research this yourself.
equanimous
Norse clairvoyant shapeshifter goddess
Password is Akida Ballista
Tothemoon24
Top 20
Perhaps the need for a password will be removed or revealed after our ‘formal’ announcement in the morningKen works
Not
Last edited:
It went back to same old verson
Boab
I wish I could paint like Vincent
I have gone back onto the site a couple of times and its obvious they are still trying to smooth it out.
Boab
I wish I could paint like Vincent
you beat me to it.It went back to same old verson
Similar threads
