BRN Discussion Ongoing

Akida mentioned (perfect fit)
Sorry forgot what pages
Need the James Webb telescope to read it.
 
  • Like
  • Fire
Reactions: 4 users
chrome_screenshot_1669087179927.png
 
  • Like
  • Love
  • Fire
Reactions: 39 users

Diogenese

Top 20
Not wishing to type cast myself as the ultimate technophobe this paper all about working with AKIDA confuses me when it says:

“Counting No. of Parameters in Deep Learning Models by Hand by Raimi Karim llustrated Guide to LSTM's and GRU's: A step by step explanation by Michael Phi.

No matter what stage you are in we recommend you to spend just 10 minutes to read the source code of TensorLayer and the Understand layer / Your layer in.

Why do we need to care of calculating number of parameters in LSTM layer since Counting No. of Parameters in Deep Learning Models by Hand by Raimi Karim.”

@Diogenese please can you emerge from your barrel and explain my obvious conundrum and answer the “Why is it so?” Question????

My opinion only DYOR
FF

AKIDA BALLISTA
Hi FF,

You are confusing me with someone who knows stuff.

My forte is knowing where to find stuff:
https://towardsdatascience.com/coun...-in-deep-learning-models-by-hand-8f1716241889

This article by Rami Karim explains that, if you need to compress the arrangement, counting the before and after layers of the NN can be useful.

That said, my serious engagement with software ended at the Fortran IV stage. Since then, I've only dabbled at the edge. You'd be better off applying to @uiux .

By the way, did I tell you about the natty program for characterizing the radiation pattern of a 12 element yagi antenna I wrote in Fortran that largely followed the yet to be created OOP* structure?

*OOP = singular of OOPS.
 
Last edited:
  • Love
  • Like
  • Fire
Reactions: 7 users

SERA2g

Founding Member
  • Like
  • Fire
  • Love
Reactions: 12 users

Boab

I wish I could paint like Vincent
Akida mentioned (perfect fit)
Sorry forgot what pages
Need the James Webb telescope to read it.
Edit Page 9
5.1. Target Hardware StereoSpike has resolutely been developed in the philosophy of spiking neural networks. As a result, it is essentially implementable on dedicated neuromorphic hardware, such as Intel Loihi [5], IBM TrueNorth [2]. These chips can leverage the binarity and sparsity of spike tensors navigating through the network. In addition, we believe that our model being feedforward and requiring a reset on all of its neurons at each timestep is not a problem, because resetting membrane potentials is actually less costly than applying a leak. Therefore, statelessness can be seen as an advantage over recurrence in spiking models with similar performances. However, we are aware that current neuromorphic chips are initially designed for the implementation of stateful units, and acknowledge that we do not leverage this feature. Consequently, we believe that it rather fits to dedicated hardware for stateless models with sparse quantized activations. We therefore consider that Brainchip’s Akida chip [40] is a good fit. As it imposes weights to take at most 8 bit, we quantized StereoSpike’s weights us
 
  • Like
  • Love
  • Fire
Reactions: 25 users

Boab

I wish I could paint like Vincent
In the top 5 for a moment anyway😁
Top 5.jpg
 
  • Like
  • Love
  • Fire
Reactions: 15 users

SERA2g

Founding Member
  • Haha
  • Like
  • Love
Reactions: 11 users
Page 9, bottom right.
Where it says:

“5.1. Target Hardware
StereoSpike has resolutely been developed in the philos- ophy of spiking neural networks. As a result, it is essen- tially implementable on dedicated neuromorphic hardware, such as Intel Loihi [5], IBM TrueNorth [2]. These chips can leverage the binarity and sparsity of spike tensors nav- igating through the network. In addition, we believe that our model being feedforward and requiring a reset on all of its neurons at each timestep is not a problem, because resetting membrane potentials is actually less costly than applying a leak. Therefore, statelessness can be seen as an advantage over recurrence in spiking models with similar performances. However, we are aware that current neuro- morphic chips are initially designed for the implementation of stateful units, and acknowledge that we do not leverage this feature. Consequently, we believe that it rather fits to dedicated hardware for stateless models with sparse quan- tized activations. We therefore consider that Brainchip’s Akida chip [40] is a good fit. As it imposes weights to take at most 8 bit, we quantized StereoSpike using PyTorch natively available post-training static quantiza- tion. The process resulted in an even lighter model with 8 bit wide unsigned integer weights, for the price of a minor performance drop (i.e., MDE of 17.1 cm on indoorflying split 1). Presumably, quantization-aware training would do even better. This demonstrates the efficient deployability on such hardware. Finally, we would like to emphasize that our class of model with sparse binary activations and less constrained weights provides a good compromise between Spiking Neural Networks (SNNs) and Binary Neural Net- works (BNNs).”
 
  • Like
  • Fire
Reactions: 6 users

alwaysgreen

Top 20
Page 9, bottom right.

Hey @Diogenese Are you interested in doing the IT for my business as a side hustle?
Can pay you in BRN shares if interested haha

EDIT: I'm actually serious.
 
Last edited:
  • Haha
  • Like
Reactions: 7 users
Edit Page 9
5.1. Target Hardware StereoSpike has resolutely been developed in the philosophy of spiking neural networks. As a result, it is essentially implementable on dedicated neuromorphic hardware, such as Intel Loihi [5], IBM TrueNorth [2]. These chips can leverage the binarity and sparsity of spike tensors navigating through the network. In addition, we believe that our model being feedforward and requiring a reset on all of its neurons at each timestep is not a problem, because resetting membrane potentials is actually less costly than applying a leak. Therefore, statelessness can be seen as an advantage over recurrence in spiking models with similar performances. However, we are aware that current neuromorphic chips are initially designed for the implementation of stateful units, and acknowledge that we do not leverage this feature. Consequently, we believe that it rather fits to dedicated hardware for stateless models with sparse quantized activations. We therefore consider that Brainchip’s Akida chip [40] is a good fit. As it imposes weights to take at most 8 bit, we quantized StereoSpike’s weights us
Sorry doubled up. 😇
 
  • Like
  • Sad
Reactions: 5 users
  • Haha
  • Like
Reactions: 4 users
But what do you do on a phone for keyword search when "find in page" is not available?
There are a couple of different approaches which I have used:

1. If near a body of water challenge yourself to see how far you can throw it;

2. Swear loudly then sob uncontrollably - swearing can be in any language seems to work equally well in this regard.

Hope these help.🤡😂🤡

Regards
FF

AKIDA BALLISTA
 
  • Haha
  • Like
Reactions: 12 users

wilzy123

Founding Member
  • Like
  • Haha
Reactions: 4 users

equanimous

Norse clairvoyant shapeshifter goddess
This could be of interest to some.


Inquiry into ASIC's capacity and capability to respond to reports of alleged misconduct​


On 27 October 2022, the Parliamentary Joint Committee on Corporations and Financial Services began an inquiry into ASIC’s capacity and capability to respond to reports of alleged misconduct.

The committee will call for written submissions in due course.

The committee currently intends to table a report in both Houses of the Parliament by June 2024.




Committee Secretariat contact:​


Committee Secretary
Parliamentary Joint Committee on Corporations and Financial Services
PO Box 6100
Parliament House
Canberra ACT 2600

Phone: +61 2 6277 3583
Fax: +61 2 6277 5719
corporations.joint@aph.gov.au
The ASIC executives will retire by May 2024
 
  • Like
  • Haha
  • Love
Reactions: 9 users
There should be some function to that effect available.
Pretty sure most PDF files I open don't have that option. When searching a webpage it's always and option for me.
 

Attachments

  • Screenshot_20221122-142313.png
    Screenshot_20221122-142313.png
    979.8 KB · Views: 73
  • Like
Reactions: 1 users

wilzy123

Founding Member
Pretty sure most PDF files I open don't have that option. When searching a webpage it's always and option for me.

Bummer. When I open it using 'Google Drive', I get the option. Can't vouch for other applications tho.

Screenshot_20221122_145925_Android System.jpg
 
  • Haha
  • Like
Reactions: 2 users

HopalongPetrovski

I'm Spartacus!
Just wondering when our next announcement may drop?
Will it be a slow and steady rise following some boring doorbell thingy or will it be another out of the blue, meteoric hellcat whammo surprise like we got in January, courtesy of MB.
Just what will we get first, hmmmm....some cool new AI glasses, a sexy new space Alexa or will some new medical testing devise nose out the boring old PCR and RAT tests ready for the next wave?
Apparently there's about 46000 of us all standing at ground zero, just waiting for the flash. 🤣
Bring it, Brainchip!

 
  • Like
  • Love
  • Haha
Reactions: 26 users
  • Haha
  • Like
Reactions: 4 users

VictorG

Member
Me: Alexa remind me to buy Gin
Alexa: Ok, remind you to go to the gym.

#upgradealexawithakidanow
 
  • Haha
  • Like
  • Fire
Reactions: 21 users
Top Bottom