You people need to learn to be a bit more positive. Making money in crypto is not that easy.
Can you help me compare DMD with GNT? 
Topic: [ANN][GNT] Golem Project - The world’s most powerful supercomputer - page 28. (Read 381022 times)
February 23, 2018, 04:30:10 PM
February 23, 2018, 03:26:10 PM
Where Golem has reached from here noway to look at back and only option to keep going on with development though slower but steady. They should not have given hard dates seeing first time they couldn't keep that up. Development in software space is always estimated and approximated.
February 23, 2018, 12:58:39 PM
Teach your Golem how to learn
Ok, so here’s the thing: machine learning is taking over the world.
Like, not in the literal, skynet’ish, space-oddysey’ish, matrix’ish way. At least for now.
How then? Well, in the ordinary, economic’ish way. Data is pouring, algorithms are trained, money is being made.
And since Golem is the future of computing, we would like to be able to run ML tasks on it.
There is, however, a problem. Ordinary machine learning algorithms are, at least at this moment, not trivially parallelizable. They can be run in parallel, but for a large part of them, it is only possible with high-bandwidth, low-latency connections.¹
But, there are algorithms, which can be easily parallelized, and which are crucial in the modern machine learning workflow, even if hidden behind the scenes. I’m talking about hyperparameter optimization stuff.
“What’s that?”, you may ask. Short answer: they are algorithms for optimizing algorithms. Let’s say, we have a toy neural network, with only one layer, for which we want to set a number of hidden neurons. If we set it too high, the network will overfit, if we set it too low — it will not learn as much as it could². Since the network’s training algorithm — gradient descent — doesn’t change the structure of the network, it just optimizes parameters (weights) of the model, we have to find it in some other way. How then?
Well, we can guess. If you are an experienced machine learning engineer or researcher, the guess will probably we quite good. But, that’s obviously not the best way to do it.
Another way is to just check all possible configurations (in some range, since there is an infinite amount of them). This will produce the best possible answer, but will take too long to be practical.
We can combine these two approaches by guessing a number of configurations, based on our intuition, then try them all and select the best.
So, how do we guess them? We can base it on our intuition, we can select them randomly… or we can use one of many hyperparameter optimization algorithms that exist to do this job for us.³
Since “trying out” every configuration in practice means training the whole network with this parameter, and then checking the result, the whole procedure can be easily parallelized.
So, this is our first, proof-of-concept machine learning algorithm working on Golem. We implemented it over the summer, so you can see how it works here. And here⁴ is the full description of the ML task, with experiments justifying various implementation choices.
If you have any comments or doubts, send them to [email protected] and we would be happy to address them.
[1] Better algorithms are developed all the time, but we have yet to see a real distributed gradient descent. However, the prospects are looking good, for example see Lian et al.
[2] Yes, we currently control the overfitting by using other techniques, but this is only a toy example.
[3] In fact, these algorithms usually perform better than random search or even human experts. See Bergstra et al. or Snoek et al.
[4] It is one of our many research resources we made publicly available, to ensure the company and our process of developing Golem are more transparent than ever. We described it here.
https://blog.golemproject.net/teach-your-golem-how-to-learn-e32fc366a976
Ok, so here’s the thing: machine learning is taking over the world.
Like, not in the literal, skynet’ish, space-oddysey’ish, matrix’ish way. At least for now.
How then? Well, in the ordinary, economic’ish way. Data is pouring, algorithms are trained, money is being made.
And since Golem is the future of computing, we would like to be able to run ML tasks on it.
There is, however, a problem. Ordinary machine learning algorithms are, at least at this moment, not trivially parallelizable. They can be run in parallel, but for a large part of them, it is only possible with high-bandwidth, low-latency connections.¹
But, there are algorithms, which can be easily parallelized, and which are crucial in the modern machine learning workflow, even if hidden behind the scenes. I’m talking about hyperparameter optimization stuff.
“What’s that?”, you may ask. Short answer: they are algorithms for optimizing algorithms. Let’s say, we have a toy neural network, with only one layer, for which we want to set a number of hidden neurons. If we set it too high, the network will overfit, if we set it too low — it will not learn as much as it could². Since the network’s training algorithm — gradient descent — doesn’t change the structure of the network, it just optimizes parameters (weights) of the model, we have to find it in some other way. How then?
Well, we can guess. If you are an experienced machine learning engineer or researcher, the guess will probably we quite good. But, that’s obviously not the best way to do it.
Another way is to just check all possible configurations (in some range, since there is an infinite amount of them). This will produce the best possible answer, but will take too long to be practical.
We can combine these two approaches by guessing a number of configurations, based on our intuition, then try them all and select the best.
So, how do we guess them? We can base it on our intuition, we can select them randomly… or we can use one of many hyperparameter optimization algorithms that exist to do this job for us.³
Since “trying out” every configuration in practice means training the whole network with this parameter, and then checking the result, the whole procedure can be easily parallelized.
So, this is our first, proof-of-concept machine learning algorithm working on Golem. We implemented it over the summer, so you can see how it works here. And here⁴ is the full description of the ML task, with experiments justifying various implementation choices.
If you have any comments or doubts, send them to [email protected] and we would be happy to address them.
[1] Better algorithms are developed all the time, but we have yet to see a real distributed gradient descent. However, the prospects are looking good, for example see Lian et al.
[2] Yes, we currently control the overfitting by using other techniques, but this is only a toy example.
[3] In fact, these algorithms usually perform better than random search or even human experts. See Bergstra et al. or Snoek et al.
[4] It is one of our many research resources we made publicly available, to ensure the company and our process of developing Golem are more transparent than ever. We described it here.
https://blog.golemproject.net/teach-your-golem-how-to-learn-e32fc366a976
February 23, 2018, 10:48:12 AM
coding activity is growing so i guess we need to have bit of patience..
https://coincheckup.com/coins/golem-network-tokens/analysis/github
https://coincheckup.com/coins/golem-network-tokens/analysis/github
February 23, 2018, 04:13:35 AM
I am also upset by the fact Golem has failed to respect their own timeline, even amended one. However, what they do is a pioneering work and it s not easy to do something like that.
This is why I have faith and respect for the Golem team. They will deliver and the smart money will be waiting for that.
This is why I have faith and respect for the Golem team. They will deliver and the smart money will be waiting for that.
February 22, 2018, 06:57:48 AM
You people need to learn to be a bit more positive. Making money in crypto is not that easy.
Agreed, making money is not easy but when it is made can easily be done 10 to 20 times the investment.Golem is not for people thinking to make it as fast as possible returns in short term.
February 22, 2018, 03:37:45 AM
You people need to learn to be a bit more positive. Making money in crypto is not that easy.
The last half year it was, that's why most take it for granted.
February 22, 2018, 03:32:20 AM
You people need to learn to be a bit more positive. Making money in crypto is not that easy.
February 22, 2018, 01:36:50 AM
I mean, it is kinda sad, but I cannot help myself
February 21, 2018, 11:57:39 PM
The issue here is that the team has thrown out their timeline, most of the current deadlines are not fixed, a lot of hype was around this project then but nothing real has been delivered by the team over a year after their ICO, which is very interesting.
February 21, 2018, 03:05:21 PM
February 21, 2018, 02:13:16 PM
https://www.reddit.com/r/GolemProject/comments/7xh8rp/welcome_to_the_february_edition_of_golem_teams/
Brass Beta will be released at the end of March, beginning of April.
Brass Beta will be released at the end of March, beginning of April.
February 20, 2018, 10:19:08 AM
Tell me, is there a purse for storing tokens?
You can use mew if you wanna store and forget for long term.. much of news is that eidoo is launching its desktoo wallet soon.
Eidoo is going to come up with a web version too similar to myetherwallet but this is not planned for Q1 as they are currently focussed on exchange and ICO engine and the desktop looks to be having a similar app like features.
February 20, 2018, 09:50:45 AM
Tell me, is there a purse for storing tokens?
You can use mew if you wanna store and forget for long term.. much of news is that eidoo is launching its desktoo wallet soon.
February 20, 2018, 09:46:43 AM
Tell me, is there a purse for storing tokens?
February 20, 2018, 06:32:02 AM
when brass will ready? what are doing team?
I am sure the whole thread is filled only with this one question and there is no answer for this from the past one year or so.I am pissed off with the team communication and feel others are pretty good in this department.
February 20, 2018, 06:10:35 AM
I think this is a situation with most of these ICOs, much promises but have delivered little, I won't go hard on Golem team for now because none of the projects in the cloud computing space have really deliver any product to the space. One thing i think will help the team is the capitalisation, they collect huge amount of Ethers during their ICO, I think the price of Ether was around $7 then and should have enough to work with
February 20, 2018, 05:58:26 AM
They cant fix the date of BRASS because they are rightfully afraid of not to release. And huge dump coming after it with a big disappoint and also disbelief.
I sold my GNT and waiting for some solid development. İf this purgatory continues, the situation of this priceless token is going worse.
I sold my GNT and waiting for some solid development. İf this purgatory continues, the situation of this priceless token is going worse.
February 20, 2018, 05:36:43 AM
when brass will ready? what are doing team?
Information from reddit: "Brass is still estimated to be released early 2018. If a more exact release date is going to be announced, it will be posted here, as well as on the blog, on twitter etc."
Who in a strong drawdown - can only wait that they will fulfill their promises. If this does not happen, they will soon be crushed by competitors.
February 20, 2018, 03:37:51 AM
team has been fooling us for a very long time now, only empty promises is what we have been hearing for a very long time 
Jump to: