If you want to support me, the best thing to do is to share out the content :)
If you want to support me financially (completely optional and voluntary, but a lot of people have asked for this): SubscribeStar: www.subscribestar.com/yannickilcher Patreon: www.patreon.com/yannickilcher Bitcoin (BTC): bc1q49lsw3q325tr58ygf8sudx2dqfguclvngvy2cq Ethereum (ETH): 0x7ad3513E3B8f66799f507Aa7874b1B0eBC7F85e2 Litecoin (LTC): LQW2TRyKYetVC8WjFkhpPhtpbDM4Vw7r9m Monero (XMR): 4ACL8AGrEo5hAir8A9CeVrW8pEauWvnp1WnSDZxW7tziCDLhZAGsgzhRQABDnFy8yuM9fWJDviJPHKRjV4FWt19CJZN9D4n
Hey man, I agree with you, lots of money will be made. But real money actually lays in pre-sales, and I think Cyberopolis is one of the best picks for the bull run.
There is one project that stands out and it is in early phases. It is Cyberopolis. 10m already raised from institutions and now public have their opportunity
Waiting after ETF for retrace, then stack some alts. In the meantime, I think the best is to invest in presale projects like Cyberopolis for maximum returns
God bless you all, I hope we all get many x returns by the end of the bull run. I am betting in 100 bucks that Cyberopolis will do 150x by the end of 2024.
You are right. My strategy is to find as many good projects in presale phase as I can. Real money is there. Cyberopolis is my number one pick, and presale is almost over.
Thanks for your video. I'm curious as to why it can outperform alpha-zero on go, because it seems to be training a simulation of an environment model, and the rest of the training may be the same with alpha-zero. The best case scenario is that it trains a simulated environment that matches the real world, but that can only achieve near-alpha-zero results.
since this "licence" has begun propagating, the development of Ai software for general use( especially in the open source arena)has shown signs of slowing down to a crawl. another dirty trick that some unscrupulous vendors will be tempted to employ is malicious code injection with a timed delay so that that, at a predetermined date the model( or software) will trigger a virus or other malware on the end users computer, forcing the user to remove or update the software or model so "solve" the problem,( i've seen this happen with anydiffusion v3 both pruned and full model)
this is VERY cool. but I believe that if given multiple views (image) of a scene, and if we are able to match enough points, it's possible to generate a photogrametric model of the scene (a little bit like stereo vision, but with many more view so the generated model is more complete), and once we have that, we can simply sample or reproject it to any point of view. Isn't that a simpler way of solving the same problem?
I see in paper they use datasets argilla/ultrafeedback-binarized-preferences-cleaned and Anthropic/hh-rlhf, but i don't quite understand how teacher forcing works here with 2 different sequences.
Reading more into parer and I think I got it: they don't add L_or to each token, but rather to whole loss from sft (gathered from generated tokens) and L_or is caluclated as probability over whole chosen and rejected sequences.
6:50 RNN: y(t+1) = sigmoid(y(t) + x(t)) State-space: Well, let’s get rid of non-linear sigmoid so that y(t+1) = y(t) + x(t) = … = y(0) + x(0) + x(1) + … + x(t) Now you change the game lol.
32:00 I think the model could still learn what actions outside the context window it should take to get future rewards. Even with a context window of 1, it could look at its training data and see 2 random agents taking different actions and having different returns to go and from this know what actions are better. If you have enough training data the model should be able to generalize over the environment and make these predictions in general
