Video wird geladen...
Video konnte nicht geladen werden
5 Quant explainers. 1 map. 🗺️ Overledger, Fusion, PayScript, Quant Flow, SATP: I covered each one alone. This video puts them together, layer by layer, plus the 3 places $QNT shows up. 10 min, straight from the docs 🧵👇 Which layer do you know least about?
68,244 Aufrufe • vor 13 Tagen •via X (Twitter)
15 Kommentare
The “approved providers” behind each pool are names you know: Alchemy, Infura, QuickNode. And the pick is a weighted score on every call, not a fixed order. Outside the six self-serve families, the networks Quant’s team wires up include XRPL, Stellar and even Bitcoin.
Small print: a leaked API key still reads your data and burns your quota; it just can’t move funds. Keys look like fk_…, up to 10 per account, each revocable. And Fusion’s on-chain firewall is strict on mainnet but permissive on testnet.
That 3.5-day wait isn’t an estimate: it’s 302,400 seconds, set in Fusion’s OptimismPortal contracts on testnet (Sepolia and Amoy). The mainnet window hasn’t been published. One more for the trust model: it’s non-custodial, but Quant itself operates the sequencer.
For agents, the point is parity: the same typed steps a person clicks through, with the risk level attached. x402 is an open standard ( here the permit is Permit2 and prices are ERC-20, so EVM only for now. So far, every published Flow App is hosted by Quant.
PayScript’s language spec is public on GitHub (June 2026): you can read exactly what reserve and lock do. The money stays the bank’s: Quant sells the rails, the bank issues the deposit. And for businesses, Quant Flow links accounts via Open Banking through Yapily.
Where SATP is right now: IETF Last Call ran from 25 Sep to 9 Oct. Next, the IESG evaluates it. A transport review on 4 Oct came back “Not Ready”. SAEP is still an individual draft (informational), not adopted by the working group.
Caveats on QNT, from the docs: staking also sets capacity tiers, but its chapter still waits on the testnet flow in Quant Connect. QNT as Fusion gas comes from the pre-June docs (testnet); today’s docs don’t name a gas token. The three journeys are illustrative composites.
What to watch next: Fusion mainnet opening beyond private access (the API reference still lists only testnet) The IESG’s decision on SATP The Clearing House network, linked to RTP and CHIPS, opening to participants in H1 2027
Sources: Overledger docs, PayScript spec (Jun 2026), IETF datatracker, Quant (2025–26), Business Wire, The Clearing House, UK Finance (Sep 2026). Unofficial explainer · not affiliated with Quant Network · not financial advice.
Full explainer:
Mate these are really good , i watched your last ones , really informative and well put together in a simplistic way ! Good work sunshine and thankyou for taking the time 🇬🇧
Thank you. I was trying to find out if the XRPL and Stellar would be connected some how.
What can I say? Mama Mia ,Bravo ,Great ? No this is not a words that describle what you doing. This word not exist in dictionary. Dont't stop your job is priceless and very motivating even for someone like me. Thank you.
Wow, thank you so much. This is art.
Yikes
![Backpropagation by hand ✍️ ~ 11 steps walkthrough below Backpropagation is the algorithm that actually trains a neural network, and it is where most people stop following along. It is not calculus you cannot do. It is matrix multiplication, working backward, one layer at a time. So I drew and calculated one entirely by hand. Goal: push the loss gradient back through a 3-layer network and land on a new value for every weight and bias. = 1. Given = A 3-layer perceptron, an input X, predictions Ypred = [0.5, 0.5, 0], and the truth Ytarget = [0, 1, 0]. = 2. Backprop gradient cells = Let us draw empty cells for every gradient we are about to compute. The shape of the answer comes first. = 3. Layer 3 softmax = We get dL/dz3 straight from Ypred minus Ytarget = [0.5, -0.5, 0]. No chain rule needed, and that shortcut is the whole reason softmax and cross-entropy are paired. = 4. Layer 3 weights and biases = Let us multiply dL/dz3 by [a2 | 1]. One multiplication gives the gradient for W3 and b3 together. = 5. Layer 2 activations = We multiply dL/dz3 by W3 to get dL/da2. The gradient moves back across a layer the same way the signal moved forward. = 6. Layer 2 ReLU = Let us pass it through the gate: keep the gradient where the activation was positive, zero it everywhere else. = 7. Layer 2 weights and biases = We multiply dL/dz2 by [a1 | 1]. The same figure as step 4, one layer up. = 8. Layer 1 activations = Let us multiply dL/dz2 by W2. = 9. Layer 1 ReLU = We apply the same gate again, now on a1. = 10. Layer 1 weights and biases = Let us multiply dL/dz1 by [x | 1], and every weight in the network now has a gradient. = 11. Update = We subtract, and the network has learned. In practice a learning rate scales this step. The gradients: dL/dz3 = [0.5, -0.5, 0] dL/da1 = [1, -2, 2, -1] dL/dz1 = [0, -2, 2, -1] The takeaway: matrix multiplication is all you need. Just like the forward pass, backpropagation is matrix multiplications end to end. You can do every one by hand, slowly and imperfectly, which is exactly why a GPU's ability to do them fast mattered so much to deep learning. 💾 Save this post!](https://image.24vids.com/tw-2079208495094677511/ext_tw_video_thumb/2079208483669315584/pu/img/dDD0x6lHYHuV1POQ.jpg)
