Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

Introducing Batch Bridge, powered by Relay A tool that allow users to bridge or swaps multiple tokens across Ethereum, Base and Arbitrum in a single batch transaction using EIP5792 and EIP7702 🔗 Tool Link : 🔗 Code : Batching very small value tokens may cause issues, so it is...

60,118 Aufrufe • vor 8 Monaten •via X (Twitter)

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

Continuous batching in LLMs, clearly explained: (a popular LLM interview question; bookmark this) In traditional ML inference, a batch is a matrix. Every input is padded to the same length, one forward pass runs, and every row finishes at the same moment. LLM decoding does not work that way. One forward pass produces one token per sequence, so a request needs as many passes as it has output tokens, and nobody knows that count until the model emits a stop token. Under static batching, membership is fixed when the batch starts. A request that finishes in 30 tokens holds its slot until the slowest request in the same batch finishes at 400. The GPU keeps paying the full weight read for a batch that is mostly empty. Loading model weights out of HBM costs the same whether four slots are producing tokens or one. Continuous batching moves the decision boundary. Instead of scheduling once per batch, the scheduler runs a single forward pass, gets control back, and decides again. A finished request leaves at the next iteration boundary, and a queued request takes its slot right there. No slot stays reserved for work that is already done. Anyscale benchmarked both OPT-13B on a single A100. With uniform generation lengths, the two policies came out about level (as expected), and as output length variance rose, static batching fell to around 81 tokens per second while vLLM reached 23x the throughput of naive Hugging Face serving. Variance drives the entire gap. Production traffic mixes 30-token replies with 400-token ones, which is exactly the condition static batching handles worst. None of this alters the model. vLLM, SGLang, TGI, and TensorRT-LLM all run it by default, and NVIDIA ships the same mechanism under the name in-flight batching. The animation below runs both policies on the same 16 requests and the same 4 slots, stepping in lockstep. The only difference is when a new request is allowed in. To dive deeper into continuous batching specifically, I wrote a full breakdown of the scheduler underneath it. It covers what happens between two forward passes, how tokens get handed out against a fixed budget, why the scheduler needs no separate path for prefill and decode, and what preemption costs you when the KV cache fills up mid-generation. Read it below.

Avi Chawla

16,140 Aufrufe • vor 29 Tagen

The most cursed sat to ever exist 149 cursed inscriptions - All on a single uncommon Satoshi The origins of Wizzy I had been meaning to post about this for awhile but thought I’d wait to see if anyone found it, and recently danny huuep found it and brought it up with ʐǟʀɖ² 🧙‍♂️ Not only is it the Sat with the most cursed inscriptions on it, but we’re pretty sure it’s the first batch reinscription All 150 inscriptions were done in one transaction, costing only $25 Ord.io link: Txn link: Wizzy was done just after batch inscribing was merged into ord (it’s one of the first batch reinscribed collections and pretty sure it was the 1st to be traded on secondary) But batch reinscription still wasn’t possible with ORD client so we did it with raw transactions. A little magic from Vannix The entire collection was done with just 3 transactions (5 if you include the parent and grandparent): 1. Batch reinscribe all 150 traits on a single sat 2. 1st batch of the collection inscribed directly to everyone’s wallets 3. 2nd batch of the collection inscribed directly to everyone’s wallets All possible thanks to recursion and GeneratOrd It’s crazy how much cool stuff is hidden onchain that most of us don’t know about Gonna have a proper look through some more cursed inscriptions this week because there were some pretty cool things done on that final day I know ORDINALLY lifofifo ◉ billy restey dpeaz (🌽,🌽) donny clutterbuck nuro and danny huuep were definitely busy, and I’m sure there’s heaps more that we don’t know about

Sanj 😶‍🌫️

13,133 Aufrufe • vor 2 Jahren

Release Notes – July 10 2025 Today’s update delivers a series of enhancements focused on friction-free cross-chain activity and a smoother mobile experience to Closed Alpha Users. 1. Unified Bridging & Swapping (BRAP) Engine -Best-rate execution: BRAP concurrently scans multiple liquidity routes and automatically selects the most favorable path for simple A → B transfers. -Single-step workflow: Move assets such as USDC (AVAX lending position → USDT on Arbitrum) or DAI on Polygon → Solana-based memecoins without manual routing or multiple approvals. Please ensure you have enough gas to execute the txs you are bridging from. 2. Bonk Launchpad Integration Recent Launchpad graduates are now surfaced directly inside the Transaction Agent, enabling one-tap participation as soon as tokens go live. 3. Flexible Signature Management Choose the security model that fits your workflow: -Session Signatures: Approve a set of transactions for a defined period or transaction count with a single signature. -Per-Transaction Signatures: Require an individual confirmation for every transaction. Switch between the two modes at any time via Settings → Authorizations. 4. Multi-Network Gas Payments via BASE PROMPT Use a single BASE PROMPT balance to cover gas fees on Arbitrum, Polygon, Base, and BNB Smart Chain—no need to maintain separate native-token reserves. 5. Mobile UI Upgrade: Collapsible Sidebar A new sidebar improves navigation on smaller screens while maximizing workspace for the tasks that matter most. 6. Contract Agent Access All free users now have access to the contract agent at a low prompt limit to test out the agent, PRO plan will have a higher prompt limit 7. Perps Agent Access -Trade perpetual futures: trade perps of tokens available on Hyperliquid -Built in BRAP function: bridge and swap from any chain to Arbitrum, and directly into your Hyperliquid perps account -Access: All free users can access perps, spot, and swaps with a lower prompt limit, PRO plan will have a higher prompt limit We value your feedback—reach out anytime via Discord or drop us a DM.

Wayfinder Foundation 🧭

51,650 Aufrufe • vor 1 Jahr