Openledger
@OpenledgerHQ
Followers
369K
Following
8K
Media
729
Statuses
5K
Openledger is the AI Blockchain, unlocking liquidity to monetize data, models and agents | Backed by @polychain, @borderless_cap & @HashKey_Capital
San Francisco, CA
Joined February 2024
OpenChat is powered by you, and built to give you credit. Every message you send, dataset you share or model you fine-tune is logged on-chain. With Proof of Attribution, your contributions aren't just remembered. They're rewarded š
880
309
2K
Core Contributor @Ramkumartweet spilled some alpha with @0xAbhiP.
Uncovered Ep.1 with Openledger. Monetization and ownership in AI remain significant pain points. Openledger is addressing these issues through its decentralized AI network, building proof of attribution to empower creators to deploy, scale, and monetize AI models. I flew out to
220
60
439
RT @nthispeaks: Let's start a riddle style story for @OpenledgerHQ š. I'll begin with a line, you quote or comment with your own continuatiā¦.
0
23
0
OpenLedger is trending. AGAIN?!.Seriously, who let the community cook this hard? š. Why do YOU think weāre trending right now?.Funniest answer wins the internet today. š
878
208
2K
OpenLedger isnāt just another blockchain. Itās the AI chain rewriting the rules:. šYou own what you build.šAttribution is built from the ground up.šModels are modular & swappable.šContributors get credit (and rewards).š Deployed with OpenLoRA (cheap, forkable, and onchain).
794
287
2K
RT @chainofseoul: šŗ We sat down with @Ramkumartweet, core contributor @OpenledgerHQ, for a Seoul-flavored conversation. We even found outā¦.
0
24
0
Excited to partner with @AlloraNetwork to bring specialized intelligence to self-improving AI agents. By integrating our specialized data and fine-tuned models, Alloraās agents will gain access to trusted, domain-specific knowledge, enhancing both precision and performance
607
273
1K
Hundreds of ideas. Millions in funding. One mission, build the AI x Web3 stack of the future. We launched OpenCircle SeedLab with $25M to support devs, researchers, and founders shaping onchain AI. If youāre building, weāre backing š.
Hundreds of ideas have already been pitched to OpenCircle SeedLab, a $25M value initiative by OpenLedger to fund the future of AI x Web3. Weāre backing builders creating onchain models, agents, and apps that push the boundaries of what's possible at the intersection of AI and
544
194
1K
RT @0xAbhiP: Flew out to Shanghai to chat with @Ramkumartweet, core contributor @OpenLedgerHQ. Openledger will be live on Mainnet soon, tesā¦.
0
57
0
Snappers, youāve been absolute legends lately šŖ. In just two weeks on @cookiedotfun:. š 14.5K+ snappers.š 2% mindshare.š 94.33K sentiment spike.š 6.87K snaps sent out. Itās time we show some love back. If you own OpenLedger snaps, flex it. Post your Snaps with.
1K
313
2K
Infini-gram is the foundation of OpenLedgerās Proof of Attribution system. By linking model outputs back to training data in real time, making AI accountable - and attribute the people who helped build it.
49
41
243
Infini-gram works using a suffix-array-based ā-gram attribution framework. It indexes every sequence from training data - no matter how long - and can instantly trace any output phrase back to where it appeared. This lets us compute influence scores without gradients and with
65
56
308
At OpenLedger, Infini-gram traces which training data shaped a modelās response - token by token. It identifies high-confidence matches across large datasets, allowing us to say, this answer came from this contributorās data. And it works without touching model internals -
15
1
83
The problem? Real language isnāt just fragments. n-grams donāt scale well to long context. They miss influence beyond their fixed window. Enter Infini-gram - a symbolic span-matching engine that goes beyond n-gram limits. It doesnāt just look at fixed lengths. It finds
10
3
77
An n-gram is a sequence of n words that helps models remember and predict language. ā¢ uni-gram (1 word): āChestā ā used in models like Naive Bayes. ā¢ bi-gram (2 words): āChest painā ā used in early translation systems. ā¢ tri-gram (3 words): āChest pain orā ā common in
16
3
92
How OpenLedger powers Proof of Attribution with Infini-gram:. Traditional Language models use n-grams to predict the next word from short patterns - fast, but blind to context. They canāt tell where predictions came from, or who contributed the data. Infini-gram enables precise
518
277
1K
RT @0xAbhiP: I flew out to Shanghai to interview @Ramkumartweet, core contributor @OpenledgerHQ . OpenLedger is a decentralised AI networkā¦.
0
67
0