MemoriaDA
103 posts
@MemoriaDA_
The permanent, decentralized memory standard for the AI Agent economy. Powered by @0G_labs Apps built on MemoriaDA: @AlphaJournal_ & SolTutor.
Google I/O hits Tuesday. Likely another Gemini, more users on servers they can't see into. The question isn't whether centralized inference scales. It's whether anyone's building the version where the operator can't read your prompts. 0G Private Computer already does.
MemoriaDA is officially live on @0G_labs Mainnet. The first decentralized memory protocol for AI agents - permanent, verifiable, and owned by no one. Every conversation your agent has is embedded, stored on 0G Storage, and anchored onchain forever. AI amnesia ends today. 🔗 memoriada.xyz #0GHackathon #BuildOn0G @0g_CN @0g_Eco @HackQuest_ 🧵👇
MemoriaDA is officially live on @0G_labs Mainnet. The first decentralized memory protocol for AI agents - permanent, verifiable, and owned by no one. Every conversation your agent has is embedded, stored on 0G Storage, and anchored onchain forever. AI amnesia ends today. 🔗 memoriada.xyz #0GHackathon #BuildOn0G @0g_CN @0g_Eco @HackQuest_ 🧵👇
Building apps used to require code, cloud accounts, and centralized providers watching your every prompt. That era just ended. Meet 0G App, the ultimate hub for decentralized AI. Prompt to app. Verified inference. 🐼 Try the Desktop App now → app.0g.ai
MemoriaDA is officially live on @0G_labs Mainnet. The first decentralized memory protocol for AI agents - permanent, verifiable, and owned by no one. Every conversation your agent has is embedded, stored on 0G Storage, and anchored onchain forever. AI amnesia ends today. 🔗 memoriada.xyz #0GHackathon #BuildOn0G @0g_CN @0g_Eco @HackQuest_ 🧵👇
训练在此，运行在此，支付在此。 0G 首个自研主权 AI 模型 0GM-1.0 正式上线。我们告别了对他人的 API 依赖，实现了从算力、微调到推理的全栈自主化。 0GM-1.0 在 AIME 数学等推理榜单上表现优异，并坚持 Apache 2.0 协议彻底开源。这是 0G 的里程碑，也是主权 AI 的新起点。
Just migrated MemoriaDA's entire AI inference stack to the latest @0G_labs Compute Router API; now running the 0GM-1.0-35B-A3B reasoning model natively on 0G. Zero dependency on OpenAI or any external provider. During integration, we discovered that 0GM is a chain-of-thought reasoning model that separates output into reasoning_content and content; if max_tokens is too low, all tokens get consumed by reasoning and content comes back null, crashing the frontend. We patched our backend with a graceful fallback that extracts the draft answer from the reasoning chain instead of failing. MemoriaDA now runs fully native on the 0G stack: Storage, Chain, and Compute. .. @0g_CN @Jtsong2 @dragon0195
