Google公布解決幻覺的AI模型DataGemma
支付動態 · 2024-09-16

為了解決模型生成錯誤資訊的幻覺問題,Google利用檢索交錯生成(RIG)以及檢索增強生成(RAG)技術,讓開源模型Gemma結合開源知識圖譜Data Commons,發展出查證能力更好的DataGemma模型

Google

第二種方式中,當用戶查詢DataGemma時,它會運用Gemini 1.5長脈絡空間,先從Data Commons取得上下文資訊,再生成回應,藉此減少幻覺、提升回應品質。

Google之前已利用開源的Gemma和Gemma 2為基礎,使用RIG和RAG微調出變種版本,而這次則是以Gemma 2為基礎,再以RIG及RAG方法,分別生成DataGemma 2個27B版本變種,現已分別在Hugging Face平臺公開。Google計畫持續改善方法,最後會將這些方法學整合到開源的輕量模型Gemma,以及Gemini家族模型。一開始將局部開放給少數人試用,再階段性逐步擴大開放。

Popular articles
Super PAC Raises $48 Million: Sports Betting Forces Ramp Up Political Push
Regulation
Online gambling, crypto pose ongoing money laundering risks in Philippines, analyst says
Southeast Asia
New Jersey July Gambling Revenue Hits $606M, Sweeps Casinos Banned
Regulation
Indiana online casino bill stalls in House committee
Regulation
PropellerAds Shared a New iGaming Case Study: 97,674 Installs and 12,701 Deposits in 3 Months
Marketing
HUIDU Invites You to Booth T70 at iGB L!VE 2026 — Let’s Ignite London This July!
HUIDU Focus
SBC Summit Canada to Make Player Safety a Key Pillar of 2026 Agenda
Marketing
British gambling levy rates confirmed for each vertical
Regulation
GAT Expo Puerto Rico Will Pulse with the New Era of Gaming in the Caribbean
Marketing
Full House at GAT Expo Cartagena 2026 Academic Agenda
Online Game
Kazakhstan plans to penalise online casino promotions
Regulation
Gaming & Technology Expo Makes a Powerful Entrance in CDMX
Marketing
GGC Awards 2026 Shines in Colombo: Honoring Leaders and Innovators in the iGaming Industry
HUIDU Focus
Brazil Proposes Raising Gambling Tax Rate to 24%, With Revenue Allocated to Social Security and Healthcare
Regulation
Vietnam’s Controlled Gaming Shift Gains Ground, But Domestic Demand Still Lags
Southeast Asia
Home
Game
Cooperation
Find
My