Unverified50% confidenceFactExact time
Gemma 2B模型拥有约20亿参数
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026
Sources
Spring AI + Ollama: Run Large Language Models Locally at Zero Cost
bilibili枫叶边城6/3/2026
Related Claims
UnverifiedGemma 4的31B模型在M2026评测中得分89.2%，代码模型的Elo评分约为250073% similarUnverifiedGemma 4的E2B版本在社区基准测试中多项任务表现堪比Gemma 3的271亿参数模型72% similarUnverifiedGemma 4采用混合专家架构（MoE），E2B后缀表示Effective 2B，每次推理只激活约20亿有效参数72% similarVerifiedGemma 4的26B MOE模型采用混合专家架构，拥有260亿总参数但任意时刻仅约40亿参数处于激活状态71% similarVerifiedGemma系列主要包含2B和7B两个参数规模版本，2025年发布的Gemma 3系列进一步扩展至27B并支持多模态输入70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/46011API
curl https://kongchang.com/api/v1/knowledge/claims/46011MCP
get_claim(id=46011)