Unverified50% confidenceEventExact time
Google DeepMind发布Gemma 4量化感知训练（QAT）CheckPoint，通过定向二比特压缩等技术使Gemma 4 E-to-B模型内存占用压缩至约1GB级别，相关权重已上线Hugging Face
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026
Sources
OpenAI确认系统Bug误封账户，多平台AI工具密集更新
bilibili橘鸦Juya6/6/2026
Related Claims
UnverifiedGoogle DeepMind发布Gemma 4量化感知训练权重，通过定向2Bit压缩技术将12B参数模型的内存占用降至约1GB81% similarUnverifiedDeepSeek-V2引入Multi-Head Latent Attention（MLA）机制和DeepSeekMoE架构，其每百万token推理成本约为同期GPT-4级别模型的百分之一72% similarUnverifiedGoogle DeepMind released Gemma 4 Quantization-Aware Training (QAT) checkpoints along with a new mobile quantization format68% similarUnverifiedGoogle DeepMind在多模态模型的GUI理解和交互方面持续投入，包括利用大规模网页截图数据训练模型以及探索Android设备操控能力68% similarUnverifiedGoogle DeepMind提出了Nested Learning（嵌套学习）相关方法67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/53826API
curl https://kongchang.com/api/v1/knowledge/claims/53826MCP
get_claim(id=53826)