Unverified50% confidenceFactExact time
Through targeted two-bit compression techniques, the Gemma 4 E-to-B model's memory footprint has been compressed to approximately 1GB
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026
Sources
Related Claims
Unverified模型量化技术将 Gemma 2B 的内存占用从约 5GB 降至 1.5GB 左右81% similarUnverifiedAfter Gemma 4 QAT optimization, the minimum memory footprint can be reduced to 1GB79% similarVerified4-bit量化将每个参数压缩至0.5字节，理论上将140GB压缩至约35-40GB75% similarVerified4-bit量化（如Q4_K_M）通常可将3B模型内存占用从约6GB压缩至约2GB，精度损失低于2-3%75% similarPartially Verified标准的FP32模型每个参数占用4字节，而INT4量化将其压缩至0.5字节，理论上可实现8倍的内存节省74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/53680API
curl https://kongchang.com/api/v1/knowledge/claims/53680MCP
get_claim(id=53680)