Unverified50% confidenceBenchmarkExact time
The total encoding and recovery time of guarded compression is 2.3-2.6 times longer than guarded BF16, but shorter than guarded block-wise 8-bit.
1
Sources
50%
Confidence
Long-term
Relevance
11/3/2026
First Seen
Sources
Related Entities
Related Claims
Unverified动态GGUF的UD-IQ2_XXS档位将原始BF16模型体积从54.7GB压缩到约9GB，体积压缩率83.5%，准确率保留82.5%66% similarUnverified从BF16到FP8内存占用减半66% similarUnverified13B–14B class models (4-bit quantization) see speeds drop to 2–5 tokens per second under this configuration66% similarUnverified2-bit量化意味着每个权重仅用4种状态表示，相比FP16压缩比高达8倍65% similarUnverifiedComparing BF16 and NVFP4 under the same harness, WebVoyager dropped from 77.72 to 66.31, DeepSWE from 24.65 to 18.2, and MMLU-ProX from 76.91 to 73.76, with other rows within the noise range.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1129381API
curl https://kongchang.com/api/v1/knowledge/claims/1129381MCP
get_claim(id=1129381)