Unverified50% confidenceFactExact time
128K上下文窗口的注意力矩阵规模是32K窗口的16倍
1
Sources
50%
Confidence
Long-term
Relevance
7/8/2026
First Seen
Sources
Context Warp Drive: A Deep Dive into Deterministic Context Folding for AI Agents
hackernewshackernews7/7/2026
Related Claims
Unverified128K长度的序列所需的注意力计算量是4K长度序列的1024倍73% similarUnverifiedIn a 170k token scenario, the attention score matrix requires approximately 54 GB of VRAM in BF16 precision (170,000 × 170,000 × 2 bytes).73% similarUnverifiedGPT-4的上下文窗口从8K扩展到128K乃至百万级别，上下文窗口越大每次调用消耗的令牌越多70% similarVerifiedThe context windows of current mainstream models are approximately 128K to 1 million tokens.70% similarUnverifiedAt a sequence length of 4096, the N×N attention matrix alone consumes approximately 64 MB of memory in FP16 precision.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/294258API
curl https://kongchang.com/api/v1/knowledge/claims/294258MCP
get_claim(id=294258)