Unverified50% confidenceFactExact time
Disk usage grows quadratically (O(n²)) because the checkpointer writes the entire conversation state at every superstep
1
Sources
50%
Confidence
Long-term
Relevance
10/30/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe KV Cache size of sliding window attention is constant, achieving O(1) memory usage rather than O(n).66% similarUnverifiedFor larger applications, a full clone may mean doubled disk usage.65% similarUnverifiedOn a machine with 2 GB of memory, what actually causes OOM is often not average memory usage, but the sudden spike in peak RSS during a particular rendering or decoding step.65% similarUnverifiedThe computational complexity and memory consumption of the self-attention mechanism are both O(n²); when the input increases from 1K to 10K tokens, the amount of computation increases approximately 100-fold64% similarUnverifiedThe strategy of substituting disk for VRAM performs much worse for dense models, because dense models require reading all parameters at every layer, multiplying the I/O pressure.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1113684API
curl https://kongchang.com/api/v1/knowledge/claims/1113684MCP
get_claim(id=1113684)