Expand description
2026-09-25: That table evaluated from the config before the checkpoint loads; preflight’s headroom check reads it. 2026-09-25: The derived-weight bytes of the native-FP8 dense route, predicted from the config and the environment before the checkpoint loads.
Owner: model-arch weight loader (Qwen3.5 dense). Invariants:
predicted_derived_bytesreads no environment and allocates nothing;Fp8RouteInputs::from_envdoes the reads.- A route whose plan keeps an NVFP4 copy (
METRALE_DENSE_FP8_KEEP_NVFP4, a CUTLASS NVFP4 attention lever,METRALE_ATTN_W4A4) getsDerivedBytesEstimate::Unavailable.
The server’s preflight (post_load_yardstick in
serve_phases/preflight/headroom.rs) adds the prediction to the
checkpoint’s on-disk size to estimate the memory in use before the KV
cache. When the prediction is unavailable, it uses pre-load free memory.
Which copies are built is fp8_residency::DenseFp8Plan::resolve; their
sizes are shape arithmetic over the config. The loader tallies NVFP4
copies through DerivedResidency::skip, never keep, so a route that
builds them is not predicted.
Structs§
- Fp8Route
Inputs - 2026-09-25: The route conditions of
load_layersthat are known before the checkpoint loads. - Predicted
Derived - 2026-09-25: The derived bytes the native-FP8 dense loader will keep, by term.
Enums§
- Derived
Bytes Estimate - 2026-09-25: A prediction, or the reason there is none.
Functions§
- predicted_
derived_ bytes - 2026-09-25: Bytes the native-FP8 dense loader will derive on top of the checkpoint, or why that is not predicted. Reads only its arguments.