Open Models Almanac
Methodsinference & servinginference kernel

general family · filed under inference & serving

Batch-invariant deterministic kernels

A family of kernels designed to produce bitwise-reproducible results independent of batch composition.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See inference kernel for the methods filed beside it.

Also called Deterministic and batch-invariant kernels.

sources
2
models
2
lab adopt it
1
strongest
core

Documented in

Evidence

3 quotes from 2 sources, one card per model, each opening with its strongest.

How sources treat it
used 2core 1

DeepSeek-V4-Pro

DeepSeekadopted2 sources

we implement end-to-end, bitwise batch-invariant, and deterministic kernels with minimal performance overhead

coreunclearin DeepSeek-V4DeepSeek
2 more from this model

we implement end-to-end, bitwise batch-invariant, and deterministic kernels with minimal performance overhead

usedsoftware implementationin DeepSeek-V4 seriesDeepSeek

Deterministic and batch-invariant kernels ensure that training is bitwise reproducible.

usedsoftware implementationin DeepSeek-V4DeepSeek

DeepSeek-V4-Flash

DeepSeekadopted1 source

we implement end-to-end, bitwise batch-invariant, and deterministic kernels with minimal performance overhead

coreunclearin DeepSeek-V4DeepSeek
1 more from this model

we implement end-to-end, bitwise batch-invariant, and deterministic kernels with minimal performance overhead

usedsoftware implementationin DeepSeek-V4 seriesDeepSeek
↑