Skip to content
KernelIndex
Search⌘K

Top k sampling from probs v200064

0 eligible runs
sampling

Top-k sampling from probabilities with vocab_size=200064. Keeps only the k highest probability tokens, renormalizes, then samples from the filtered distribution. Captured from MiniMax M2.

Current records

No published measurement for the selected workload.

Implementations

Implementation
Runtime
Best latency
Evidence
Availability

Semantics

Inputs and outputs
probsfp32 [batch_size, vocab_size]
top_kint32 [batch_size]
samplesint64 [batch_size]
Axes and behavior
batch_sizevariable
vocab_sizeconstant = 200064
determinismunspecified
constraintsNo mutation or aliasing
Identity
aliastop_k_sampling_from_probs_v200064modelminimax-m2sha256aa6f1b848751…
No source imports for this operation yet.How records are decidedJSON