Search Ctrl K
Home / Tools
THE FEUTEX TOOLKIT

Training Sequence Padding Waste

Estimate token compute wasted by padding a batch to a common sequence length.

AI Infrastructure workspace● Runs in your browser
01 Your input
02 Your result
View or copy raw result data
FormulaPadding tokens per batch = batch*(pad-mean); Padding share = 100*(1-mean/pad)

Made for real tasks

  • Estimate token compute wasted by padding a batch to a common sequence length.
Explore step-by-step workflows →

Worked example

Inputs

Sequences in batch
32 sequences
Mean real sequence length
700 tokens
Padded sequence length
1024 tokens

Expected results

Padding tokens per batch
10368 tokens
Padding share
31.640625 %

Load the example in the workspace to reproduce these results. Display rounding does not change the underlying calculation.

How to use this tool

  1. Enter sequences in batch, mean real sequence length, padded sequence length in the displayed units.
  2. Calculate the scenario, then compare it with measurements before changing a system.

Try this example

{
  "batch": 32,
  "mean": 700,
  "pad": 1024
}

Accuracy and scope

Uses mean token count, not quadratic attention work; packing and attention masks may change actual compute.

Is my data uploaded?

This tool runs locally in your browser. Input and output are not sent to a processing server. Copy or download your result before leaving the page.