Glossary · term163 of 202
RTK-ML ranked compression
A machine-learning variant that ranks tool-output tokens by estimated importance before removing the lowest-ranked ones, tested against a rule-based baseline in the same study.
also written as rtk-ml
Related terms
RTK rule-based compressionA fixed, rule-driven method for stripping tokens from tool output before it reaches the model, one of the compression arms tested in a large paired-cost study.view termheadroom proxy compressionA compression arm that estimates how much a payload can shrink using a proxy signal rather than measuring actual cost impact, and which raised paired cost the most in testing.view term
At a glance
- cited by
- 1 post
- categories
- 1
- first used
- jul 2026
