Decoupled Contrastive Decoding (DCD) uses EAGLE3-style expert-aligned drafting to reach 1.65-1.95x speedup over vanilla contrastive decoding. Code is available on GitHub.
DCD: decoupled contrastive decoding hits 1.65-1.95x speedup, code released
2026-08-13Frontier watchWATCH
Decoupled Contrastive Decoding (DCD) uses EAGLE3-style expert-aligned drafting to reach 1.65-1.95x speedup over vanilla contrastive decoding. Code is available on GitHub.
Imported from the existing research archive. Migration preserves the recorded judgment and is not a fresh verification.
What this means for you
For anyone working on inference acceleration, this is a low-cost paper to verify: the method is practical and the code is released.
FRONTIER WATCH
Public artifacts and validated results are different.
The recorded fact is the public work and its artifacts. Author-reported outcomes do not establish independent reproduction. Next checks include usable code, task conditions and third-party validation.
Evidence & related material
Technical details & scores
| Speedup | 1.65-1.95x vs vanilla contrastive decoding |
|---|---|
| Code | github.com/chadlzx/dcd |
Technical impact 3
Engineering value 3
Adoption signal 1
Maturity 2
Verification cost 3
Risk 2