DFlash split 3: wire engine draft-verify loop - #25
Draft
dusterbloom wants to merge 1 commit into
Draft
Conversation
This was referenced Jun 23, 2026
dusterbloom
force-pushed
the
codex/dflash-2-drafter
branch
from
June 23, 2026 15:19
d28386a to
3e2f133
Compare
dusterbloom
force-pushed
the
codex/dflash-3-engine-loop
branch
from
June 23, 2026 15:19
9c9dafd to
a27bcb7
Compare
dusterbloom
force-pushed
the
codex/dflash-2-drafter
branch
from
June 23, 2026 17:13
3e2f133 to
f515b86
Compare
dusterbloom
force-pushed
the
codex/dflash-3-engine-loop
branch
3 times, most recently
from
June 24, 2026 12:46
444555b to
bc2d4f8
Compare
dusterbloom
force-pushed
the
codex/dflash-2-drafter
branch
from
June 24, 2026 15:57
f515b86 to
1237cfa
Compare
dusterbloom
force-pushed
the
codex/dflash-3-engine-loop
branch
from
June 24, 2026 15:57
bc2d4f8 to
b377882
Compare
dusterbloom
force-pushed
the
codex/dflash-2-drafter
branch
from
June 24, 2026 16:07
1237cfa to
b12e1ed
Compare
dusterbloom
force-pushed
the
codex/dflash-3-engine-loop
branch
from
June 24, 2026 16:07
b377882 to
64f08f2
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Split from panbanda#204. Adds engine-side DFlash loading,
SimpleEnginedispatch, the draft-verify generation loop (auto-calibrated realized-speedup gate / t_ar EMA / adaptive block + EOS & max-token handling), and the local DFlash benchmark harness.The decode here is the canonical buffered loop. Per-request selection (
Speculation) and DFlash streaming arrive in the stacked split #27, which refactors this loop behind aDflashSinkso buffered and streamed paths share one implementation.Base
Stacked on #24.
Validation
cargo test -p higgs-engine --lib— 287 passcargo clippy+cargo fmt --check— clean