Repository navigation
SKEEP-008: Precision and encoding conversion (Draft) - #1434
Merged
michalharakal merged 1 commit intoOct 21, 2026
Merged
Conversation
Proposal for the mixed-precision DSL stubs (dense<TLayer>/stage<TStage>, six TODO()s in NetworkBuilder.kt) and TensorStorage.repackTo, which throws for every encoding change. The DSL takes the dtype as a value behind one PrecisionBoundary module built on TensorOps.convert; repackTo stays lossless and a new requantizeTo does the lossy direction through Requantizers. The unused MixedPrecisionModule scaffold is retired. Tracking issue #1433; the stub allowlist now points at it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
1 of 3 tasks
|
📖 Documentation Preview The documentation has been built successfully for this PR. Generated Files:
Artifacts:
This comment will be updated automatically when the PR is updated. |
aharakal
approved these changes
Oct 21, 2026
michalharakal
deleted the
feature/skeep-008-precision-and-encoding-conversion
branch
October 21, 2026 13:15
michalharakal
added a commit
that referenced
this pull request
Oct 21, 2026
CHANGELOG [0.62.1] - 2026-10-21 over everything develop holds since 0.62.0: the FFM Q8_0/Q4_0/Q5_0/Q5_1 binding fix (#1436) as the headline, the YOLO move-out (#1437), the StableHLO softmax axis and logSoftmax fixes, the compile-flexibility explanation (#1432) and the SKEEP-008 draft (#1434). VERSION_NAME, CITATION, Antora version, README bom coordinate, What's New and contributors, Android tutorial coordinates; regenerated operator and kernel-support matrices.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Adds SKEEP-008: Precision and encoding conversion at
Status: Draft. Tracking issue: #1433.Covers the two stub areas the op-coverage inventory (#1370) left without an issue, because both change public API semantics:
dense<TLayer>()×2 andstage<TStage>()areTODO()in both DSL implementations, and they can't be implemented as declared: an interface method's type parameter can't bereified, so there's noKClassfor the layer. Proposal: take the dtype as a value (dense(64, dtype = FP16)) behind onePrecisionBoundarymodule built on the existingTensorOps.convert, which is already traced and lowers tostablehlo.convert. Inference first;convertbecomes@Diffin a later phase.TensorStorage.repackTothrows for every encoding change. Proposal:repackTostays lossless (packed → dense, narrow → wide dense), and a new, explicitly lossyrequantizeToreusesRequantizers, so it accepts exactly what loader quantize-at-load accepts.MixedPrecisionModule/InternalMixedPrecisionModule/MixedPrecisionTensorOpsscaffold. Nothing subclasses or implements it.Docs only, plus the
config/stub-allowlist.txtreason now pointing at SKEEP-008 / #1433. Registered innav.adocand the Current Proposals table.Open questions for review are in the proposal:
reifiedsugar, cast-pair elision, whererequantizeToshould live, andDTypePolicymetadata on boundaries.🤖 Generated with Claude Code