Repository navigation
Release 0.64.2: LFM2 decode step for an NPU - #534
Merged
Merged
Conversation
…e 0.63.0 Version 0.64.2 in gradle.properties, the README release paragraph and the BOM snippets in README and docs. CHANGELOG 0.64.2: the decode-step export, the multiply-sum short convolution, exportPorts, and the external-bf16 mode. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Against engine 0.63.0, transformers-only and additive. This ships #533, which the
generator-lfm25-350m-torqblueprint needs for its export step.Added
Lfm2DecodeStep,exportLfm2DecodeStepandLfm2DecodeStepExportCli: a fixed-capacity single-token step with Q8_0 / Q4_0 weights kept packed. LFM2.5-350M compiled with Torq 2.3.0 is one NPU dispatch on an SL2610 at 533 ms per step, with the same top-1 as FP32.--parameters external-bf16: weights as model-scope BF16 parameters, for a GPU that serves them from an archive.ShortConvLowering.MULTIPLY_SUM,DecodeStepExport.exportPorts, andloadLfm2Model(..., weightFormFor)withlfm2ProjectionsAs.Also included from develop: agp 9.5.1 and json-schema-validator 3.0.9 (#528, #531).
Checked locally
:llm-inference:lfm2:jvmTestand:llm-core:jvmTestpass.Release steps: after merge, tag
0.64.2on develop, which triggerspublish.yml.🤖 Generated with Claude Code