Repository navigation
Pin SparkGLM fabric and refusal-content maintenance - #70
Merged
Merged
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Pin the published SparkGLM maintenance image so fresh installs retain full blocking assistant content when Atlas classifies a refusal. The image includes the existing per-node RDMA configuration and ambiguity checks.
Update source/image provenance, documentation, and exact pin assertions. Preserve the model and DFlash checkpoint revisions, 0.91 utilization, display carveout, profile settings, and all SparkQwen sidecar settings.
Validation: full
npm test,check,format:check,lint,interchange:check, andpackage:checkpassed. The image's native arm64 CUDA build passed all configured gates, including 25 refusal and 43 streaming tests. Product PR: Enntity/sparkglm#49. Live GLM activation awaits a coordinated window around the separate MTP experiment on 01/02.