Repository navigation
Conversation
Disable chunked prefill because no-ID multi-token forwards reset model-owned RGLRU state while preserving stable-ID continuation. Add a loaded one-layer regression that compares reset prompts with a fresh model and stable-ID split execution with one-pass logits. Focused CUDA validation is running under the orchestrator. Refs #2288
…ent-gemma-chunk-contract
inureyes
added
status:review
Preserve the newly merged MLX pin and overlay baseline declarations before validating the model prefill correction against current main. The model-specific implementation and synthetic regression are unchanged.
This branch has not been deployed
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
State and memory contract
Current DirectEngine prefill carries one stable sequence ID, so its state handling does not enter the no-ID reset. The capability is global, however, so returning
falsemakes every honoring caller, including DirectEngine, use a whole-prompt graph. Long prompts can therefore use more peak transient memory than chunk-bounded prefill.Test plan
test-fastbuild graph: passed in 14 minutes 25 secondsrecurrent_gemma_focused suite: 5 passed in 0.84 seconds, including the loaded-model contract and existing model-owned state testsgit diff --checkCloses #2288