Repository navigation
Conversation
nxphi47
marked this pull request as ready for review
October 14, 2026 19:43
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Add registered Triton per-32 MXFP8 quantization and BF16 dequantization, with E4M3 payloads and UE8M0 RCEIL scales. Support strided dense/expert weights, signed zeros, empty tensors, and deterministic stochastic rounding keyed by seed, tensor stream, and global element offset.
Validation: 14 GPU component tests passed, covering all finite BF16 magnitudes against the native vLLM cast, an independent exact scale reference, Philox/neighbor-reference SR bytes, shard invariance, changing streams, on-grid invariance, and unbiased midpoint rounding.
The serving compatibility layer and QAT integration are tracked separately in PrimeIntellect-ai/prime-rl#4095.