fix(imf): cut greedy-KV generation when the token window cycles - #61
Merged
Conversation
Live failure: short inputs (كتاب) made the int8 students echo the same phrase until max_len. The guard stops generation when the last 24 tokens repeat verbatim earlier in the output; goldens and parity unaffected (90/90).
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Flat-byte student models loop under greedy KV decode on short inputs
"input":"كتاب"} returned the phrase echoed repeatedly until max_len.
onnx_greedy_kv now stops when the trailing 24-token window repeats
verbatim earlier in the generated sequence. Normal generation is
untouched (dedicated no-loop test), and the full suite — including
the WO03 parity gate over real exports — passes 90/90.