feat(imf): int4 export tier (MatMulNBits) + distill eval fixes - #31
Merged
Conversation
added 7 commits
August 23, 2026 10:47
Qwen3.8-style quantization-ladder move for the client tier: int4 halves int8 again (~246MB -> ~130MB expected for ByT5-small) at some quality cost, measured by the parity gate (int4 limit 3.0pp). Client crystals only — the Ruby gem's bundled ORT cannot execute MatMulNBits; int4 zips must fail loudly on old runtimes, never fall back silently.
…her_volume ara-diac-tiny writes its checkpoints to the secryst volume; the eval hardcoded the teacher's volume and from_pretrained fell through to the hub (repo_type error).
Custom students inherit T5's default max_length=20; 1400-byte windows clamp max_new_tokens to zero and generate raises.
1 task
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Test plan