Repository navigation
Follow-up to #28: README speaker clause and patch 0003 subject - #30
Draft
plotarmordev wants to merge 4 commits into
Draft
plotarmordev wants to merge 4 commits into
plotarmordev wants to merge 4 commits into
Conversation
…saurus/Drama-tensorfold into fix/pr28-followups
Credits @krunkosaurus (#28). Lists speaker tags 1-15 (0015, CUDA 0003, Metal 0006) as a flag-free request feature in the README, and gives CUDA patch 0003 the series' 'server:' subject prefix (SHA256SUMS updated; patch tree unchanged).
4 of 5 tasks
The full-model paragraph named only request sampling as flag-free. Speaker tags 0 to 15 are too, and CUDA NVFP4 0003 and Metal NVFP4 0006 carry them. I keep the 0003 subject fix from my earlier commit; its old README hunk targeted the patch-series paragraph that #33 replaced.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Small follow-up fixes from my review of @krunkosaurus's stack (#26, #27, #28), kept out of his branches. Credit for the features goes to him.
I merged
maininto this branch after #33 and #34 landed (no rebase, no force push). #33 replaced the full-model patch series with TensorFold v1.1.9, so the README paragraph my first commit edited no longer exists; I dropped that hunk and rewrote the clause against the new paragraph. What remains againstmain:Full-model engine: the paragraph named only request sampling as needing no launch flag. It now lists speaker tags 0 to 15 in request text beside it, and names the NVFP4 patches that carry both features: CUDA 0002 and 0003, Metal 0004 to 0006.engine/nvfp4/patches/0003-dialogue-speakers.patch: subject changed toserver: preserve dialogue speakers across long-form requeststo match theserver:prefix of 0001, 0002 and 0004 in the CUDA NVFP4 series. Only the patch header changes;SHA256SUMSupdated for 0003.Diff against
main: 3 files,README.md, the 0003 patch header and itsSHA256SUMSline.Checks on this head:
git diff --check origin/main HEADclean;sha256sum -c SHA256SUMSOK forengine/nvfp4/patches(4/4) andengine/nvfp4/macos/patches(6/6), andengine/patchesis the empty full-model manifest as onmain;python3 -m unittest discover -s recipe -p 'test_*.py'ran 50 tests, OK (skipped=15);python3 bench/stream_check.py --self-testpassed. I ran no GPU check; the patch content is unchanged.