Skip to content

Add VibeVoice 1.5B Transformers release tooling - #438

Open
Justin Chu (justinchuby) wants to merge 4 commits into
microsoft:mainfrom
justinchuby:justinchuby-publish-vibevoice-1-5b-hf
Open

Justin Chu (justinchuby) wants to merge 4 commits into
microsoft:mainfrom
justinchuby:justinchuby-publish-vibevoice-1-5b-hf

Conversation

@justinchuby

@justinchuby Justin Chu (justinchuby) commented Sep 4, 2026

Copy link
Copy Markdown

I realized microsoft/VibeVoice-1.5B currently is not compatible with the version in HF transformers.

Adds owner-only, reproducible tooling and documentation for publishing microsoft/VibeVoice-1.5B-HF while preserving microsoft/VibeVoice-1.5B as the legacy source/provenance repository.

The converter pins the official checkpoint, canonical Transformers implementation/converter, and Qwen tokenizer. It emits native weights and all required config/tokenizer/processor/chat-template/generation assets, records a conversion manifest, and fails closed unless all 1,204 source tensors map one-to-one with identical shapes and dtypes.

Changed files:

  • README.md
  • docs/vibevoice-tts.md
  • docs/publish-vibevoice-1.5b-hf.md
  • tools/release/convert_vibevoice_1_5b_hf.py
  • tests/test_publish_vibevoice_1_5b_hf.py

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Signed-off-by: Justin Chu <justinchuby@users.noreply.github.com>
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Signed-off-by: Justin Chu <justinchuby@users.noreply.github.com>
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Signed-off-by: Justin Chu <justinchuby@users.noreply.github.com>
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Signed-off-by: Justin Chu <justinchuby@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant