GitLab

[Feature] Support Multiturn Inference for Audio Conversation task.

Add multiturn audio processing in infer.py and update KimiAPromptManager to handle audio-text messages. Introduce new audio and text files for testing multiturn cases.

Add install via pip in README.md

Add a minimal pyproject.toml

This reverts commit c3cface8.

update README

lm generate add torch inference mode wrap

Signed-off-by: weedge <weege007@gmail.com>

Fix bug： cannot load from model-id

Update README.md to include Kimi-Audio-7B model.