Instructions to use cudabenchmarktest/personaplex-7b-v1-raw-mirror with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Moshi
How to use cudabenchmarktest/personaplex-7b-v1-raw-mirror with Moshi:
# pip install moshi # Run the interactive web server python -m moshi.server --hf-repo "cudabenchmarktest/personaplex-7b-v1-raw-mirror" # Then open https://localhost:8998 in your browser
# pip install moshi import torch from moshi.models import loaders # Load checkpoint info from HuggingFace checkpoint = loaders.CheckpointInfo.from_hf_repo("cudabenchmarktest/personaplex-7b-v1-raw-mirror") # Load the Mimi audio codec mimi = checkpoint.get_mimi(device="cuda") mimi.set_num_codebooks(8) # Encode audio (24kHz, mono) wav = torch.randn(1, 1, 24000 * 10) # [batch, channels, samples] with torch.no_grad(): codes = mimi.encode(wav.cuda()) decoded = mimi.decode(codes) - Notebooks
- Google Colab
- Kaggle
Third-party notices
PersonaPlex is based on Moshi/Moshiko. The upstream PersonaPlex model card lists
kyutai/moshiko-pytorch-bf16 as the base model and identifies CC BY 4.0 as
additional governing information. A copy of CC BY 4.0 is included under
LICENSES/CC-BY-4.0.txt.
The authoritative upstream model card is preserved byte-for-byte as
UPSTREAM_README.md. Components with separate notices remain governed by those
notices.