Instructions to use xirui824/credit-worth-models with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Stable Audio Tools
How to use xirui824/credit-worth-models with Stable Audio Tools:
import torch import torchaudio from einops import rearrange from stable_audio_tools import get_pretrained_model from stable_audio_tools.inference.generation import generate_diffusion_cond device = "cuda" if torch.cuda.is_available() else "cpu" # Download model model, model_config = get_pretrained_model("xirui824/credit-worth-models") sample_rate = model_config["sample_rate"] sample_size = model_config["sample_size"] model = model.to(device) # Set up text and timing conditioning conditioning = [{ "prompt": "128 BPM tech house drum loop", }] # Generate stereo audio output = generate_diffusion_cond( model, conditioning=conditioning, sample_size=sample_size, device=device ) # Rearrange audio batch to a single sequence output = rearrange(output, "b d n -> d (b n)") # Peak normalize, clip, convert to int16, and save to file output = output.to(torch.float32).div(torch.max(torch.abs(output))).clamp(-1, 1).mul(32767).to(torch.int16).cpu() torchaudio.save("output.wav", output, sample_rate) - Notebooks
- Google Colab
- Kaggle
What's a Credit Worth? β trained checkpoints
Research checkpoints for What's a Credit Worth? A Market Framework for Attribution-Aware Compensation in Generative Music.
Code, numerical results, and the recovery guide: GitHub archive branch.
Upload status
The public checkpoint archive is complete and verified: all 473 planned
files are present, including 182 weight files, totaling about 398.20 GiB.
Every weight file matches the local upload SHA-256 record. The repository also
contains this README and the automatically generated .gitattributes.
See the final verification report for the exact byte count and inventory checks. Training recordings and private experiment inputs are stored separately in the private dataset, which is complete and requires authorized access. They are not exposed by this public model repository.
Model and layout
The directory name sao_ip_100 is historical: this experiment uses 50 catalogs.
The uploaded full model was trained with seed 42 for 50 epochs, with per-device
batch size 4 and gradient accumulation 2. The checkpoint bytes and numerical
precision are preserved unchanged. The accompanying resolved architecture
configuration comes from the original fit.
Paths are relative to the source repository root. Download into a matching
checkout; see PRESERVATION.md there for environment and data requirements.
Licenses
The Stable Audio Open-derived component remains subject to the Stable Audio Community License. The source repository contains its code licenses. This upload does not relicense third-party base models or redistribute the training recordings.
The Anticipatory Music Transformer-derived component uses the upstream Apache-2.0 license.
- Downloads last month
- -
Model tree for xirui824/credit-worth-models
Base model
stabilityai/stable-audio-open-1.0