Search results

Document available Hugging Face checkpoints and quality/VRAM tradeoffs

Status: Open.

#9 In NVIDIA/diffusion-audio-restoration;

· tapstoop opened on May 27, 2026

Potential memory optimization: avoid storing all intermediate diffusion predictions during inference

Status: Open.

#8 In NVIDIA/diffusion-audio-restoration;

· tapstoop opened on May 27, 2026

Optional evaluation/plotting dependencies are imported globally and can break inference setup

Status: Open.

#7 In NVIDIA/diffusion-audio-restoration;

· tapstoop opened on May 27, 2026

Low-VRAM inference settings are not documented; predict_batch_size is critical

Status: Open.

#6 In NVIDIA/diffusion-audio-restoration;

· tapstoop opened on May 27, 2026

Hardcoded /debug checkpoint/log path causes PermissionError on normal Linux installations

Status: Open.

#5 In NVIDIA/diffusion-audio-restoration;

· tapstoop opened on May 27, 2026

Single-file inference workflow is unclear and CLI/config behavior is under-documented

Status: Open.

#4 In NVIDIA/diffusion-audio-restoration;

· tapstoop opened on May 27, 2026

Installation docs are outdated for modern Python / PyTorch / CUDA environments

Status: Open.

#3 In NVIDIA/diffusion-audio-restoration;

· tapstoop opened on May 27, 2026