What I observed:

When running docs2db pipeline on my system, the embedding step got stuck at 0% for 4+ hours. I'm using a CPU-only setup (Intel GPU + Linux, PyTorch CPU version).

System details:

• OS: Arch Linux
• CPU: x86_64 (Intel)
• GPU: Intel UHD Graphics 620 (not supported by PyTorch on Linux)
• PyTorch: CPU-only version

What I tried:

Running docs2db embed --workers 1 fixed the issue and completed successfully.

Questions:

1. Is this expected behavior for CPU-only systems?
2. The --workers 1 flag was added in v0.4.3 - should this be mentioned more prominently in docs?
3. I noticed the default is still 2 workers - would it make sense to auto-detect CPU-only and default to 1 worker, or add a warning message?
4. Without the --workers 1 flag, users might not realize the option exists and could get stuck waiting indefinitely

Possible improvements (just suggestions):

• Change default to 1 worker for CPU-only systems, OR
• Add a warning message like "CPU-only detected, consider using --workers 1", OR
• Add a note in README/Getting Started about this flag

Just wanted to share this observation - maybe it's helpful for improving the user experience!