warning
Legacy managed-runtime reference. Command examples below describe the older desktop/managed llama.cpp preview and are legacy, not the current BCCLI v0.4 command surface. Use BCCLI and Local AI for current instructions.Download problems
Symptoms and fixes
| Symptom | Cause | Fix |
|---|---|---|
| Search returns nothing | Hub unreachable | Check connectivity; botconnector doctor shows HF status |
| Download stalls at 0% | Hub rate limit / proxy | Retry; corporate proxies may need exceptions |
| Speed drops mid-download | Hub throttling | Resume continues from the partial file instead of starting over |
| Checksum mismatch on load | Corrupted file / disk issue | Delete and re-download (Installed models) |
| "Disk full" | Store location too small | Free space or move store (Model storage) |
| 401/403 from Hub | Gated repository | Accept license on huggingface.co (Gated models) |
Resuming downloads
botconnector get resumes partial downloads. If a resume loop fails:
powershell
botconnector models remove <publisher>/<model>
botconnector get <publisher>/<model>@Q4_K_MSharded models
Files like -00002-of-00005.gguf are parts of one model; all parts download together. A missing part fails the set — re-run get to complete it.
Slow speeds
- Hub throughput varies by region and load — trying again later usually fixes it.
- VPNs can slow Hub CDN routes; test without.
- Verify disk health if a specific file repeatedly corrupts.
Verification after download
powershell
botconnector lsCompare the size to the file list shown on the model detail page (Model catalog). Mismatch → corrupted → re-download.