14. Community Resources

Where to Find Answers

GitHub Issues: Before posting, search closed issues. Most common problems have been solved. Include your system specs, exact error message, and steps to reproduce.

Hugging Face Discord: Active community with channels for specific model families. Use the search function before asking.

Stack Overflow: Tag questions with hugging-face or llama-cpp or vllm. High-quality questions with minimal code get answers. "It doesn't work" gets closed.

Local AI Discord servers: Serve specific deployment technologies (Ollama Discord, LocalAI Discord, text-generation-webui Discord).

Filing Effective Bug Reports

A useful bug report contains:

1. Environment: OS, GPU model, driver version, CUDA version, Python version, library versions
2. Reproduction: Exact commands to reproduce the issue
3. Expected behavior: What should happen
4. Actual behavior: What happens instead
5. Full error traceback: Not a screenshot of the error, the actual text

# Generate environment report
echo "=== System ===" > bug_report.txt
uname -a >> bug_report.txt
echo "=== GPU ===" >> bug_report.txt
nvidia-smi >> bug_report.txt
echo "=== CUDA ===" >> bug_report.txt
nvcc --version >> bug_report.txt
echo "=== Python ===" >> bug_report.txt
python --version >> bug_report.txt
pip freeze >> bug_report.txt

Local verification checkpoint

Run the smallest example from this chapter in a local workspace and record the package version, runtime, data path, and observed output. If the result depends on model size, vector count, CPU/GPU backend, or available memory, note that constraint beside the exercise so the lesson remains reproducible.

Local verification checkpoint

Run the smallest example from this chapter in a local workspace and record the package version, runtime, data path, and observed output. If the result depends on model size, vector count, CPU/GPU backend, or available memory, note that constraint beside the exercise so the lesson remains reproducible.