Reduced memory use and minimal boilerplate for open models
The Hardware platform column shows where an issue is most relevant. "All hardware platforms" applies to every platform listed in this playbook.
| Symptom | Hardware platform | Cause | Fix |
|---|---|---|---|
nvcc not found or wrong CUDA version | All hardware platforms | CUDA toolkit missing or not on PATH | Install or configure CUDA 13.0; confirm with nvcc --version |
| Container fails to start with GPU error | All hardware platforms | NVIDIA Container Toolkit not configured | Run nvidia-ctk runtime configure --runtime=docker and restart Docker |
| "permission denied" when running docker | All hardware platforms | User not in docker group | Run sudo usermod -aG docker $USER && newgrp docker |
| Pip install or import errors for Unsloth / Triton | All hardware platforms | Dependency or compiler mismatch inside the container | Use the playbook container image and install commands; retry in a fresh container |
| CUDA out of memory / training OOM | All hardware platforms | Batch size, sequence length, or model too large for available memory | Lower per_device_train_batch_size, max_seq_length, or use a smaller 4-bit model |
| Training loss does not decrease / job exits early | All hardware platforms | Dataset, config, or interrupted download | Confirm network access; re-run with the default test_unsloth.py settings |
| Memory pressure within capacity | DGX Spark | UMA buffer cache not released | See UMA note below |
NOTE
Unified memory (UMA). On hardware platforms with unified memory, GPU and CPU share memory dynamically. Some applications have not yet been updated for UMA, so you may hit memory issues even within capacity. If that happens, manually flush the buffer cache:
sudo sh -c 'sync; echo 3 > /proc/sys/vm/drop_caches'
For latest known issues, see the documentation linked under Resources for your hardware platform.