Ollama CLI Commands and Parameters Cheat Sheet
This quick reference guide summarizes CLI actions and Modelfile configurations.
1. CLI Commands Quick Reference
| Command | Action Description | Example Usage |
ollama run <model> | Pull and start interactive terminal session | ollama run qwen2.5:7b |
ollama pull <model> | Download model from registry | ollama pull llama3 |
ollama list | List downloaded local models | ollama list |
ollama rm <model> | Delete local model file | ollama rm llama3 |
ollama create <name> | Build custom model from Modelfile | ollama create my-bot -f ./Modelfile |
ollama show <model> | Inspect model parameters info | ollama show qwen2.5 |
ollama ps | View currently running model info | ollama ps |
2. Modelfile Parameters Reference
Define these parameters under the PARAMETER directive to customize model runtime behavior:
| Parameter Key | Purpose and Description | Recommended Default |
num_ctx | Set the context window token size | 4096 or 8192 |
temperature | Control model creativity levels | 0.7 |
top_p | Top probability nucleus sampling | 0.9 |
top_k | Reduce random word choices | 40 |
num_predict | Set maximum tokens response output | -1 (infinite) or 1024 |
num_gpu | Set layers to offload to GPU | Let Ollama set this |
repeat_penalty | Prevent model repeating identical words | 1.1 |
Published on Last updated: