tomoul rm

Delete cached model weights to free disk. Cache lives at `~/.cache/tomoul/` by default.

Usage

$ tomoul rm inkubalm-0.4b
✓  Removed inkubalm-0.4b (240 MB)

Flag	Notes
`--all`	Remove every cached model. Confirms first.
`--quant <q>`	Only remove a specific quantization variant of the named model.
`--force`	Skip confirmation on `--all`.
`--dry-run`	Print what would be removed, do nothing.

If a tomoul serve process has the model loaded, rm fails — the weights are mmapped. Stop the server first.

The cache is regenerable — tomoul pull or tomoul serve will re-download — so removal is safe by default.