Showing entries 1-6 out of 6.
Cache mode for VRAM savings
List of fake model names sent via the /v1/models endpoint
Automatically allocate resources to GPUs
Max sequence length
Directory to look for models
The initial model to load on startup
option-set