目标

/root/.ollama

默认值

/mnt/user/appdata/ollama

价值

/mnt/user/appdata/ollama

Port number where ollama listens on.

目标

11434

默认值

11434

价值

11434

IP and Port the server binds to. Set to 127.0.0.1:11434 for internal only access.

默认值

0.0.0.0:11434

价值

0.0.0.0:11434

Comma-separated list of allowed CORS origins.

默认值

*

价值

*

How long a model stays in VRAM, e.g. 60m or 24h (Set to -1 for infinite, 0 for none).

默认值

5m

价值

5m

Timeout for stall detection during model loads.

默认值

5m

价值

5m

Max number of parallel requests a single model can handle.

默认值

1

价值

1

Default context window (tokens) if not specified by the model.

默认值

4096

价值

4096

Quantization type for the K/V cache, e.g. f16, q8_0, q4_0.

默认值

f16

价值

f16

The path where model weights and blobs are stored.

默认值

/root/.ollama/models

价值

/root/.ollama/models

Maximum number of models loaded per GPU at once (Set to 0 for infinite).

默认值

0

价值

0

Max requests that can wait in line when the server is busy.

默认值

512

价值

512

Log detail level: 0 for INFO, 1 for DEBUG, 2 for TRACE.

默认值

0|1|2

Reserved VRAM (in bytes) to leave empty on each GPU.

默认值

0

价值

0

Enables experimental Flash Attention optimizations.

默认值

false|true

If true, always spreads model layers across all visible GPUs.

默认值

false|true

Optimizes prompt caching when multiple users share a model.

默认值

false|true

If true, does not delete unused model blobs on startup.

默认值

false|true

Disables the readline history in the interactive CLI.

默认值

false|true

Enables the experimental new Ollama engine.

默认值

false|true

Enables experimental Vulkan hardware acceleration.

默认值

false|true

Proxy for downloading models over HTTP.

Proxy for downloading models over HTTPS.

Comma-separate list of hosts/IPs that bypass the proxy.