Feature Request: Default config for --model-presets

[eleqtrizit]

Prerequisites

• I am running the latest code. Mention the version if possible as well.
• I carefully followed the README.md.
• I searched using keywords relevant to my issue to make sure that I am creating a new issue that is not already open (or closed).
• I reviewed the Discussions, and have a new and useful enhancement to share.

Feature Description

Can we get a catch-all wild card for the preset.ini?

e.g.

[*]
c = 128000

Right now, the CLI overrides the context in the ini, so there is no way to set a default context limit.

Motivation

Laziness.

Possible Implementation

No response

[ngxson]

@eleqtrizit could you confirm if #17959 works?

[ServeurpersoCom]

I tried moving my CLI arguments into a [*] section but it didn't worked for me :

llama-server --port 8082 --models-max 1 --models-preset config.ini

[*]
ngl = 999
ctk = q8_0
ctv = q8_0
fa = on
mlock = on
np = 4
kvu = on

[Dense-Devstral-Small-2-24B-Instruct-2512]
m = unsloth/Devstral-Small-2-24B-Instruct-2512-GGUF/Devstral-Small-2-24B-Instruct-2512-UD-Q5_K_XL.gguf
c = 131072

srv  log_server_r: request: GET /v1/models 127.0.0.1 200
srv          load: spawning server instance with name=Dense-Devstral-Small-2-24B-Instruct-2512 on port 39055
srv          load: spawning server instance with args:
srv          load:   /root/llama.cpp.pascal/build/bin/llama-server
srv          load:   --host
srv          load:   127.0.0.1
srv          load:   --port
srv          load:   39055
srv          load:   --alias
srv          load:   Dense-Devstral-Small-2-24B-Instruct-2512
srv          load:   --ctx-size
srv          load:   131072
srv          load:   --model
srv          load:   unsloth/Devstral-Small-2-24B-Instruct-2512-GGUF/Devstral-Small-2-24B-Instruct-2512-UD-Q5_K_XL.gguf
srv  log_server_r: request: POST /models/load 127.0.0.1 200
srv  log_server_r: request: GET /v1/models 127.0.0.1 200

I get OOM because no ctk/ctv

[eleqtrizit]

I tried a very simple

[*]
c = 128000

and it still loaded default 4096 for command llama-server --host 0.0.0.0 --models-preset llama.ini

srv  log_server_r: request: GET /v1/models 10.221.42.205 200
srv          load: spawning server instance with name=unsloth/gpt-oss-20b-GGUF:UD-Q8_K_XL on port 39849
srv          load: spawning server instance with args:
srv          load:   /home/arivera/repos/llama.cpp/build/bin/llama-server
srv          load:   --host
srv          load:   127.0.0.1
srv          load:   --port
srv          load:   39849
srv          load:   --alias
srv          load:   unsloth/gpt-oss-20b-GGUF:UD-Q8_K_XL
srv          load:   --hf-repo
srv          load:   unsloth/gpt-oss-20b-GGUF:UD-Q8_K_XL

[ngxson]

I think it need a refactoring to support this (I already had an idea but will see when I have time). The current system make it quite hard to merge configs from different model sources

[ngxson]

@eleqtrizit could you try #18169 ?

[eleqtrizit]

Sorry, I missed your message, but I see it's been merged. Thank you, @ngxson ! Works great.

[disarticulate]

Along this same line: Is there any resource that provides per model sample configs?