llama swap update
This commit is contained in:
+5
-10
@@ -48,7 +48,8 @@
|
||||
batch = "-b 1024 -ub 1024"; # default 512 512
|
||||
hf_repo = "";
|
||||
image-tokens = "--image-min-tokens 256 --image-max-tokens 1536";
|
||||
tools = "--tools 'all'";
|
||||
#tools = "--tools 'all'";
|
||||
tools = "";
|
||||
thinking = "--reasoning on";
|
||||
no-thinking = "--reasoning off";
|
||||
};
|
||||
@@ -79,18 +80,18 @@
|
||||
};
|
||||
|
||||
"gemma4-26b-a4b" = {
|
||||
cmd = "${llama-server} --port $\{PORT\} --host 0.0.0.0 --ctx-size $\{ctx\} -fa 0 -ngl $\{ngl\} --hf-repo $\{hf_repo\} $\{tools\}";
|
||||
cmd = "${llama-server} --port $\{PORT\} --host 0.0.0.0 --ctx-size $\{ctx\} $\{kv_cache\} -ngl $\{ngl\} --hf-repo $\{hf_repo\} --no-mmap $\{tools\}";
|
||||
aliases = [
|
||||
"gemma4"
|
||||
];
|
||||
ttl = 900;
|
||||
macros = {
|
||||
hf_repo = "unsloth/gemma-4-26B-A4B-it-GGUF:UD-IQ3_XXS";
|
||||
ctx = 36000;
|
||||
ctx = 48000;
|
||||
};
|
||||
};
|
||||
"gemma4-12b" = {
|
||||
cmd = "${llama-server} --port $\{PORT\} --host 0.0.0.0 --ctx-size $\{ctx\} -fa 0 -ngl $\{ngl\} --hf-repo $\{hf_repo\} --spec-type draft-mtp $\{tools\}";
|
||||
cmd = "${llama-server} --port $\{PORT\} --host 0.0.0.0 --ctx-size $\{ctx\} -fa 1 -ngl $\{ngl\} --hf-repo $\{hf_repo\} --spec-type draft-mtp --no-mmap $\{tools\}";
|
||||
ttl = 900;
|
||||
macros = {
|
||||
hf_repo = "unsloth/gemma-4-12B-it-qat-GGUF:UD-Q4_K_XL";
|
||||
@@ -116,12 +117,6 @@
|
||||
};
|
||||
};
|
||||
|
||||
"ministal-3-3b" = {
|
||||
cmd = "${llama-server} --port $\{PORT\} --host 0.0.0.0 --ctx-size $\{ctx\} -ngl $\{ngl\} --hf-repo $\{hf_repo\} $\{tools\} ";
|
||||
aliases = [ "ministral3-mini" ];
|
||||
ttl = 900;
|
||||
macros.hf_repo = "mistralai/Ministral-3-3B-Instruct-2512-GGUF";
|
||||
};
|
||||
|
||||
"z-image-turbo" = {
|
||||
cmd = "${sd-server} --listen-port $\{PORT\} --diffusion-model $\{diffusion_model\} --vae $\{vae\} --llm $\{llm\} --offload-to-cpu --cfg-scale 1.0 --height 1024 --width 1024 --steps 4";
|
||||
|
||||
Reference in New Issue
Block a user