Swapped volatile Tailscale ips for fixed wireguard ips
This commit is contained in:
@@ -1,4 +1,3 @@
|
||||
{ pkgs, ... }:
|
||||
{
|
||||
# Avante: AI-powered coding assistant (Cursor-like experience in Neovim)
|
||||
programs.nixvim = {
|
||||
@@ -29,7 +28,7 @@
|
||||
};
|
||||
|
||||
ollama = {
|
||||
endpoint = "http://100.109.179.25:11434"; # tailscale IP, no /v1 suffix
|
||||
endpoint = "http://10.10.0.2:11434";
|
||||
model = "qwen2.5:3b"; # swap for "llama3.2:3b" or "deepseek-r1:1.5b"
|
||||
timeout = 60000; # local + small model can be slow on first load
|
||||
disable_tools = true; # these small models aren't reliable at tool calling
|
||||
|
||||
+1
-1
@@ -12,7 +12,7 @@ _: {
|
||||
npm = "@ai-sdk/openai-compatible";
|
||||
name = "Ollama (tailscale)";
|
||||
options = {
|
||||
baseURL = "http://100.109.179.25:11434/v1";
|
||||
baseURL = "http://10.10.0.2:11434/v1";
|
||||
};
|
||||
models = {
|
||||
"llama3.2:3b" = {
|
||||
|
||||
Reference in New Issue
Block a user