From 85d59e20ecfc204b4ddc13f87dc44c956e17ff56 Mon Sep 17 00:00:00 2001 From: Salman Paracha Date: Fri, 20 Jun 2025 17:12:12 -0700 Subject: [PATCH] local support for Arch-Router via Ollama --- .../brightstaff/src/router/router_model_v1.rs | 1 + .../arch_config_local.yaml | 45 +++++++++++++++++++ 2 files changed, 46 insertions(+) create mode 100644 demos/use_cases/preference_based_routing/arch_config_local.yaml diff --git a/crates/brightstaff/src/router/router_model_v1.rs b/crates/brightstaff/src/router/router_model_v1.rs index 08f742ba..dc623f0a 100644 --- a/crates/brightstaff/src/router/router_model_v1.rs +++ b/crates/brightstaff/src/router/router_model_v1.rs @@ -144,6 +144,7 @@ impl RouterModel for RouterModelV1 { content: Some(ContentType::Text(messages_content)), role: USER_ROLE.to_string(), }], + temperature: Some(0.01), ..Default::default() } } diff --git a/demos/use_cases/preference_based_routing/arch_config_local.yaml b/demos/use_cases/preference_based_routing/arch_config_local.yaml new file mode 100644 index 00000000..607d180a --- /dev/null +++ b/demos/use_cases/preference_based_routing/arch_config_local.yaml @@ -0,0 +1,45 @@ +version: v0.1.0 + +routing: + model: arch-router + +listeners: + egress_traffic: + address: 0.0.0.0 + port: 12000 + message_format: openai + timeout: 30s + +llm_providers: + + - name: arch-router + access_key: $ARCH_API_KEY + provider_interface: arch + model: hf.co/katanemo/Arch-Router-1.5B.gguf:Q4_K_M + endpoint: host.docker.internal:11434 + + - name: gpt-4o-mini + provider_interface: openai + access_key: $OPENAI_API_KEY + model: gpt-4o-mini + + - name: gpt-4.1 + provider_interface: openai + access_key: $OPENAI_API_KEY + model: gpt-4.1 + default: true + + - name: code_generation + access_key: $OPENAI_API_KEY + provider_interface: openai + model: gpt-4.1 + usage: generating new code snippets, functions, or boilerplate based on user prompts or requirements + + - name: code_understanding + provider_interface: openai + access_key: $OPENAI_API_KEY + model: gpt-4.1 + usage: understand and explain existing code snippets, functions, or libraries + +tracing: + random_sampling: 100