CUBE-182 - Replace all references to tiny llama with the new default model (#188)

* Replace refences of tinyllama with new default model

Signed-off-by: Jilks Smith <smithjilks@gmail.com>

* Revert config.json

Signed-off-by: Jilks Smith <smithjilks@gmail.com>

* Revert vll-compose.yaml

Signed-off-by: Jilks Smith <smithjilks@gmail.com>

---------

Signed-off-by: Jilks Smith <smithjilks@gmail.com>

Author: smithjilks

README.md

@@ -174,7 +174,7 @@ Cube AI uses TEEs to protect user data and AI models from unauthorized access. T
      -H "Content-Type: application/json" \
      -H "Authorization: Bearer YOUR_ACCESS_TOKEN" \
      -d '{
-       "model": "tinyllama:1.1b",
+       "model": "llama3.2:3b",
        "messages": [
          {
            "role": "user",
@@ -222,7 +222,7 @@ Example:
 curl -k https://localhost/proxy/YOUR_DOMAIN_ID/v1/chat/completions \
   -H "Authorization: Bearer YOUR_ACCESS_TOKEN" \
   -H "Content-Type: application/json" \
-  -d '{"model":"tinyllama:1.1b","messages":[{"role":"user","content":"Hello"}]}'
+  -d '{"model":"llama3.2:3b","messages":[{"role":"user","content":"Hello"}]}'
 
 # Ollama API endpoint
 curl -k https://localhost/proxy/YOUR_DOMAIN_ID/api/tags \

api/guardrails.yaml

@@ -462,7 +462,7 @@ components:
             models:
               - type: main
                 engine: ollama
-                model: tinyllama
+                model: llama3.2:3b
         prompts_yaml:
           type: string
           description: YAML content for prompts.yml

docker/.env

@@ -244,7 +244,7 @@ VLLM_LOGGING_LEVEL=INFO
 
 # UI LLM Configuration (supports both ollama and vllm via proxy)
 UV_CUBE_UI_LLM_BASE_URL=http://cube-proxy:${UV_CUBE_PROXY_PORT}
-UV_CUBE_UI_LLM_DEFAULT_MODEL=tinyllama:1.1b
+UV_CUBE_UI_LLM_DEFAULT_MODEL=llama3.2:3b
 
 # UI Attestation Configuration
 CUBE_AI_ATTESTATION_URL=http://cube-proxy:${UV_CUBE_PROXY_PORT}

guardrails/src/drivers/rest/routers/schemas.py

@@ -123,7 +123,7 @@ class ChatRequest(BaseModel):
     """Schema for chat completion request."""
 
     messages: List[ChatMessage]
-    model: Optional[str] = "tinyllama"
+    model: Optional[str] = "llama3.2:3b"
     temperature: Optional[float] = 0.1
     max_tokens: Optional[int] = 150
 

hal/buildroot/linux/README.md

@@ -87,7 +87,7 @@ Navigate to **Target packages → Cube AI Services**:
   - **Client CA Certificates**: Path to file
 
 ##### Ollama Backend (if selected)
-- **Install default models**: Automatically pulls tinyllama:1.1b, starcoder2:3b, nomic-embed-text:v1.5
+- **Install default models**: Automatically pulls llama3.2:3b, starcoder2:3b, nomic-embed-text:v1.5
 - **Custom models**: Space-separated list (e.g., `llama2:7b mistral:7b codellama:13b`)
 - **GPU Support**: Enable NVIDIA or AMD GPU acceleration
 
@@ -198,7 +198,7 @@ ollama list
 ollama pull llama2:7b
 
 # Remove models
-ollama rm tinyllama:1.1b
+ollama rm llama3.2:3b
 ```
 
 ### vLLM Models

hal/buildroot/linux/package/ollama/Config.in

@@ -15,7 +15,7 @@ config BR2_PACKAGE_OLLAMA_MODELS
 	default y
 	help
 	  Automatically pull default models on first boot
-	  (tinyllama:1.1b, starcoder2:3b, nomic-embed-text:v1.5)
+	  (llama3.2:3b, starcoder2:3b, nomic-embed-text:v1.5)
 
 config BR2_PACKAGE_OLLAMA_CUSTOM_MODELS
 	string "Custom models to install"

hal/ubuntu/qemu.sh

@@ -141,7 +141,7 @@ write_files:
         sleep 2
       done
       # Pull models
-      /usr/local/bin/ollama pull tinyllama:1.1b
+      /usr/local/bin/ollama pull llama3.2:3b
       /usr/local/bin/ollama pull starcoder2:3b
       /usr/local/bin/ollama pull nomic-embed-text:v1.5
     permissions: '0755'