Merge branch 'main' into llama-ui

Author: yejseo01

README.md

@@ -32,7 +32,7 @@ The stack includes:
 
 | Parameter        | Description                                            | Default                              |
 | ---------------- | ------------------------------------------------------ | ------------------------------------ |
-| `HF_MODEL`       | Hugging Face model repo ID containing `.gguf` files    | `bartowski/Qwen_Qwen3.5-0.8B-GGUF`   |
+| `HF_MODEL`       | Hugging Face model repo ID containing `.gguf` files    | `unsloth/SmolLM2-135M-Instruct-GGUF` |
 | `HF_MODEL_FILE`  | Optional explicit GGUF filename                        | `""`                                |
 | `ENABLE_SVE`     | Enable SVE optimizations                               | `OFF`                                |
 
@@ -56,7 +56,7 @@ topo deploy --target <ip-address-of-target>
 Use a different model:
 ```bash
 topo deploy --target <ip-address-of-target> \
-  --arg HF_MODEL=unsloth/SmolLM2-135M-Instruct-GGUF
+  --arg HF_MODEL=bartowski/Qwen_Qwen3.5-0.8B-GGUF
 ```
 
 Force an exact GGUF file:

compose.yaml

@@ -5,7 +5,7 @@ services:
       context: ./llama-inference
       args:
         ENABLE_SVE: OFF
-        HF_MODEL: bartowski/Qwen_Qwen3.5-0.8B-GGUF
+        HF_MODEL: unsloth/SmolLM2-135M-Instruct-GGUF
         HF_MODEL_FILE: ""
     ports:
       - "8080:8080"
@@ -31,7 +31,7 @@ x-topo:
     - Built-in web chat interface
     - No GPU required - pure CPU inference
 
-    Perfect for demos and testing! The bundled Qwen3.5-0.8B model allows the
+    Perfect for demos and testing! The bundled SmolLM2 135M model allows the
     project to run immediately without downloading additional models.
 
     Ideal for testing LLM workloads on Arm hardware without GPU dependencies,
@@ -45,8 +45,8 @@ x-topo:
   args:
     HF_MODEL:
       description: "Hugging Face model repo ID containing a supported single-file GGUF model"
-      default: "bartowski/Qwen_Qwen3.5-0.8B-GGUF"
-      example: "unsloth/SmolLM2-135M-Instruct-GGUF"
+      default: "unsloth/SmolLM2-135M-Instruct-GGUF"
+      example: "bartowski/Qwen_Qwen3.5-0.8B-GGUF"
 
     HF_MODEL_FILE:
       description: "Exact supported GGUF filename to download; sharded and mmproj files are rejected"