Merge branch 'main' into adil/agent_format

arch/arch_config_schema.yaml

@@ -7,7 +7,8 @@ properties:
       - v0.1
       - v0.1.0
       - 0.1-beta
-      - v0.2.0
+      - 0.2.0
+      - v0.3.0
 
   agents:
     type: array
@@ -111,6 +112,19 @@ properties:
       additionalProperties: false
       required:
         - model
+
+  model_aliases:
+    type: object
+    patternProperties:
+      "^.*$":
+        type: object
+        properties:
+          target:
+            type: string
+        additionalProperties: false
+        required:
+          - target
+
   overrides:
     type: object
     properties:

arch/tools/README.md

@@ -19,7 +19,7 @@ source venv/bin/activate
 
 ### Step 3: Run the build script
 ```bash
-pip install archgw==0.3.11
+pip install archgw==0.3.12
 ```
 
 ## Uninstall Instructions: archgw CLI

arch/tools/cli/config_generator.py

@@ -301,6 +301,16 @@ def validate_and_render_schema():
             listener["llm_providers"] = deepcopy(updated_llm_providers)
     config_yaml["llm_providers"] = updated_llm_providers
 
+    # Validate model aliases if present
+    if "model_aliases" in config_yaml:
+        model_aliases = config_yaml["model_aliases"]
+        for alias_name, alias_config in model_aliases.items():
+            target = alias_config.get("target")
+            if target not in model_name_keys:
+                raise Exception(
+                    f"Model alias '{alias_name}' targets '{target}' which is not defined as a model. Available models: {', '.join(sorted(model_name_keys))}"
+                )
+
     arch_config_string = yaml.dump(config_yaml)
     arch_llm_config_string = yaml.dump(config_yaml)
 

arch/tools/cli/consts.py

@@ -10,4 +10,4 @@ SERVICE_NAME_MODEL_SERVER = "model_server"
 SERVICE_ALL = "all"
 MODEL_SERVER_LOG_FILE = "~/archgw_logs/modelserver.log"
 ARCHGW_DOCKER_NAME = "archgw"
-ARCHGW_DOCKER_IMAGE = os.getenv("ARCHGW_DOCKER_IMAGE", "katanemo/archgw:0.3.11")
+ARCHGW_DOCKER_IMAGE = os.getenv("ARCHGW_DOCKER_IMAGE", "katanemo/archgw:0.3.12")

arch/tools/poetry.lock

@@ -2,7 +2,7 @@
 
 [[package]]
 name = "archgw_modelserver"
-version = "0.3.11"
+version = "0.3.12"
 description = "A model server for serving models"
 optional = false
 python-versions = "*"
@@ -587,4 +587,4 @@ files = [
 [metadata]
 lock-version = "2.0"
 python-versions = "^3.10"
-content-hash = "f4339b8be7bc43b0506bc703dd9a13cdbeb33bfb1cf85e48880d48b77e4ae424"
+content-hash = "47de0f12b9da376ac1ef76c1efa696debf064399ec21040449303f7b6ec65c65"

arch/tools/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "archgw"
-version = "0.3.11"
+version = "0.3.12"
 description = "Python-based CLI tool to manage Arch Gateway."
 authors = ["Katanemo Labs, Inc."]
 packages = [
@@ -10,7 +10,7 @@ readme = "README.md"
 
 [tool.poetry.dependencies]
 python = "^3.10"
-archgw_modelserver = "^0.3.11"
+archgw_modelserver = "^0.3.12"
 click = "^8.1.7"
 jinja2 = "^3.1.4"
 jsonschema = "^4.23.0"