Add chartTemplate config value

Also fix missed renames of model_ -> hf_model_ config options.

Author: sd109

chart/azimuth-ui.schema.yaml

@@ -5,7 +5,7 @@ controls:
   /huggingface/token:
     type: TextControl
     secret: true
-  /ui/appSettings/model_instruction:
+  /ui/appSettings/hf_model_instruction:
     type: TextControl
   /ui/appSettings/page_title:
     type: TextControl

chart/templates/api/deployment.yml

@@ -28,6 +28,10 @@ spec:
         args:
           - --model
           - {{ .Values.huggingface.model }}
+          {{- if .Values.huggingface.chatTemplate }}
+          - --chat-template
+          - {{ quote .Values.huggingface.chatTemplate }}
+          {{- end -}}
           {{- if .Values.api.extraArgs -}}
           {{- .Values.api.extraArgs | toYaml | nindent 10 }}
           {{- end -}}

chart/values.schema.json

@@ -25,13 +25,13 @@
                 "appSettings": {
                     "type": "object",
                     "properties": {
-                        "model_name": {
+                        "hf_model_name": {
                             "type": "string",
                             "title": "Model Name",
                             "description": "Model name supplied to the OpenAI client in frontend web app. Should match huggingface.model above.",
                             "default": "mistralai/Mistral-7B-Instruct-v0.2"
                         },
-                        "model_instruction": {
+                        "hf_model_instruction": {
                             "type": "string",
                             "title": "Instruction",
                             "description": "The initial model prompt (i.e. the hidden instructions) to use when generating responses.",
@@ -75,7 +75,7 @@
                         }
 
                     },
-                    "required": ["model_name", "model_instruction"]
+                    "required": ["hf_model_name", "hf_model_instruction"]
                 }
             }
         }

chart/values.yaml

@@ -6,6 +6,15 @@ huggingface:
   # The name of the HuggingFace model to use
   # Use a yaml anchor to avoid duplication elsewhere
   model: &model-name ise-uiuc/Magicoder-S-DS-6.7B
+  # A Jinja formatted chat template to provide to the language model.
+  # See https://huggingface.co/blog/chat-templates for background info.
+  # If not provided, the default template specified in the HuggingFace
+  # model repository's tokenizer_config.json file is used. As explained
+  # in the above blog post, the HF template key in tokenizer_config.json
+  # is relatively new and not all HF models include a template in their
+  # repo files yet. This chart value provides a hook to manually apply the
+  # correct chat template for such models.
+  chatTemplate:
 
   # For private/gated huggingface models (e.g. Meta's Llama models)
   # you must provide your own huggingface token, for details see:
@@ -71,8 +80,8 @@ ui:
   # The values to be written to settings.yml for parsing as frontend app setting
   # (see example_app.py and config.py for example using pydantic-settings to configure app)
   appSettings:
-    model_name: *model-name
-    model_instruction: "You are a helpful AI assistant. Please response appropriately."
+    hf_model_name: *model-name
+    hf_model_instruction: "You are a helpful AI assistant. Please response appropriately."
   # Container image config
   image:
     repository: ghcr.io/stackhpc/azimuth-llm-ui-base