Change model to Qwen3-14B

app.py

@@ -42,7 +42,7 @@ def run_workflow(input, history, compute_mode, thread_id, session_hash):
     if compute_mode == "local":
         if not torch.cuda.is_available():
             raise gr.Error(
-                "Local mode requires GPU. Please select remote mode.",
+                "Local mode requires GPU.",
                 print_exception=False,
             )
 
@@ -244,7 +244,11 @@ with gr.Blocks(
         ],
         value=("local" if torch.cuda.is_available() else "remote"),
         label="Compute Mode",
-        info=(None if torch.cuda.is_available() else "NOTE: local mode requires GPU"),
+        info=(
+            "NOTE: remote mode is available even if you have exceeded your ZeroGPU quota"
+            if torch.cuda.is_available()
+            else "NOTE: local mode requires GPU"
+        ),
         render=False,
     )
 
@@ -355,8 +359,8 @@ with gr.Blocks(
         if compute_mode == "local":
             status_text = f"""
             📍 Now in **local** mode, using ZeroGPU hardware<br>
-            ⌛ Response time is around 2 minutes<br>
-            ✨ [Nomic](https://huggingface.co/nomic-ai/nomic-embed-text-v1.5) embeddings and [{model_id}](https://huggingface.co/{model_id})<br>
+            ⌛ Response time is around 1 minute<br>
+            ✨ [nomic-embed-text-v1.5](https://huggingface.co/nomic-ai/nomic-embed-text-v1.5) and [{model_id.split("/")[-1]}](https://huggingface.co/{model_id})<br>
             🏠 See the project's [GitHub repository](https://github.com/jedick/R-help-chat)
             """
         return status_text

graph.py

@@ -9,7 +9,7 @@ import os
 
 # Local modules
 from retriever import BuildRetriever
-from prompts import query_prompt, generate_prompt, gemma_tools_template
+from prompts import query_prompt, generate_prompt, generic_tools_template
 from mods.tool_calling_llm import ToolCallingLLM
 
 # Local modules
@@ -81,7 +81,7 @@ def ToolifyHF(chat_model, system_message, system_message_suffix="", think=False)
     #    system_message = "/no_think\n" + system_message
 
     # Combine system prompt and tools template
-    tool_system_prompt_template = system_message + gemma_tools_template
+    tool_system_prompt_template = system_message + generic_tools_template
 
     class HuggingFaceWithTools(ToolCallingLLM, ChatHuggingFace):
 

main.py

@@ -40,7 +40,8 @@ openai_model = "gpt-4o-mini"
 model_id = os.getenv("MODEL_ID")
 if model_id is None:
     # model_id = "HuggingFaceTB/SmolLM3-3B"
-    model_id = "google/gemma-3-12b-it"
+    # model_id = "google/gemma-3-12b-it"
+    model_id = "Qwen/Qwen3-14B"
 
 # Suppress these messages:
 # INFO:httpx:HTTP Request: POST https://api.openai.com/v1/embeddings "HTTP/1.1 200 OK"

prompts.py

@@ -84,9 +84,9 @@ You must always select one of the above tools and respond with only a JSON objec
 
 """
 
-# Prompt template for Gemma 3 with tools
+# Prompt template for Gemma/Qwen with tools
 # Based on https://ai.google.dev/gemma/docs/capabilities/function-calling
-gemma_tools_template = """
+generic_tools_template = """
 
 ### Functions