Fixed Phi 3, Fixed Function Call Result Message and changed some exam…

…ples.

examples/02_Structured_Output/output_knowledge_graph.py

@@ -6,9 +6,9 @@
 from llama_cpp_agent import LlamaCppAgent
 from llama_cpp_agent.llm_output_settings import LlmStructuredOutputSettings, LlmStructuredOutputType
 from llama_cpp_agent import MessagesFormatterType
-from llama_cpp_agent.providers import TGIServerProvider
+from llama_cpp_agent.providers import LlamaCppServerProvider
 
-provider = TGIServerProvider("http://localhost:8080")
+provider = LlamaCppServerProvider("http://localhost:8080")
 
 
 class Node(BaseModel):
@@ -54,7 +54,7 @@ def visualize_knowledge_graph(kg):
 
 
 def generate_graph(user_input: str):
-    prompt = f"""Help me understand the following by describing it as a extremely detailed knowledge graph with at least 20 nodes: {user_input}""".strip()
+    prompt = f"""Help me understand the following by describing it as a extremely detailed knowledge graph with at least 40 nodes: {user_input}""".strip()
     response = agent.get_chat_response(
         message=prompt,
         structured_output_settings=output_settings
@@ -63,5 +63,5 @@ def generate_graph(user_input: str):
     return response
 
 
-graph = generate_graph("Teach me about quantum mechanics")
+graph = generate_graph("large language models.")
 visualize_knowledge_graph(graph)

examples/03_Tools_And_Function_Calling/function_calling_agent.py

@@ -8,9 +8,9 @@
 from llama_cpp_agent import LlamaCppFunctionTool
 from llama_cpp_agent import FunctionCallingAgent
 from llama_cpp_agent import MessagesFormatterType
-from llama_cpp_agent.providers import TGIServerProvider
+from llama_cpp_agent.providers import LlamaCppServerProvider
 
-model = TGIServerProvider("http://localhost:8080")
+model = LlamaCppServerProvider("http://localhost:8080")
 
 
 # Simple tool for the agent, to get the current date and time in a specific format.
@@ -93,7 +93,7 @@ def get_current_weather(location, unit):
 
 # Callback for receiving messages for the user.
 def send_message_to_user_callback(message: str):
-    print(message)
+    print("Assistant: " + message.strip())
 
 # First we create the calculator tool.
 calculator_function_tool = LlamaCppFunctionTool(calculator)
@@ -110,8 +110,16 @@ def send_message_to_user_callback(message: str):
     llama_cpp_function_tools=[calculator_function_tool, current_datetime_function_tool, get_weather_function_tool],
     send_message_to_user_callback=send_message_to_user_callback,
     allow_parallel_function_calling=True,
+    debug_output=False,
     messages_formatter_type=MessagesFormatterType.CHATML)
 
-user_input = "What is the current weather in London celsius?"
-function_call_agent.generate_response(user_input)
+user_input = '''Get the date and time in '%d-%m-%Y %H:%M' format. Get the current weather in celsius in London, New York and at the North Pole. Solve the following calculations: 42 * 42, 74 + 26, 7 * 26, 4 + 6  and 96/8.'''
+print("User: " + user_input)
+
+settings = model.get_provider_default_settings()
+settings.add_additional_stop_sequences(["<|end|>"])
+settings.stream = False
+settings.temperature = 0.65
+
+function_call_agent.generate_response(user_input, llm_sampling_settings=settings)
 

examples/04_Chains/math_operation_greeting.py

@@ -5,9 +5,9 @@
 from llama_cpp_agent import AgentChainElement, AgentChain
 from llama_cpp_agent import LlamaCppAgent
 from llama_cpp_agent import MessagesFormatterType
-from llama_cpp_agent.providers import LlamaCppServerProvider
+from llama_cpp_agent.providers import TGIServerProvider
 
-model = LlamaCppServerProvider("http://127.0.0.1:8080")
+model = TGIServerProvider("http://127.0.0.1:8080")
 
 agent = LlamaCppAgent(
     model,

examples/05_Rag/example_synthetic_diamonds_bars.py

@@ -39,7 +39,7 @@
     rag.add_document(split)
 
 # Define a llamacpp server endpoint.
-from llama_cpp_agent.providers.llama_cpp_server import LlamaCppServerProvider
+from llama_cpp_agent.providers import LlamaCppServerProvider
 
 model = LlamaCppServerProvider("http://127.0.0.1:8080")
 

src/llama_cpp_agent/chat_history/messages.py

@@ -95,7 +95,7 @@ def convert_messages_to_list_of_dictionaries(
                 else:
                     content = f"Function Call:\nFunction: {message.tool_calls[0].function.name}\nArguments: {message.tool_calls[0].function.arguments}\n"
         elif isinstance(message, ToolMessage):
-            content = f"Function Call Result:\nResult: {message.content}\n"
+            content = f"{message.content}\n"
         else:
             content = f"{message.content}"
         # Construct the dictionary for the current message

src/llama_cpp_agent/messages_formatter.py

@@ -127,12 +127,12 @@
 FUNCTION_PROMPT_END_OPEN_CHAT = """"""
 DEFAULT_OPEN_CHAT_STOP_SEQUENCES = ["<|end_of_turn|>"]
 
-SYS_PROMPT_START_PHI_3 = """"""
-SYS_PROMPT_END_PHI_3 = """\n\n"""
-USER_PROMPT_START_PHI_3 = """<|user|>"""
-USER_PROMPT_END_PHI_3 = """<|end|>\n"""
-ASSISTANT_PROMPT_START_PHI_3 = """<|assistant|>"""
-ASSISTANT_PROMPT_END_PHI_3 = """<|end|>\n"""
+SYS_PROMPT_START_PHI_3 = """<|system|>\n"""
+SYS_PROMPT_END_PHI_3 = """<|endoftext|>\n"""
+USER_PROMPT_START_PHI_3 = """<|user|>\n"""
+USER_PROMPT_END_PHI_3 = """<|endoftext|>\n"""
+ASSISTANT_PROMPT_START_PHI_3 = """<|assistant|>\n"""
+ASSISTANT_PROMPT_END_PHI_3 = """<|endoftext|>\n"""
 FUNCTION_PROMPT_START_PHI_3 = """"""
 FUNCTION_PROMPT_END_PHI_3 = """"""
 DEFAULT_PHI_3_STOP_SEQUENCES = ["<|end|>", "<|end_of_turn|>"]
@@ -546,11 +546,11 @@ def as_dict(self) -> dict:
     USER_PROMPT_END_PHI_3,
     ASSISTANT_PROMPT_START_PHI_3,
     ASSISTANT_PROMPT_END_PHI_3,
-    True,
+    False,
     DEFAULT_PHI_3_STOP_SEQUENCES,
     True,
     FUNCTION_PROMPT_START_PHI_3,
-    FUNCTION_PROMPT_END_PHI_3,
+    FUNCTION_PROMPT_END_PHI_3
 )
 
 predefined_formatter = {