Update app.py

app.py

@@ -1,190 +1,188 @@
-# Importing required libraries
-import json
-import subprocess
-from llama_cpp import Llama
-from llama_cpp_agent import LlamaCppAgent
-from llama_cpp_agent import MessagesFormatterType
-from llama_cpp_agent.providers import LlamaCppPythonProvider
-from llama_cpp_agent.chat_history import BasicChatHistory
-from llama_cpp_agent.chat_history.messages import Roles
-import gradio as gr
-from huggingface_hub import hf_hub_download
-
-
-# Download gguf model files
-llm = None
-llm_model = None
-
-hf_hub_download(
-    repo_id="bartowski/Dolphin3.0-Llama3.2-1B-GGUF",
-    filename="Dolphin3.0-Llama3.2-1B-Q6_K.gguf",
-    local_dir="./models",
-)
-hf_hub_download(
-    repo_id="bartowski/Dolphin3.0-Qwen2.5-1.5B-GGUF",
-    filename="Dolphin3.0-Qwen2.5-1.5B-Q6_K.gguf",
-    local_dir="./models",
-)
-hf_hub_download(
-    repo_id="bartowski/Dolphin3.0-Qwen2.5-0.5B-GGUF",
-    filename="Dolphin3.0-Qwen2.5-0.5B-Q6_K.gguf",
-    local_dir="./models",
-)
-
-# Set the title and description
-title = "Dolphin-3 Llama.cpp 🐬"
-description = """Dolphin 3.0 is a powerful, general-purpose local AI model designed for coding, math, and various other tasks, aiming similar to the models like ChatGPT and Claude."""
-examples = [
-    ["Provide a historical fact about the capital of France."],
-    ["Suggest a three-day itinerary for exploring Tokyo and nearby areas?"],
-    ["Can you write a short, imaginative story about exploring a forest?"],
-]
-
-
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    model,
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-    top_k,
-    repeat_penalty,
-):
-    """
-    Respond to a message using the Dolphin-3 model via Llama.cpp.
-
-    Args:
-        - message (str): The message to respond to.
-        - history (list[tuple[str, str]]): The chat history.
-        - model (str): The model to use.
-        - max_tokens (int): The maximum number of tokens to generate.
-        - temperature (float): The temperature of the model.
-        - top_p (float): The top-p of the model.
-        - top_k (int): The top-k of the model.
-        - repeat_penalty (float): The repetition penalty of the model.
-
-    Returns:
-        str: The response to the message.
-    """
-    # Load the global variables
-    global llm
-    global llm_model
-
-    # Load the model
-    if llm is None or llm_model != model:
-        llm = Llama(
-            model_path=f"models/{model}",
-            flash_attn=False,
-            n_gpu_layers=0,
-            n_batch=32,
-            n_ctx=8192,
-        )
-        llm_model = model
-    provider = LlamaCppPythonProvider(llm)
-
-    # Create the agent
-    agent = LlamaCppAgent(
-        provider,
-        system_prompt=f"{system_message}",
-        predefined_messages_formatter_type=MessagesFormatterType.CHATML,
-        debug_output=True,
-    )
-
-    # Set the settings like temperature, top-k, top-p, max tokens, etc.
-    settings = provider.get_provider_default_settings()
-    settings.temperature = temperature
-    settings.top_k = top_k
-    settings.top_p = top_p
-    settings.max_tokens = max_tokens
-    settings.repeat_penalty = repeat_penalty
-    settings.stream = True
-
-    messages = BasicChatHistory()
-
-    # Add the chat history
-    for msn in history:
-        user = {"role": Roles.user, "content": msn[0]}
-        assistant = {"role": Roles.assistant, "content": msn[1]}
-        messages.add_message(user)
-        messages.add_message(assistant)
-
-    # Get the response stream
-    stream = agent.get_chat_response(
-        message,
-        llm_sampling_settings=settings,
-        chat_history=messages,
-        returns_streaming_generator=True,
-        print_output=False,
-    )
-
-    # Generate the response
-    outputs = ""
-    for output in stream:
-        outputs += output
-        yield outputs
-
-
-# Create a chat interface
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs_accordion=gr.Accordion(
-        label="⚙️ Parameters", open=False, render=False
-    ),
-    additional_inputs=[
-        gr.Dropdown(
-            [
-                "Dolphin3.0-Llama3.2-1B-Q6_K.gguf",
-                "Dolphin3.0-Qwen2.5-1.5B-Q6_K.gguf",
-                "Dolphin3.0-Qwen2.5-0.5B-Q6_K.gguf",
-            ],
-            value="Dolphin3.0-Llama3.2-1B-Q6_K.gguf",
-            label="Model",
-        ),
-        gr.Textbox(
-            value="You are Dolphin an AI assistant that helps humanity.",
-            label="System message",
-        ),
-        gr.Slider(minimum=1, maximum=8192, value=8192, step=1, label="Max tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p",
-        ),
-        gr.Slider(
-            minimum=0,
-            maximum=100,
-            value=40,
-            step=1,
-            label="Top-k",
-        ),
-        gr.Slider(
-            minimum=0.0,
-            maximum=2.0,
-            value=1.1,
-            step=0.1,
-            label="Repetition penalty",
-        ),
-    ],
-    theme="Glass",
-    retry_btn="Retry",
-    undo_btn="Undo",
-    clear_btn="Clear",
-    submit_btn="Send",
-    title=title,
-    description=description,
-    chatbot=gr.Chatbot(scale=1, show_copy_button=True),
-    examples=examples,
-    cache_examples=True,
-    cache_mode="lazy",
-    flagging_mode="never",
-)
-
-
-# Launch the chat interface
-if __name__ == "__main__":
-    demo.launch(debug=False)
+# Importing required libraries
+import json
+import subprocess
+from llama_cpp import Llama
+from llama_cpp_agent import LlamaCppAgent
+from llama_cpp_agent import MessagesFormatterType
+from llama_cpp_agent.providers import LlamaCppPythonProvider
+from llama_cpp_agent.chat_history import BasicChatHistory
+from llama_cpp_agent.chat_history.messages import Roles
+import gradio as gr
+from huggingface_hub import hf_hub_download
+
+
+# Download gguf model files
+llm = None
+llm_model = None
+
+hf_hub_download(
+    repo_id="bartowski/Dolphin3.0-Llama3.2-1B-GGUF",
+    filename="Dolphin3.0-Llama3.2-1B-Q6_K.gguf",
+    local_dir="./models",
+)
+hf_hub_download(
+    repo_id="bartowski/Dolphin3.0-Qwen2.5-1.5B-GGUF",
+    filename="Dolphin3.0-Qwen2.5-1.5B-Q6_K.gguf",
+    local_dir="./models",
+)
+hf_hub_download(
+    repo_id="bartowski/Dolphin3.0-Qwen2.5-0.5B-GGUF",
+    filename="Dolphin3.0-Qwen2.5-0.5B-Q6_K.gguf",
+    local_dir="./models",
+)
+
+# Set the title and description
+title = "Dolphin-3 Llama.cpp 🐬"
+description = """Dolphin 3.0 is a powerful, general-purpose local AI model designed for coding, math, and various other tasks, aiming similar to the models like ChatGPT and Claude."""
+examples = [
+    ["Provide a historical fact about the capital of France."],
+    ["Suggest a three-day itinerary for exploring Tokyo and nearby areas?"],
+    ["Can you write a short, imaginative story about exploring a forest?"],
+]
+
+
+def respond(
+    message,
+    history: list[tuple[str, str]],
+    model,
+    system_message,
+    max_tokens,
+    temperature,
+    top_p,
+    top_k,
+    repeat_penalty,
+):
+    """
+    Respond to a message using the Dolphin-3 model via Llama.cpp.
+
+    Args:
+        - message (str): The message to respond to.
+        - history (list[tuple[str, str]]): The chat history.
+        - model (str): The model to use.
+        - max_tokens (int): The maximum number of tokens to generate.
+        - temperature (float): The temperature of the model.
+        - top_p (float): The top-p of the model.
+        - top_k (int): The top-k of the model.
+        - repeat_penalty (float): The repetition penalty of the model.
+
+    Returns:
+        str: The response to the message.
+    """
+    # Load the global variables
+    global llm
+    global llm_model
+
+    # Load the model
+    if llm is None or llm_model != model:
+        llm = Llama(
+            model_path=f"models/{model}",
+            flash_attn=False,
+            n_gpu_layers=0,
+            n_batch=32,
+            n_ctx=8192,
+        )
+        llm_model = model
+    provider = LlamaCppPythonProvider(llm)
+
+    # Create the agent
+    agent = LlamaCppAgent(
+        provider,
+        system_prompt=f"{system_message}",
+        predefined_messages_formatter_type=MessagesFormatterType.CHATML,
+        debug_output=True,
+    )
+
+    # Set the settings like temperature, top-k, top-p, max tokens, etc.
+    settings = provider.get_provider_default_settings()
+    settings.temperature = temperature
+    settings.top_k = top_k
+    settings.top_p = top_p
+    settings.max_tokens = max_tokens
+    settings.repeat_penalty = repeat_penalty
+    settings.stream = True
+
+    messages = BasicChatHistory()
+
+    # Add the chat history
+    for msn in history:
+        user = {"role": Roles.user, "content": msn[0]}
+        assistant = {"role": Roles.assistant, "content": msn[1]}
+        messages.add_message(user)
+        messages.add_message(assistant)
+
+    # Get the response stream
+    stream = agent.get_chat_response(
+        message,
+        llm_sampling_settings=settings,
+        chat_history=messages,
+        returns_streaming_generator=True,
+        print_output=False,
+    )
+
+    # Generate the response
+    outputs = ""
+    for output in stream:
+        outputs += output
+        yield outputs
+
+
+# Create a chat interface
+demo = gr.ChatInterface(
+    respond,
+    additional_inputs_accordion=gr.Accordion(
+        label="⚙️ Parameters", open=False, render=False
+    ),
+    additional_inputs=[
+        gr.Dropdown(
+            [
+                "Dolphin3.0-Llama3.2-1B-Q6_K.gguf",
+                "Dolphin3.0-Qwen2.5-1.5B-Q6_K.gguf",
+                "Dolphin3.0-Qwen2.5-0.5B-Q6_K.gguf",
+            ],
+            value="Dolphin3.0-Llama3.2-1B-Q6_K.gguf",
+            label="Model",
+        ),
+        gr.Textbox(
+            value="You are Dolphin an AI assistant that helps humanity.",
+            label="System message",
+        ),
+        gr.Slider(minimum=1, maximum=8192, value=8192, step=1, label="Max tokens"),
+        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
+        gr.Slider(
+            minimum=0.1,
+            maximum=1.0,
+            value=0.95,
+            step=0.05,
+            label="Top-p",
+        ),
+        gr.Slider(
+            minimum=0,
+            maximum=100,
+            value=40,
+            step=1,
+            label="Top-k",
+        ),
+        gr.Slider(
+            minimum=0.0,
+            maximum=2.0,
+            value=1.1,
+            step=0.1,
+            label="Repetition penalty",
+        ),
+    ],
+    theme="Glass",
+    submit_btn="Send",
+    stop_btn="Stop",
+    title=title,
+    description=description,
+    chatbot=gr.Chatbot(scale=1, show_copy_button=True),
+    examples=examples,
+    cache_examples=True,
+    cache_mode="lazy",
+    flagging_mode="never",
+)
+
+
+# Launch the chat interface
+if __name__ == "__main__":
+    demo.launch(debug=False)