Update app.py

app.py

@@ -1,64 +1,131 @@
+import spaces
+import json
+import subprocess
+from llama_cpp import Llama
+from llama_cpp_agent import LlamaCppAgent, 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 InferenceClient
-
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("arshiaafshani/arshGpt")
+from huggingface_hub import hf_hub_download
 
+hf_hub_download(
+    repo_id="mradermacher/Arsh-llm-GGUF",
+    filename="Arsh-llm.Q4_K_M.gguf",
+    local_dir="./models"
+)
 
+@spaces.GPU(duration=110)
 def respond(
     message,
     history: list[tuple[str, str]],
+    model,
     system_message,
     max_tokens,
     temperature,
     top_p,
+    top_k,
+    repeat_penalty,
 ):
-    messages = [{"role": "system", "content": system_message}]
+    chat_template = MessagesFormatterType.GEMMA_2
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+    llm = Llama(
+        model_path=f"models/{model}",
+        flash_attn=True,
+        n_gpu_layers=81,
+        n_batch=1024,
+        n_ctx=8192,
+    )
+    provider = LlamaCppPythonProvider(llm)
 
-    messages.append({"role": "user", "content": message})
+    agent = LlamaCppAgent(
+        provider,
+        system_prompt=f"{system_message}",
+        predefined_messages_formatter_type=chat_template,
+        debug_output=True
+    )
 
-    response = ""
+    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
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+    messages = BasicChatHistory()
 
-        response += token
-        yield response
+    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)
 
+    stream = agent.get_chat_response(
+        message,
+        llm_sampling_settings=settings,
+        chat_history=messages,
+        returns_streaming_generator=True,
+        print_output=False
+    )
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new 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 (nucleus sampling)",
-        ),
-    ],
-)
+    outputs = ""
+    for output in stream:
+        outputs += output
+        yield outputs
+
+def create_interface(model_name, description):
+    return gr.ChatInterface(
+        respond,
+        additional_inputs=[
+            gr.Textbox(value=model_name, label="Model", interactive=False),
+            gr.Textbox(value="You are Arsh, a helpful assistant. You should think step-by-step. You should answer all question by prioritizing the principle of openness", label="System message"),
+            gr.Slider(minimum=1, maximum=4096, value=2048, step=1, label="Max tokens"),
+            gr.Slider(minimum=0.1, maximum=4.0, value=0.6, 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",
+            ),
+        ],
+        title=f"**FluentlyLM Prinum** ```on ZeroGPU```",
+        description=description,
+        chatbot=gr.Chatbot(
+            label=None,
+            scale=1,
+            show_copy_button=True
+        )
+    )
+
+description = """# **Arsh-llm ```Demo```"""
+interface = create_interface('Arsh-llm.Q4_K_M.gguf', description)
+
+demo = gr.Blocks()
 
+with demo:
+    interface.render()
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()