Update app.py

app.py

@@ -1,64 +1,65 @@
 #!/usr/bin/env python
 import gradio as gr
-from huggingface_hub import InferenceClient
 import spaces
-"""
-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("xi0v/aether-7b-chat-v1.0")
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline, StoppingCriteria, StoppingCriteriaList, TextIteratorStreamer
+import time
+import numpy as np
+from torch.nn import functional as F
+import os
+from threading import Thread
 
-@spaces.GPU
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+print(f"Starting to load the model to memory")
+m = AutoModelForCausalLM.from_pretrained(
+    "xi0v/aether-7b-chat-v1.0", torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32, trust_remote_code=False)
+tok = AutoTokenizer.from_pretrained("xi0v/aether-7b-chat-v1.0", trust_remote_code=False)
+# using CUDA for an optimal experience
+device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+m = m.to(device)
+print(f"Sucessfully loaded the model to the memory")
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
 
-    messages.append({"role": "user", "content": message})
+start_message = ""
 
-    response = ""
+def user(message, history):
+    # Append the user's message to the conversation history
+    return "", history + [[message, ""]]
 
-    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
+@spaces.GPU
+def chat(message, history):
+    chat = []
+    for item in history:
+        chat.append({"role": "user", "content": item[0]})
+        if item[1] is not None:
+            chat.append({"role": "assistant", "content": item[1]})
+    chat.append({"role": "user", "content": message})
+    messages = tok.apply_chat_template(chat, tokenize=False, add_generation_prompt=True)
+    # Tokenize the messages string
+    model_inputs = tok([messages], return_tensors="pt").to(device)
+    streamer = TextIteratorStreamer(
+        tok, timeout=10., skip_prompt=True, skip_special_tokens=True)
+    generate_kwargs = dict(
+        model_inputs,
+        streamer=streamer,
+        max_new_tokens=1024,
+        do_sample=True,
+        top_p=0.95,
+        top_k=1000,
+        temperature=0.75,
+        num_beams=1,
+    )
+    t = Thread(target=m.generate, kwargs=generate_kwargs)
+    t.start()
 
-        response += token
-        yield response
+    # Initialize an empty string to store the generated text
+    partial_text = ""
+    for new_text in streamer:
+        # print(new_text)
+        partial_text += new_text
+        # Yield an empty string to cleanup the message textbox and the updated conversation history
+        yield partial_text
 
-"""
-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="Your name is Aether, a 7 billion parameter large language model, you are a helpful and friendly chat assistant", label="System message"),
-        gr.Slider(minimum=1, maximum=4096, 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)",
-        ),
-    ],
-)
 
 
-if __name__ == "__main__":
-    demo.launch()
+demo = gr.ChatInterface(fn=chat, examples=["hello", "hola", "merhaba"], title="Stable LM 2 Zephyr 1.6b")
+demo.launch()