Update app.py

app.py

@@ -1,52 +1,63 @@
+from threading import Thread
+
 import gradio as gr
 from huggingface_hub import InferenceClient
+import spaces
+from transformers import AutoTokenizer, AutoModelForCausalLM, TextIteratorStreamer
 
-"""
-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("HuggingFaceH4/zephyr-7b-beta")
+model_id = "CohereForAI/aya-expanse-8b"
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id).to("cuda")
 
 
-def respond(
+@spaces.GPU
+def generate(
     message,
     history: list[tuple[str, str]],
     system_message,
     max_tokens,
     temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
-
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+    top_p):
+    
+    conversation = [{"role": "system", "content": system_message}]
+    for user, assistant in history:
+        conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
+    conversation.append({"role": "user", "content": message})
 
-    messages.append({"role": "user", "content": message})
+    input_ids = tokenizer.apply_chat_template(conversation, tokenize=True, add_generation_prompt=True, return_tensors="pt")
 
-    response = ""
+    # if input_ids.shape[1] > MAX_INPUT_TOKEN_LENGTH:
+    #     input_ids = input_ids[:, -MAX_INPUT_TOKEN_LENGTH:]
+    #     gr.Warning(f"Trimmed input from conversation as it was longer than {MAX_INPUT_TOKEN_LENGTH} tokens.")
+    input_ids = input_ids.to(model.device)
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
+    streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)
+            
+    generate_kwargs = dict(
+        {"input_ids": input_ids},
+        streamer=streamer,
+        max_new_tokens=max_tokens,
+        do_sample=True,
         top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+        temperature=temperature,
+    )
+    t = Thread(target=model.generate, kwargs=generate_kwargs)
+    t.start()
 
-        response += token
-        yield response
+    outputs = []
+    for text in streamer:
+        outputs.append(text)
+        yield "".join(outputs)
 
 
 """
 For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
 """
 demo = gr.ChatInterface(
-    respond,
+    generate,
+    cache_examples=False,
     additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
+        gr.Textbox(value="Je bent een vriendelijke, behulpzame 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(
@@ -57,6 +68,15 @@ demo = gr.ChatInterface(
             label="Top-p (nucleus sampling)",
         ),
     ],
+    examples=[
+        ["""Vraagje: welk woord hoort er niet in dit rijtje thuis: "auto, vliegtuig, geit, bus"?"""],
+        ["Schrijf een nieuwsbericht voor De Speld over de inzet van een kudde geiten door het Nederlands Forensisch Instituut"],
+        ["Wat zijn 3 leuke dingen om te doen als ik een weekendje naar Friesland ga?"],
+        ["Met wie trad clown Bassie op?"],
+        ["Kan je naar de maan fietsen?"],
+        ["Wat is het belang van open source taalmodellen?"],
+    ],
+    title="Aya Expanse 8B demo",
 )