Upload 3 files

app.py

@@ -1,64 +1,125 @@
-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("HuggingFaceH4/zephyr-7b-beta")
+#!/usr/bin/env python
 
+import os
+from threading import Thread
+from typing import Iterator
+import spaces
+import gradio as gr
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+MAX_MAX_NEW_TOKENS = 2048
+DEFAULT_MAX_NEW_TOKENS = 1024
+MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "8192"))
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+model_id = "utter-project/EuroLLM-1.7B-Instruct"
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")
 
-    messages.append({"role": "user", "content": message})
+@spaces.GPU
+def generate(
+    message: str,
+    chat_history: list[tuple[str, str]],
+    max_new_tokens: int = 1024,
+    temperature: float = 0.06,
+    top_p: float = 0.95,
+    top_k: int = 40,
+    repetition_penalty: float = 1.2,
+) -> Iterator[str]:
+    conversation = []
+    for user, assistant in chat_history:
+        conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
+    conversation.append({"role": "user", "content": message})
 
-    response = ""
+    input_ids = tokenizer.apply_chat_template(conversation, add_generation_prompt=True, return_tensors="pt")
+    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_new_tokens,
+        do_sample=True,
         top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+        top_k=top_k,
+        temperature=temperature,
+        num_beams=1,
+        repetition_penalty=repetition_penalty,
+    )
+    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,
+chat_interface = gr.ChatInterface(
+    fn=generate,
+    chatbot=gr.Chatbot(height=450,
+                      label="utter-project/EuroLLM-1.7B-Instruct",
+                      show_share_button=True,
+                      ),
+    cache_examples=False,
     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,
+            label="Max new tokens",
+            minimum=1,
+            maximum=MAX_MAX_NEW_TOKENS,
+            step=1,
+            value=DEFAULT_MAX_NEW_TOKENS,
+        ),
+        gr.Slider(
+            label="Temperature",
+            minimum=0.05,
+            maximum=1.2,
             step=0.05,
+            value=0.2,
+        ),
+        gr.Slider(
             label="Top-p (nucleus sampling)",
+            minimum=0.05,
+            maximum=1.0,
+            step=0.05,
+            value=0.9,
+        ),
+        gr.Slider(
+            label="Top-k",
+            minimum=1,
+            maximum=1000,
+            step=1,
+            value=50,
         ),
+        gr.Slider(
+            label="Repetition penalty",
+            minimum=1.0,
+            maximum=2.0,
+            step=0.05,
+            value=1.2,
+        ),
+    ],
+    examples=[
+        ["Describe the significance of the Eiffel Tower in French culture and history."],
+        ["Что такое 'загадочная русская душа' и как это понятие отражается в русской литературе?"],  # Russian: What is the "mysterious Russian soul" and how is this concept reflected in Russian literature?
+        ["Jakie są najbardziej znane polskie tradycje bożonarodzeniowe?"],  # Polish: What are the most well-known Polish Christmas traditions?
+        ["Welche Rolle spielte die Hanse im mittelalterlichen Europa?"],  # German: What role did the Hanseatic League play in medieval Europe?
+        ["日本の茶道の精神と作法について説明してください。"]  # Japanese: Please explain the spirit and etiquette of Japanese tea ceremony.
     ],
+    title="utter-project/EuroLLM-1.7B-Instruct",
+    description="""utter-project/EuroLLM-1.7B-Instruct quick demo""",
+    submit_btn="Generate",
+    stop_btn="Stop",
+    retry_btn="🔄 Retry",
+    undo_btn="↩️ Undo",
+    clear_btn="🗑️ Clear",
 )
 
+with gr.Blocks(css="style.css") as demo:
+    chat_interface.render()
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.queue(max_size=20).launch()

requirements.txt

@@ -1 +1,9 @@
-huggingface_hub==0.25.2
+accelerate
+bitsandbytes
+gradio
+scipy
+sentencepiece
+spaces
+torch~=2.2.0
+transformers
+numpy<2

style.css

@@ -0,0 +1,16 @@
+h1 {
+  text-align: center;
+}
+
+#duplicate-button {
+  margin: auto;
+  color: white;
+  background: #1565c0;
+  border-radius: 100vh;
+}
+
+.contain {
+  max-width: 900px;
+  margin: auto;
+  padding-top: 1.5rem;
+}