Update app.py

app.py

@@ -1,104 +1,100 @@
 from huggingface_hub import InferenceClient
 import gradio as gr
 import re
-from nltk.tokenize import sent_tokenize
 
 client = InferenceClient(
     "mistralai/Mixtral-8x7B-Instruct-v0.1"
 )
 
-def tokenize_sentences(file_content):
-    sentences = sent_tokenize(file_content.decode())
-    return sentences
-
-def generate_synthetic_data(prompt, sentences, data_size, toxicity_level, use_emoji):
-    synthetic_data = []
-    for sentence in sentences[:data_size]:
-        # Apply the prompt instructions to generate synthetic data from the sentence
-        synthetic_sentence = f"{prompt}: {sentence}"
-        
-        # Adjust toxicity level
-        if toxicity_level == "High":
-            synthetic_sentence = add_toxic_content(synthetic_sentence)
-        elif toxicity_level == "Low":
-            synthetic_sentence = remove_toxic_content(synthetic_sentence)
-        
-        # Add or remove emoji
-        if use_emoji:
-            synthetic_sentence = add_emojis(synthetic_sentence)
-        else:
-            synthetic_sentence = remove_emojis(synthetic_sentence)
-        
-        synthetic_data.append(synthetic_sentence)
-    return "\n".join(synthetic_data)
+def format_prompt(message, history):
+  prompt = "<s>"
+  for user_prompt, bot_response in history:
+    prompt += f"[INST] {user_prompt} [/INST]"
+    prompt += f" {bot_response}</s> "
+  prompt += f"[INST] {message} [/INST]"
+  return prompt
 
-def add_toxic_content(text):
-    # Add code to make the text more toxic
-    return text
+def generate(
+    prompt, history, system_prompt, temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0,
+):
+    temperature = float(temperature)
+    if temperature < 1e-2:
+        temperature = 1e-2
+    top_p = float(top_p)
 
-def remove_toxic_content(text):
-    # Add code to remove toxic content from the text
-    return text
+    generate_kwargs = dict(
+        temperature=temperature,
+        max_new_tokens=max_new_tokens,
+        top_p=top_p,
+        repetition_penalty=repetition_penalty,
+        do_sample=True,
+        seed=42,
+    )
 
-def add_emojis(text):
-    # Add code to add emojis to the text
-    return text
+    formatted_prompt = format_prompt(f"{system_prompt}, {prompt}", history)
+    stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
+    output = ""
 
-def remove_emojis(text):
-    # Add code to remove emojis from the text
-    return text
+    for response in stream:
+        output += response.token.text
+        yield output
+    return output
 
-def generate(prompt, max_data_size=100, toxicity_level="Neutral", use_emoji=False, files=None):
-    if files is not None:
-        file_contents = [file.decode() for file in files]
-        sentences = []
-        for content in file_contents:
-            sentences.extend(tokenize_sentences(content))
-        synthetic_data = generate_synthetic_data(prompt, sentences, max_data_size, toxicity_level, use_emoji)
-        return synthetic_data
-    else:
-        return "Please upload a file to generate synthetic data."
+def process_file(file):
+    text = file.decode("utf-8")
+    sentences = re.split(r'[.!?]+', text)
+    sentences = [s.strip() for s in sentences if s.strip()]
+    return sentences
 
 additional_inputs=[
     gr.Textbox(
-        label="Prompt for Synthetic Data Generation",
+        label="System Prompt",
         max_lines=1,
         interactive=True,
     ),
     gr.Slider(
-        label="Max Data Size",
-        value=100,
-        minimum=10,
-        maximum=1000,
-        step=10,
+        label="Temperature",
+        value=0.9,
+        minimum=0.0,
+        maximum=1.0,
+        step=0.05,
         interactive=True,
-        info="The maximum number of sentences to include in the synthetic data",
+        info="Higher values produce more diverse outputs",
     ),
-    gr.Radio(
-        label="Toxicity Level",
-        choices=["High", "Low", "Neutral"],
-        value="Neutral",
+    gr.Slider(
+        label="Max new tokens",
+        value=256,
+        minimum=0,
+        maximum=5120,
+        step=64,
         interactive=True,
-        info="Adjust the toxicity level of the synthetic data",
+        info="The maximum numbers of new tokens",
     ),
-    gr.Checkbox(
-        label="Use Emoji",
-        value=False,
+    gr.Slider(
+        label="Top-p (nucleus sampling)",
+        value=0.90,
+        minimum=0.0,
+        maximum=1,
+        step=0.05,
         interactive=True,
-        info="Add or remove emojis in the synthetic data",
+        info="Higher values sample more low-probability tokens",
     ),
-    gr.File(
-        label="Upload PDF or Document",
-        file_count="multiple",
-        file_types=[".pdf", ".doc", ".docx", ".txt"],
+    gr.Slider(
+        label="Repetition penalty",
+        value=1.2,
+        minimum=1.0,
+        maximum=2.0,
+        step=0.05,
         interactive=True,
-    )
+        info="Penalize repeated tokens",
+    ),
+    gr.File(label="Upload File", file_count="single"),
 ]
 
-gr.Interface(
+gr.ChatInterface(
     fn=generate,
-    inputs=additional_inputs,
-    outputs="text",
+    chatbot=gr.Chatbot(show_label=False, show_share_button=False, show_copy_button=True, likeable=True, layout="panel"),
+    additional_inputs=additional_inputs,
     title="Synthetic-data-generation-aze",
-    description="Generate synthetic data from uploaded files based on a given prompt and customization options.",
-).launch(show_api=False) 
+    concurrency_limit=20,
+).launch(show_api=False)