Update app.py

app.py

@@ -1,62 +1,67 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+from huggingface_hub import InferenceClient, login
+from transformers import TextStreamer
+import torch
+from unsloth import FastLanguageModel
 
 """
 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")
+login()
 
+model, tokenizer = FastLanguageModel.from_pretrained(
+    model_name = "aayushpuri01/Llama-3.1-8B-Threat-Intelligent-v2",
+    max_seq_length = 2048,
+    device_map = "auto"
+)
+
+
+FastLanguageModel.for_inference(model)
+
+prompt_style = """You are a cybersecurity genius and expert threat hunter and analyst who can answer about any level of cybersecurity scenarios.
+Based on the given Instruction and Input, generate appropriate Output.
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+### Instruction:
+Please analyse the given scenario, provide diagnosis of the situation in between <diagnosis></diagnosis>. Write Solutions in between <solution></solution>.
+
+### Input:
+{}
+
+### Output:
+{}
+"""
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+def generate_response(scenario):
+    formatted_prompt = prompt_style.format(scenario, "")
 
-    messages.append({"role": "user", "content": message})
+    inputs = tokenizer(
+        [formatted_prompt],
+        return_tensors = "pt",
+    ).to("cuda" if torch.cuda.is_available() else "cpu")
 
-    response = ""
+    text_streamer = TextStreamer(tokenizer)
 
-    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
+    outputs = model.generate(**inputs, streamer=text_streamer, max_new_tokens=1028)
 
-        response += token
-        yield response
+    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
 
+    output_start = response.find("### Output:") + len("### Output:\n")
+    output_text = response[output_start:].strip()
 
+    return output_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="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)",
-        ),
-    ],
+    fn = generate_response,
+    inputs=gr.Textbox(
+        label="Cyberthreat scenario",
+        placeholder="Enter a scene (e.g, 'Cryptowall 2.0 began using the Tor anonymity network...')",
+        lines=5),
+    outputs = gr.Markdown(label="Analysis and Solutions"),
+    title = "Threat Intelligence with Llama-3.1-8B",
+    description = "Enter a cybersecurity scenario to get a detailed analysis and solutions from a fine tuned LLM model",
+    theme = "huggingface"
 )