model changes

app.py

@@ -1,32 +1,46 @@
-import gradio as gr
-from transformers import GPT2LMHeadModel, GPT2Tokenizer
 import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import gradio as gr
 
-# Load pre-trained model and tokenizer
-model_name = "gpt2"  # You can also try 'distilgpt2' for faster results
-tokenizer = GPT2Tokenizer.from_pretrained(model_name)
-model = GPT2LMHeadModel.from_pretrained(model_name)
+# Load the DeepSeek model and tokenizer
+model_name = "deepseek-ai/deepseek-llm-7b-base"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(
+    model_name,
+    torch_dtype=torch.float16,
+    device_map="auto"
+)
 model.eval()
 
 # Chat function
-def chat_with_gpt2(message, history=[]):
-    # Combine history with current input
-    input_text = " ".join([msg for pair in history for msg in pair]) + " " + message
-    inputs = tokenizer.encode(input_text, return_tensors="pt")
-
-    outputs = model.generate(inputs, max_length=150, pad_token_id=tokenizer.eos_token_id,
-                             no_repeat_ngram_size=2, do_sample=True, top_p=0.9, temperature=0.8)
+def chat(message, history=[]):
+    history_text = "".join([f"User: {u}\nAssistant: {a}\n" for u, a in history])
+    prompt = history_text + f"User: {message}\nAssistant:"
+    
+    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+    outputs = model.generate(
+        **inputs,
+        max_new_tokens=200,
+        temperature=0.7,
+        do_sample=True,
+        top_p=0.9,
+        repetition_penalty=1.1,
+        eos_token_id=tokenizer.eos_token_id,
+        pad_token_id=tokenizer.eos_token_id
+    )
 
-    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
-    response = response[len(input_text):].strip()
-    history.append((message, response))
-    return response, history
+    output_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    reply = output_text.split("Assistant:")[-1].strip()
+    history.append((message, reply))
+    return reply, history
 
-# Gradio Interface
-chat_interface = gr.ChatInterface(fn=chat_with_gpt2, title="GPT-2 Resume Chatbot",
-                                  description="A chatbot powered by GPT-2 for answering resume/CV-related queries.",
-                                  theme="default")
+# Gradio UI
+iface = gr.ChatInterface(
+    fn=chat,
+    title="DeepSeek Chatbot",
+    description="Chatbot using DeepSeek 7B LLM",
+    theme="default"
+)
 
-# Launch
 if __name__ == "__main__":
-    chat_interface.launch()
+    iface.launch()

requirements.txt

@@ -1,3 +1,4 @@
-torch
 transformers
+torch
 gradio
+accelerate