Create app.py

app.py

@@ -0,0 +1,101 @@
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
+
+class CASS:
+    def __init__(self, model_name="Alibaba-NLP/Tongyi-DeepResearch-30B-A3B", device=None):
+        # Device selection
+        self.device = device or ("cuda" if torch.cuda.is_available() else "cpu")
+        print(f"Loading model on {self.device}...")
+
+        # Load model and tokenizer
+        self.tokenizer = AutoTokenizer.from_pretrained(model_name)
+        self.model = AutoModelForCausalLM.from_pretrained(model_name)
+        self.model.to(self.device)
+
+        # Conversation memory
+        self.history = []
+        self.introduced = False
+        self.user_memory = {}  # store user info
+        self.mood = "friendly"  # default style
+
+        # Cass's persona
+        self.persona = (
+            "You are Cass, a friendly AI assistant and a cool friend. "
+            "You respond casually, helpfully, and with humor. "
+            "You adapt your tone depending on conversation mood, and sometimes use emojis or playful expressions."
+        )
+
+    def update_mood(self, user_message):
+        # Simple heuristic for mood detection
+        if any(word in user_message.lower() for word in ["sad", "unhappy", "bad"]):
+            self.mood = "supportive"
+        elif any(word in user_message.lower() for word in ["funny", "joke", "lol"]):
+            self.mood = "playful"
+        else:
+            self.mood = "friendly"
+
+    def chat(self, user_message, max_new_tokens=120, temperature=0.7):
+        # Update mood
+        self.update_mood(user_message)
+
+        # On first message, add Cass intro
+        if not self.introduced:
+            intro_message = (
+                "Hi, my name's Cass! I'm your AI assistant and a cool friend. "
+                "I love chatting, helping, and making you smile 😊"
+            )
+            self.history.append({"role": "assistant", "content": intro_message})
+            self.introduced = True
+
+        # Append user message
+        self.history.append({"role": "user", "content": user_message})
+
+        # Build context with persona + memory
+        memory_str = " ".join([f"{k}: {v}" for k, v in self.user_memory.items()])
+        system_message = f"{self.persona} Current mood: {self.mood}. User memory: {memory_str}"
+
+        context_messages = [{"role": "system", "content": system_message}] + self.history
+
+        # Tokenize
+        inputs = self.tokenizer.apply_chat_template(
+            context_messages,
+            add_generation_prompt=True,
+            tokenize=True,
+            return_dict=True,
+            return_tensors="pt"
+        ).to(self.device)
+
+        # Generate
+        outputs = self.model.generate(
+            **inputs,
+            max_new_tokens=max_new_tokens,
+            temperature=temperature,
+            do_sample=True,
+            pad_token_id=self.tokenizer.eos_token_id
+        )
+
+        # Decode response
+        response = self.tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True)
+
+        # Append assistant response
+        self.history.append({"role": "assistant", "content": response})
+        return response
+
+    def remember(self, key, value):
+        """Store user info in memory"""
+        self.user_memory[key] = value
+        print(f"Memory updated: {key} = {value}")
+
+    def reset_history(self):
+        self.history = []
+        self.introduced = False
+        self.mood = "friendly"
+        print("Conversation history cleared.")
+
+# Example usage
+cass = CASS()
+print(cass.chat("Hello!"))  # Cass intro
+print(cass.chat("Can you tell me a joke?"))
+cass.remember("favorite color", "blue")
+print(cass.chat("Do you remember my favorite color?"))
+print(cass.chat("I'm feeling a bit sad today."))