Update app.py

app.py

@@ -1,8 +1,14 @@
 from flask import Flask, request, jsonify
-import ollama
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+MODEL_NAME = "meta-llama/Llama-3.1-8B-Instruct"
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+model = AutoModelForCausalLM.from_pretrained(MODEL_NAME).to("cpu")
 
 app = Flask(__name__)
 DEFAULT_TEMPERATURE = 0.7
+DEFAULT_MAX_TOKENS = 150
 DEFAULT_TOP_P = 0.95
 
 def generate_journal_suggestion(current_page):
@@ -12,13 +18,17 @@ def generate_journal_suggestion(current_page):
             Întrebarea ar trebui să încurajeze reflecția personală mai profundă, explorarea sentimentelor sau clarificarea obiectivelor."""
         )
         
-        response = ollama.chat(
-            model="llama3",
-            messages=[{"role": "user", "content": suggestion_prompt}],
-            options={"temperature": DEFAULT_TEMPERATURE, "top_p": DEFAULT_TOP_P}
+        input_ids = tokenizer(suggestion_prompt, return_tensors="pt").input_ids.to("cpu")
+
+        output_ids = model.generate(
+            input_ids,
+            max_length=DEFAULT_MAX_TOKENS,
+            temperature=DEFAULT_TEMPERATURE,
+            top_p=DEFAULT_TOP_P,
+            do_sample=True,
         )
-        
-        return response["message"]["content"]
+        suggestion_response = tokenizer.decode(output_ids[0], skip_special_tokens=True)
+
     except Exception as e:
         return f"Error: {str(e)}"