Update app.py

app.py

@@ -1,27 +1,32 @@
-from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
+from flask import Flask, request, jsonify
+from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
-from flask import Flask, render_template, request, jsonify
 
-app = Flask(__name__)
+# Modell: Deutsch, öffentlich zugänglich
+MODEL_NAME = "dbmdz/german-gpt2"
 
-MODEL_NAME = "ml6team/german-gpt2"
+# Tokenizer & Modell laden
 tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
 model = AutoModelForCausalLM.from_pretrained(MODEL_NAME)
-generator = pipeline("text-generation", model=model, tokenizer=tokenizer, device=0 if torch.cuda.is_available() else -1)
 
-@app.route("/")
-def index():
-    return render_template("index.html")
+app = Flask(__name__)
 
 @app.route("/chat", methods=["POST"])
 def chat():
-    data = request.get_json()
+    data = request.json
     user_input = data.get("message", "")
-    if not user_input:
-        return jsonify({"response": "Bitte geben Sie eine Nachricht ein."})
 
-    output = generator(user_input, max_length=200, num_return_sequences=1, do_sample=True)[0]["generated_text"]
-    return jsonify({"response": output.strip()})
+    # Eingabe tokenisieren
+    input_ids = tokenizer.encode(user_input, return_tensors="pt")
+
+    # Modell antworten lassen
+    output_ids = model.generate(input_ids, max_new_tokens=100, do_sample=True, top_k=50)
+    response = tokenizer.decode(output_ids[0], skip_special_tokens=True)
+
+    # Antwort ohne ursprüngliche Eingabe zurückgeben
+    answer_only = response[len(user_input):].strip()
+
+    return jsonify({"response": answer_only})
 
 if __name__ == "__main__":
-    app.run(debug=True, host="0.0.0.0", port=8080)
+    app.run(debug=True)