Update app.py

app.py

@@ -6,7 +6,7 @@ def install_and_import(package):
     try:
         __import__(package)
     except ImportError:
-        print(f"{package} yüklü değil, kuruluyor...")
+        print(f"{package} not installed, installing...")
         subprocess.check_call([sys.executable, "-m", "pip", "install", package])
 
 install_and_import("gradio")
@@ -17,11 +17,24 @@ import gradio as gr
 from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
 
-model_name = "Bertug1911/BrtGPT-124m-Base"
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(model_name)
+MODEL_OPTIONS = {
+    "BrtGPT-124m-Base": "Bertug1911/BrtGPT-124m-Base",
+}
+
+loaded_models = {}
+
+def load_model_and_tokenizer(model_key):
+    if model_key not in loaded_models:
+        print(f"Loading model {model_key}...")
+        model_name = MODEL_OPTIONS[model_key]
+        tokenizer = AutoTokenizer.from_pretrained(model_name)
+        model = AutoModelForCausalLM.from_pretrained(model_name)
+        loaded_models[model_key] = (tokenizer, model)
+    return loaded_models[model_key]
+
+def generate_text(prompt, temperature, top_k, max_new_tokens, model_key):
+    tokenizer, model = load_model_and_tokenizer(model_key)
 
-def generate_text(prompt, temperature, top_k, max_new_tokens):
     inputs = tokenizer(prompt, return_tensors="pt")
 
     start_time = time.time()
@@ -36,23 +49,33 @@ def generate_text(prompt, temperature, top_k, max_new_tokens):
 
     generated_text = tokenizer.decode(output[0], skip_special_tokens=False)
     generated_text = generated_text.replace(" ", "").replace("Ġ", " ")
-    return generated_text
 
-arayuz = gr.Interface(
+    tokenized_output = tokenizer.tokenize(generated_text)
+    token_count = len(tokenized_output)
+
+    elapsed = end_time - start_time
+    tokens_per_sec = token_count / elapsed if elapsed > 0 else 0
+
+    details = f"Token count: {token_count} | Tokens per second: {tokens_per_sec:.2f}"
+    return generated_text, details
+
+interface = gr.Interface(
     fn=generate_text,
     inputs=[
-        gr.Textbox(lines=2, placeholder="Write something here..."),
+        gr.Textbox(lines=2, placeholder="Enter your prompt here..."),
         gr.Slider(minimum=0.01, maximum=1.0, value=0.5, step=0.01, label="Temperature"),
         gr.Slider(minimum=1, maximum=50, value=10, step=1, label="Top-K"),
-        gr.Slider(minimum=1, maximum=150, value=15, step=1, label="Max New Tokens"),
+        gr.Slider(minimum=1, maximum=100, value=15, step=1, label="Max New Tokens"),
+        gr.Dropdown(choices=list(MODEL_OPTIONS.keys()), value="BrtGPT-124m-Base (Smartest and Fastest)", label="Select Model"),
     ],
     outputs=[
         "text",
-        gr.Textbox(label="Performans", interactive=False)
+        gr.Textbox(label="Performance", interactive=False)
     ],
     title="BrtGPT-124m-Base",
-    description="For CREATIVITY, Temperature: 0.4/0.7, Top-k: 10-15; For ACCUARYC: Temperature: 0.1/0.15, Top-k: 1-5"
+    description="Adjust the parameters, select the model, and generate text. (0.7 Temp and Top-k = 10 is good for CREATIVITY, 0.1/0.15 Temp. and Top-k = 1-5 is  good for ACCURACY"
 )
 
-arayuz.launch()
+interface.launch()
+