Update app.py

app.py

@@ -2,13 +2,14 @@ import gradio as gr
 from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
 
-# Load model (choose Falcon or Mistral)
-model_name = "mistralai/Mistral-7B-Instruct-v0.2"
+# ✅ Free, open Llama 3 instruct model
+model_name = "meta-llama/Meta-Llama-3-8B-Instruct"
 tokenizer = AutoTokenizer.from_pretrained(model_name)
+
 model = AutoModelForCausalLM.from_pretrained(
     model_name,
-    device_map="auto",
-    torch_dtype=torch.float16
+    device_map="auto",          # auto = CPU in free Spaces
+    low_cpu_mem_usage=True
 )
 
 # Humanizer function
@@ -24,7 +25,7 @@ def humanize_text(ai_text):
         "Humanized version:"
     )
 
-    inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
+    inputs = tokenizer(prompt, return_tensors="pt")
     outputs = model.generate(
         **inputs,
         max_new_tokens=400,
@@ -41,7 +42,7 @@ def humanize_text(ai_text):
         f"Humanized: {humanized}\n\n"
         "Feedback:"
     )
-    fb_inputs = tokenizer(feedback_prompt, return_tensors="pt").to("cuda")
+    fb_inputs = tokenizer(feedback_prompt, return_tensors="pt")
     fb_outputs = model.generate(**fb_inputs, max_new_tokens=250, temperature=0.8, top_p=0.9)
     feedback = tokenizer.decode(fb_outputs[0], skip_special_tokens=True)