Update app.py

app.py

@@ -56,15 +56,20 @@ if article_file and synopsis_file:
             # Anonymization
             article_anon = anonymize_text(article)
             synopsis_anon = anonymize_text(synopsis)
+            # Estimate n_ctx
+            total_text = article_anon + synopsis_anon
+            estimated_tokens = int(len(total_text)/3.5)
+            n_ctx = estimated_tokens + 500
             
-            article_limit = 350000 # max_chars = 128000 * 3.5 (approx_chars_per_token) ≈ 448,000 characters;  448,000 - 98000(space for synopsis) = 350000
+            article_limit = 80000 # max_article_chars = 32,000 tokens×3.5 (approx_chars_per_token)≈112,000 characters; 112,000 - 32000(space for synopsis)= 80000
             
             # LLM feedback
             try:
                 llm = Llama(
                 model_path="./gemma-3-4b-it-qat-q4_0/gemma-3-4b-it-q4_0.gguf",
-                    n_ctx=5000,
-                   # n_batch=128
+                    n_ctx=n_ctx,
+                    n_threads=2,
+                    n_batch=128
                 )
                 
                 prompt = (