Update phishing_email_detection_gpt2.py

try a low seq_len

Author: david-thrower

phishing_email_detection_gpt2.py

@@ -34,7 +34,7 @@
 
 # Text encoding / embedding related constants
 
-MAX_SEQ_LENGTH = 1536 # (baseline from text classification model)
+MAX_SEQ_LENGTH = 500 # (baseline from text classification model)
 
 tokenizer_checkpoint = "HuggingFaceTB/SmolLM3-3B" # "HuggingFaceTB/SmolLM2-1.7B-Instruct" 
 tokenizer = AutoTokenizer.from_pretrained(tokenizer_checkpoint)