Update app.py
Browse files
app.py
CHANGED
|
@@ -64,8 +64,13 @@ def simplify_document(pdf_file):
|
|
| 64 |
|
| 65 |
takeaways = []
|
| 66 |
for idx, chunk in enumerate(chunks, 1):
|
| 67 |
-
#
|
| 68 |
-
prompt =
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 69 |
inputs = tokenizer(
|
| 70 |
prompt, return_tensors="pt", max_length=512, truncation=True
|
| 71 |
).to(device)
|
|
@@ -83,9 +88,9 @@ def simplify_document(pdf_file):
|
|
| 83 |
)
|
| 84 |
|
| 85 |
summary = tokenizer.decode(output_tokens[0], skip_special_tokens=True)
|
| 86 |
-
|
| 87 |
-
# Ensure output finishes cleanly without abrupt cutoffs
|
| 88 |
summary = summary.strip()
|
|
|
|
|
|
|
| 89 |
if summary and not summary.endswith((".", "!", "?")):
|
| 90 |
last_period = summary.rfind(".")
|
| 91 |
if last_period > 20:
|
|
|
|
| 64 |
|
| 65 |
takeaways = []
|
| 66 |
for idx, chunk in enumerate(chunks, 1):
|
| 67 |
+
# Explicit structured prompt instruction
|
| 68 |
+
prompt = (
|
| 69 |
+
"Summarize the RTI/civic document in simple, clear bullet points for an"
|
| 70 |
+
" ordinary citizen. Separate the Subject, Timeframe, Fees Paid, and"
|
| 71 |
+
f" Delivery Mode clearly: {chunk}"
|
| 72 |
+
)
|
| 73 |
+
|
| 74 |
inputs = tokenizer(
|
| 75 |
prompt, return_tensors="pt", max_length=512, truncation=True
|
| 76 |
).to(device)
|
|
|
|
| 88 |
)
|
| 89 |
|
| 90 |
summary = tokenizer.decode(output_tokens[0], skip_special_tokens=True)
|
|
|
|
|
|
|
| 91 |
summary = summary.strip()
|
| 92 |
+
|
| 93 |
+
# Clean up any trailing broken sentences
|
| 94 |
if summary and not summary.endswith((".", "!", "?")):
|
| 95 |
last_period = summary.rfind(".")
|
| 96 |
if last_period > 20:
|