Update app.py
Browse files
app.py
CHANGED
|
@@ -532,25 +532,18 @@ class Hit(TypedDict):
|
|
| 532 |
score: float
|
| 533 |
text: str
|
| 534 |
|
| 535 |
-
# Best b 0.9
|
| 536 |
-
# Best k1: 0.4
|
| 537 |
-
best_b = 0.9
|
| 538 |
-
best_k1 = 0.4
|
| 539 |
-
sciq = load_sciq()
|
| 540 |
-
csc_bm25_index = CSCBM25Index.build_from_documents(
|
| 541 |
-
documents=iter(sciq.corpus),
|
| 542 |
-
ndocs=12160,
|
| 543 |
-
show_progress_bar=True,
|
| 544 |
-
k1=best_k1,
|
| 545 |
-
b=best_b
|
| 546 |
-
)
|
| 547 |
-
csc_bm25_index.save("output/csc_bm25_index")
|
| 548 |
-
|
| 549 |
demo: Optional[gr.Interface] = None # Assign your gradio demo to this variable
|
| 550 |
return_type = List[Hit]
|
| 551 |
|
| 552 |
## YOUR_CODE_STARTS_HERE
|
| 553 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 554 |
doc2text = {doc.collection_id: doc.text for doc in sciq.corpus}
|
| 555 |
|
| 556 |
def retrieve(query: str) -> List[Hit]:
|
|
|
|
| 532 |
score: float
|
| 533 |
text: str
|
| 534 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 535 |
demo: Optional[gr.Interface] = None # Assign your gradio demo to this variable
|
| 536 |
return_type = List[Hit]
|
| 537 |
|
| 538 |
## YOUR_CODE_STARTS_HERE
|
| 539 |
+
# Use default b, k1
|
| 540 |
+
csc_bm25_index = CSCBM25Index.build_from_documents(
|
| 541 |
+
documents=iter(sciq.corpus),
|
| 542 |
+
ndocs=12160,
|
| 543 |
+
show_progress_bar=True
|
| 544 |
+
)
|
| 545 |
+
csc_bm25_index.save("output/csc_bm25_index_default")
|
| 546 |
+
csc_bm25_retriever = CSCBM25Retriever(index_dir="output/csc_bm25_index_default")
|
| 547 |
doc2text = {doc.collection_id: doc.text for doc in sciq.corpus}
|
| 548 |
|
| 549 |
def retrieve(query: str) -> List[Hit]:
|