Spaces:
Runtime error
Runtime error
domenicrosati
commited on
Commit
Β·
b7e15be
1
Parent(s):
a14da38
improve tokenization
Browse files
app.py
CHANGED
@@ -90,7 +90,7 @@ def find_source(text, docs):
|
|
90 |
for snippet in doc[1]:
|
91 |
if text in remove_html(snippet.get('snippet', '')):
|
92 |
new_text = text
|
93 |
-
for sent in remove_html(snippet.get('snippet', ''))
|
94 |
if text in sent:
|
95 |
new_text = sent
|
96 |
return {
|
@@ -103,7 +103,7 @@ def find_source(text, docs):
|
|
103 |
}
|
104 |
if text in remove_html(doc[3]):
|
105 |
new_text = text
|
106 |
-
for sent in remove_html(doc[3])
|
107 |
if text in sent:
|
108 |
new_text = sent
|
109 |
return {
|
@@ -206,8 +206,8 @@ with st.expander("Settings (strictness, context limit, top hits)"):
|
|
206 |
use_reranking = st.radio(
|
207 |
"Use Reranking? Reranking will rerank the top hits using semantic similarity of document and query.",
|
208 |
('yes', 'no'))
|
209 |
-
top_hits_limit = st.slider('Top hits? How many documents to use for reranking. Larger is slower but higher quality', 10, 300,
|
210 |
-
context_lim = st.slider('Context limit? How many documents to use for answering from. Larger is slower but higher quality', 10, 300,
|
211 |
|
212 |
# def paraphrase(text, max_length=128):
|
213 |
# input_ids = queryexp_tokenizer.encode(text, return_tensors="pt", add_special_tokens=True)
|
@@ -216,6 +216,22 @@ with st.expander("Settings (strictness, context limit, top hits)"):
|
|
216 |
# preds = '\n * '.join(queries)
|
217 |
# return preds
|
218 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
219 |
def run_query(query):
|
220 |
# if use_query_exp == 'yes':
|
221 |
# query_exp = paraphrase(f"question2question: {query}")
|
@@ -224,6 +240,10 @@ def run_query(query):
|
|
224 |
# * {query_exp}
|
225 |
# """)
|
226 |
|
|
|
|
|
|
|
|
|
227 |
# could also try fallback if there are no good answers by score...
|
228 |
limit = top_hits_limit or 100
|
229 |
context_limit = context_lim or 10
|
@@ -280,12 +300,9 @@ def run_query(query):
|
|
280 |
"score": result['score'],
|
281 |
"doi": support["supporting"]
|
282 |
})
|
283 |
-
|
284 |
-
|
285 |
-
|
286 |
-
}.values())
|
287 |
-
sorted_result = sorted(
|
288 |
-
sorted_result, key=lambda x: x['score'], reverse=True)
|
289 |
|
290 |
if confidence_threshold == 0:
|
291 |
threshold = 0
|
@@ -299,9 +316,11 @@ def run_query(query):
|
|
299 |
|
300 |
for r in sorted_result:
|
301 |
answer = r["answer"]
|
302 |
-
ctx = remove_html(r["context"])
|
303 |
-
|
304 |
-
|
|
|
|
|
305 |
score = round(r["score"], 4)
|
306 |
card(title, ctx, score, r['link'], r['doi'])
|
307 |
|
|
|
90 |
for snippet in doc[1]:
|
91 |
if text in remove_html(snippet.get('snippet', '')):
|
92 |
new_text = text
|
93 |
+
for sent in nltk.sent_tokenize(remove_html(snippet.get('snippet', ''))):
|
94 |
if text in sent:
|
95 |
new_text = sent
|
96 |
return {
|
|
|
103 |
}
|
104 |
if text in remove_html(doc[3]):
|
105 |
new_text = text
|
106 |
+
for sent in nltk.sent_tokenize(remove_html(doc[3])):
|
107 |
if text in sent:
|
108 |
new_text = sent
|
109 |
return {
|
|
|
206 |
use_reranking = st.radio(
|
207 |
"Use Reranking? Reranking will rerank the top hits using semantic similarity of document and query.",
|
208 |
('yes', 'no'))
|
209 |
+
top_hits_limit = st.slider('Top hits? How many documents to use for reranking. Larger is slower but higher quality', 10, 300, 10)
|
210 |
+
context_lim = st.slider('Context limit? How many documents to use for answering from. Larger is slower but higher quality', 10, 300, 5)
|
211 |
|
212 |
# def paraphrase(text, max_length=128):
|
213 |
# input_ids = queryexp_tokenizer.encode(text, return_tensors="pt", add_special_tokens=True)
|
|
|
216 |
# preds = '\n * '.join(queries)
|
217 |
# return preds
|
218 |
|
219 |
+
|
220 |
+
def group_results_by_context(results):
|
221 |
+
result_groups = {}
|
222 |
+
for result in results:
|
223 |
+
if result['context'] not in result_groups:
|
224 |
+
result_groups[result['context']] = result
|
225 |
+
result_groups[result['context']]['texts'] = []
|
226 |
+
|
227 |
+
result_groups[result['context']]['texts'].append(
|
228 |
+
result['answer']
|
229 |
+
)
|
230 |
+
if result['score'] > result_groups[result['context']]['score']:
|
231 |
+
result_groups[result['context']]['score'] = result['score']
|
232 |
+
return list(result_groups.values())
|
233 |
+
|
234 |
+
|
235 |
def run_query(query):
|
236 |
# if use_query_exp == 'yes':
|
237 |
# query_exp = paraphrase(f"question2question: {query}")
|
|
|
240 |
# * {query_exp}
|
241 |
# """)
|
242 |
|
243 |
+
# address period in highlitht avoidability. Risk factors
|
244 |
+
# address poor tokenization Deletions involving chromosome region 4p16.3 cause WolfHirschhorn syndrome (WHS, OMIM 194190) [Battaglia et al, 2001].
|
245 |
+
# address highlight html
|
246 |
+
|
247 |
# could also try fallback if there are no good answers by score...
|
248 |
limit = top_hits_limit or 100
|
249 |
context_limit = context_lim or 10
|
|
|
300 |
"score": result['score'],
|
301 |
"doi": support["supporting"]
|
302 |
})
|
303 |
+
|
304 |
+
grouped_results = group_results_by_context(results)
|
305 |
+
sorted_result = sorted(grouped_results, key=lambda x: x['score'], reverse=True)
|
|
|
|
|
|
|
306 |
|
307 |
if confidence_threshold == 0:
|
308 |
threshold = 0
|
|
|
316 |
|
317 |
for r in sorted_result:
|
318 |
answer = r["answer"]
|
319 |
+
ctx = remove_html(r["context"])
|
320 |
+
for answer in r['texts']:
|
321 |
+
ctx = ctx.replace(answer, f"<mark>{answer}</mark>")
|
322 |
+
# .replace( '<cite', '<a').replace('</cite', '</a').replace('data-doi="', 'href="https://scite.ai/reports/')
|
323 |
+
title = r.get("title", '')
|
324 |
score = round(r["score"], 4)
|
325 |
card(title, ctx, score, r['link'], r['doi'])
|
326 |
|