Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -23,6 +23,7 @@ def get_datasets():
|
|
23 |
|
24 |
all_datasets = get_datasets()
|
25 |
|
|
|
26 |
def get_split(dataset_name):
|
27 |
if dataset_name == "Communication Networks: unseen questions":
|
28 |
split = load_dataset("Short-Answer-Feedback/saf_communication_networks_english", split="test_unseen_questions")
|
@@ -37,6 +38,7 @@ def get_split(dataset_name):
|
|
37 |
if dataset_name == "Legal Domain: unseen answers":
|
38 |
split = load_dataset("Short-Answer-Feedback/saf_legal_domain_german", split="test_unseen_answers")
|
39 |
return split
|
|
|
40 |
|
41 |
def get_model(datasetname):
|
42 |
if datasetname == "Communication Networks: unseen questions" or datasetname == "Communication Networks: unseen answers":
|
@@ -47,6 +49,7 @@ def get_model(datasetname):
|
|
47 |
model = "Short-Answer-Feedback/mbart-finetuned-saf-legal-domain"
|
48 |
return model
|
49 |
|
|
|
50 |
def get_tokenizer(datasetname):
|
51 |
if datasetname == "Communication Networks: unseen questions" or datasetname == "Communication Networks: unseen answers":
|
52 |
tokenizer = "Short-Answer-Feedback/bart-finetuned-saf-communication-networks"
|
@@ -234,6 +237,102 @@ def load_data():
|
|
234 |
|
235 |
df = pd.concat([df, new_row])
|
236 |
return df
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
237 |
|
238 |
dataframe = load_data()
|
239 |
|
|
|
23 |
|
24 |
all_datasets = get_datasets()
|
25 |
|
26 |
+
'''
|
27 |
def get_split(dataset_name):
|
28 |
if dataset_name == "Communication Networks: unseen questions":
|
29 |
split = load_dataset("Short-Answer-Feedback/saf_communication_networks_english", split="test_unseen_questions")
|
|
|
38 |
if dataset_name == "Legal Domain: unseen answers":
|
39 |
split = load_dataset("Short-Answer-Feedback/saf_legal_domain_german", split="test_unseen_answers")
|
40 |
return split
|
41 |
+
'''
|
42 |
|
43 |
def get_model(datasetname):
|
44 |
if datasetname == "Communication Networks: unseen questions" or datasetname == "Communication Networks: unseen answers":
|
|
|
49 |
model = "Short-Answer-Feedback/mbart-finetuned-saf-legal-domain"
|
50 |
return model
|
51 |
|
52 |
+
'''
|
53 |
def get_tokenizer(datasetname):
|
54 |
if datasetname == "Communication Networks: unseen questions" or datasetname == "Communication Networks: unseen answers":
|
55 |
tokenizer = "Short-Answer-Feedback/bart-finetuned-saf-communication-networks"
|
|
|
237 |
|
238 |
df = pd.concat([df, new_row])
|
239 |
return df
|
240 |
+
'''
|
241 |
+
|
242 |
+
def get_rows(datasetname):
|
243 |
+
if datasetname == "Communication Networks: unseen questions":
|
244 |
+
row = pd.DataFrame(
|
245 |
+
{
|
246 |
+
'Model': get_model(datasetname),
|
247 |
+
'Dataset': datasetname,
|
248 |
+
'SacreBLEU': [2.4],
|
249 |
+
'ROUGE-2': [20.1],
|
250 |
+
'METEOR': [28.5],
|
251 |
+
'BERTScore': [36.6],
|
252 |
+
'Accuracy': [51.6],
|
253 |
+
'Weighted F1': [41.0],
|
254 |
+
'Macro F1': [27.9],
|
255 |
+
}
|
256 |
+
)
|
257 |
+
|
258 |
+
if datasetname == "Communication Networks: unseen answers":
|
259 |
+
row = pd.DataFrame(
|
260 |
+
{
|
261 |
+
'Model': get_model(datasetname),
|
262 |
+
'Dataset': datasetname,
|
263 |
+
'SacreBLEU': [36.0],
|
264 |
+
'ROUGE-2': [49.1],
|
265 |
+
'METEOR': [60.8],
|
266 |
+
'BERTScore': [69.5],
|
267 |
+
'Accuracy': [76.0],
|
268 |
+
'Weighted F1': [73.0],
|
269 |
+
'Macro F1: [53.4]'
|
270 |
+
}
|
271 |
+
)
|
272 |
+
if datasetname == "Micro Job: unseen questions":
|
273 |
+
row = pd.DataFrame(
|
274 |
+
{
|
275 |
+
'Model': get_model(datasetname),
|
276 |
+
'Dataset': datasetname,
|
277 |
+
'SacreBLEU': [0.3],
|
278 |
+
'ROUGE-2': [0.5],
|
279 |
+
'METEOR': [33.8],
|
280 |
+
'BERTScore': [31.3],
|
281 |
+
'Accuracy': [48.7],
|
282 |
+
'Weighted F1': [46.5],
|
283 |
+
'Macro F1': [40.6],
|
284 |
+
}
|
285 |
+
)
|
286 |
+
if datasetname == "Micro Job: unseen answers":
|
287 |
+
row = pd.DataFrame(
|
288 |
+
{
|
289 |
+
'Model': get_model(datasetname),
|
290 |
+
'Dataset': datasetname,
|
291 |
+
'SacreBLEU': [39.5],
|
292 |
+
'ROUGE-2': [29.8],
|
293 |
+
'METEOR': [63.3],
|
294 |
+
'BERTScore': [63.1],
|
295 |
+
'Accuracy': [80.1],
|
296 |
+
'Weighted F1': [80.3],
|
297 |
+
'Macro F1': [80.7],
|
298 |
+
}
|
299 |
+
)
|
300 |
+
if datasetname == "Legal Domain: unseen questions":
|
301 |
+
row = pd.DataFrame(
|
302 |
+
{
|
303 |
+
'Model': get_model(datasetname),
|
304 |
+
'Dataset': datasetname,
|
305 |
+
'SacreBLEU': [3.2],
|
306 |
+
'ROUGE-2': [5.0],
|
307 |
+
'METEOR': [20.0],
|
308 |
+
'BERTScore': [14.8],
|
309 |
+
'Accuracy': [60.7],
|
310 |
+
'Weighted F1': [55.3],
|
311 |
+
'Macro F1': [55.4],
|
312 |
+
}
|
313 |
+
)
|
314 |
+
if datasetname == "Legal Domain: unseen answers":
|
315 |
+
row = pd.DataFrame(
|
316 |
+
{
|
317 |
+
'Model': get_model(datasetname),
|
318 |
+
'Dataset': datasetname,
|
319 |
+
'SacreBLEU': [42.8],
|
320 |
+
'ROUGE-2': [43.7],
|
321 |
+
'METEOR': [58.2],
|
322 |
+
'BERTScore': [57.5],
|
323 |
+
'Accuracy': [81.0],
|
324 |
+
'Weighted F1': [80.1],
|
325 |
+
'Macro F1': [74.6],
|
326 |
+
}
|
327 |
+
)
|
328 |
+
return row
|
329 |
+
|
330 |
+
def load_data():
|
331 |
+
df = pd.DataFrame(columns=['Model', 'Dataset', 'SacreBLEU', 'ROUGE-2', 'METEOR', 'BERTScore', 'Accuracy', 'Weighted F1', 'Macro F1'])
|
332 |
+
for ds in all_datasets:
|
333 |
+
new_row = get_rows(ds)
|
334 |
+
df = pd.concat([df, new_row])
|
335 |
+
return df
|
336 |
|
337 |
dataframe = load_data()
|
338 |
|