gardarjuto
commited on
Commit
•
6aa2b20
1
Parent(s):
1d38d16
add blacklisted examples for inflection
Browse files
quiz.py
CHANGED
@@ -22,6 +22,19 @@ BENCHMARKS = {
|
|
22 |
"name": "Fallbeygingar",
|
23 |
"path": "mideind/icelandic-inflection-all-flat",
|
24 |
"type": "free_text",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
},
|
26 |
"icelandic-belebele": {
|
27 |
"name": "Lesskilningur",
|
@@ -182,6 +195,13 @@ class BenchmarkQuiz:
|
|
182 |
icelandic_sentence_gec_preprocessing(sample) for sample in samples
|
183 |
]
|
184 |
elif benchmark_name == "icelandic-inflection-all":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
185 |
samples = [inflection_all_preprocessing(sample) for sample in samples]
|
186 |
elif benchmark_name == "icelandic-belebele":
|
187 |
samples = [belebele_preprocessing(sample) for sample in samples]
|
|
|
22 |
"name": "Fallbeygingar",
|
23 |
"path": "mideind/icelandic-inflection-all-flat",
|
24 |
"type": "free_text",
|
25 |
+
"blacklisted_noun_phrases": [
|
26 |
+
"hágæða sprengjutilræði",
|
27 |
+
"óstöðvandi geðröskun",
|
28 |
+
"allsber meirihluti",
|
29 |
+
"geðsjúkt álagsstýrikerfi",
|
30 |
+
"kynþokkafullt starfsvið",
|
31 |
+
"lettneskur þræll",
|
32 |
+
"nígerískt meyjarhaft",
|
33 |
+
"kynæsandi málvísindamaður",
|
34 |
+
"kynþokkafullur menntaskólakennari",
|
35 |
+
"lóðrétt forhúð",
|
36 |
+
"vandþrædd hvatabuska",
|
37 |
+
],
|
38 |
},
|
39 |
"icelandic-belebele": {
|
40 |
"name": "Lesskilningur",
|
|
|
195 |
icelandic_sentence_gec_preprocessing(sample) for sample in samples
|
196 |
]
|
197 |
elif benchmark_name == "icelandic-inflection-all":
|
198 |
+
while any(
|
199 |
+
sample["noun_phrase"] in BENCHMARKS[benchmark_name]["blacklisted_noun_phrases"]
|
200 |
+
for sample in samples
|
201 |
+
):
|
202 |
+
print(samples)
|
203 |
+
random_indices = random.sample(range(len(dataset)), 5)
|
204 |
+
samples = dataset.select(random_indices)
|
205 |
samples = [inflection_all_preprocessing(sample) for sample in samples]
|
206 |
elif benchmark_name == "icelandic-belebele":
|
207 |
samples = [belebele_preprocessing(sample) for sample in samples]
|