⚡️ Benchmark with stanza

Author: nipunsadvilkar

benchmarks/stanza_benchmark.py

@@ -0,0 +1,29 @@
+import stanza
+# stanza.download('en')
+nlp = stanza.Pipeline('en')
+
+from .english_golden_rules import GOLDEN_EN_RULES
+
+def stanza_english_benchmark(golden_rules):
+    global percent_score
+    total_rules = len(golden_rules)
+    score = 0
+    for rule in golden_rules:
+        text, expected = rule
+        doc = nlp(text)
+        segments = [e.text for e in doc.sentences]
+        if segments == expected:
+            score += 1
+    percent_score = (score / total_rules) * 100.0
+
+if __name__ == "__main__":
+    import timeit
+    time_taken = timeit.timeit("stanza_english_benchmark(GOLDEN_EN_RULES)",
+                  setup="from __main__ import stanza_english_benchmark, GOLDEN_EN_RULES",
+                  number=10)
+    print(f'{stanza.__name__} - v{stanza.__version__}')
+    print(f'GRS score: {percent_score:0.2f}%')
+    print(f'Speed: {time_taken*1000:>10.2f} ms')
+    stanza - v1.0.1
+    GRS score: 72.92%
+    Speed:  109839.63 ms