Create batch_test.py
Browse files- batch_test.py +40 -0
batch_test.py
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env python3
|
| 2 |
+
from model import AcoliModel
|
| 3 |
+
import json
|
| 4 |
+
|
| 5 |
+
def batch_test():
|
| 6 |
+
print("Loading model and test data...")
|
| 7 |
+
model = AcoliModel("./acoli-model")
|
| 8 |
+
|
| 9 |
+
# Test with first 10 examples from your dataset
|
| 10 |
+
test_samples = []
|
| 11 |
+
with open('../acoli-dataset/data.jsonl', 'r', encoding='utf-8') as f:
|
| 12 |
+
for i, line in enumerate(f):
|
| 13 |
+
if i >= 10: # Test only first 10 samples
|
| 14 |
+
break
|
| 15 |
+
data = json.loads(line)
|
| 16 |
+
test_samples.append(data)
|
| 17 |
+
|
| 18 |
+
print(f"\n=== Batch Testing on {len(test_samples)} samples ===")
|
| 19 |
+
|
| 20 |
+
correct = 0
|
| 21 |
+
for i, sample in enumerate(test_samples, 1):
|
| 22 |
+
text = sample['text']
|
| 23 |
+
true_label = sample['label']
|
| 24 |
+
|
| 25 |
+
prediction = model.predict(text)
|
| 26 |
+
predicted_label = int(prediction[0]['label'].split('_')[-1])
|
| 27 |
+
|
| 28 |
+
is_correct = (predicted_label == true_label)
|
| 29 |
+
if is_correct:
|
| 30 |
+
correct += 1
|
| 31 |
+
|
| 32 |
+
status = "✅" if is_correct else "❌"
|
| 33 |
+
print(f"{status} Sample {i}: True={true_label}, Predicted={predicted_label}")
|
| 34 |
+
print(f" Text: {text[:60]}...")
|
| 35 |
+
|
| 36 |
+
accuracy = correct / len(test_samples) * 100
|
| 37 |
+
print(f"\n📊 Accuracy: {correct}/{len(test_samples)} ({accuracy:.1f}%)")
|
| 38 |
+
|
| 39 |
+
if __name__ == "__main__":
|
| 40 |
+
batch_test()
|