File size: 14,886 Bytes
d742fde
 
 
 
e05ebfd
c0f38e6
e05ebfd
c0f38e6
e05ebfd
c0f38e6
e05ebfd
 
 
 
 
 
c0f38e6
e05ebfd
c0f38e6
e05ebfd
c0f38e6
e05ebfd
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d742fde
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
---
license: apache-2.0
---

from gliner import GLiNER

from utca.core import RenameAttribute

from utca.implementation.predictors import GLiNERPredictor, GLiNERPredictorConfig

from utca.implementation.tasks import (
    GLiNER as UTCAGLiNER,
    GLiNERPreprocessor,
    GLiNERRelationExtraction,
    GLiNERRelationExtractionPreprocessor,
)

import time

from typing import Dict, List

import json

def measure_time(func):
    def wrapper(*args, **kwargs):
        start_time = time.time()
        result = func(*args, **kwargs)
        end_time = time.time()
        execution_time = end_time - start_time
        print(f"Execution time of {func.__name__}: {execution_time:.6f} seconds")
        return result

    return wrapper



class GLiNERTester:
    def __init__(self, model_name: str = r"C:\Users\doren\PycharmProjects\GlinerFineTuning\data\checkpoint-100000", device: str = "cuda:0"):
        # Initialize the basic model for most tasks
        self.model = GLiNER.from_pretrained(model_name)

        # Initialize the relation extraction pipeline
        self.predictor = GLiNERPredictor(
            GLiNERPredictorConfig(
                model_name=model_name,
                device=device
            )
        )

        # Build the relation extraction pipeline
        self.relation_pipe = (
                UTCAGLiNER(
                    predictor=self.predictor,
                    preprocess=GLiNERPreprocessor(threshold=0.5)
                )
                | RenameAttribute("output", "entities")
                | GLiNERRelationExtraction(
            predictor=self.predictor,
            preprocess=(
                    GLiNERPreprocessor(threshold=0.5)
                    | GLiNERRelationExtractionPreprocessor()
            )
        )
        )

        self.results = {}
    @measure_time
    def test_ner(self) -> Dict:
        """Test Named Entity Recognition capabilities"""
        print("\nTesting NER...")

        text = """
        Microsoft was founded by Bill Gates and Paul Allen on April 4, 1975 to develop and sell BASIC interpreters 
        for the Altair 8800. During his career at Microsoft, Gates held the positions of chairman, 
        chief executive officer, president and chief software architect, while also being the largest 
        individual shareholder until May 2014.
        """

        labels = ["founder", "computer", "software", "position", "date"]

        start_time = time.time()
        entities = self.model.predict_entities(text, labels)
        duration = time.time() - start_time

        return {
            "task": "ner",
            "entities": [{"text": e["text"], "label": e["label"], "score": e["score"]} for e in entities],
            "duration": duration
        }
    @measure_time
    def test_relation_extraction(self) -> Dict:
        """Test Relation Extraction capabilities"""
        print("\nTesting Relation Extraction...")

        text = """
        Microsoft was founded by Bill Gates and Paul Allen on April 4, 1975 to develop and sell BASIC interpreters 
        for the Altair 8800. During his career at Microsoft, Gates held the positions of chairman, 
        chief executive officer, president and chief software architect.
        """

        start_time = time.time()
        result = self.relation_pipe.run({
            "text": text,
            "labels": ["organisation", "founder", "position", "date"],
            "relations": [{
                "relation": "founder",
                "pairs_filter": [("organisation", "founder")],
                "distance_threshold": 100,
            }, {
                "relation": "inception date",
                "pairs_filter": [("organisation", "date")],
            }, {
                "relation": "held position",
                "pairs_filter": [("founder", "position")],
            }]
        })
        duration = time.time() - start_time

        return {
            "task": "relation_extraction",
            "relations": result["output"],
            "duration": duration
        }
    @measure_time
    def test_qa(self) -> Dict:
        """Test Question Answering capabilities"""
        print("\nTesting Question Answering...")

        question = "Who was the CEO of Microsoft?"
        text = """
        Microsoft was founded by Bill Gates and Paul Allen on April 4, 1975, to develop and sell BASIC interpreters 
        for the Altair 8800. During his career at Microsoft, Gates held the positions of chairman, 
        chief executive officer, president and chief software architect, while also being the largest 
        individual shareholder until May 2014.
        """

        input_ = question + text
        labels = ["answer"]

        start_time = time.time()
        answers = self.model.predict_entities(input_, labels)
        duration = time.time() - start_time



        return {
            "task": "question_answering",
            "answers": [{"text": a["text"], "score": a["score"]} for a in answers],
            "duration": duration
        }
    @measure_time
    def test_summarization(self) -> Dict:
        """Test Summarization capabilities"""
        print("\nTesting Summarization...")

        text = """
        Several studies have reported its pharmacological activities, including anti-inflammatory, 
        antimicrobial, and antitumoral effects. The effect of E-anethole was studied in the osteosarcoma 
        MG-63 cell line, and the antiproliferative activity was evaluated by an MTT assay. It showed 
        a GI50 value of 60.25 μM with apoptosis induction through the mitochondrial-mediated pathway.
        """

        prompt = "Summarize the given text, highlighting the most important information:\n"
        input_ = prompt + text
        labels = ["summary"]

        start_time = time.time()
        summaries = self.model.predict_entities(input_, labels, threshold=0.1)
        duration = time.time() - start_time

        return {
            "task": "summarization",
            "summaries": [{"text": s["text"], "score": s["score"]} for s in summaries],
            "duration": duration
        }
    @measure_time
    def test_sentiment_extraction(self) -> Dict:
        """Test Sentiment Extraction capabilities"""
        print("\nTesting Sentiment Extraction...")

        text = """
        I recently purchased the Sony WH-1000XM4 headphones and I'm thoroughly impressed. 
        The noise-canceling is excellent, though the price is a bit high. The sound quality is amazing 
        but the app could use some improvements.
        """

        labels = ["positive sentiment", "negative sentiment"]

        start_time = time.time()
        sentiments = self.model.predict_entities(text, labels)
        duration = time.time() - start_time

        return {
            "task": "sentiment_extraction",
            "sentiments": [{"text": s["text"], "label": s["label"], "score": s["score"]} for s in sentiments],
            "duration": duration
        }
    @measure_time
    def test_entity_disambiguation(self) -> Dict:
        """Test Entity Disambiguation capabilities"""
        print("\nTesting Entity Disambiguation...")

        text = """
        Paris is the capital of France. Paris Hilton is an American media personality.
        Mercury is a planet in our solar system. Mercury is also a chemical element.
        """

        labels = ["location Paris", "person Paris", "planet Mercury", "element Mercury"]

        start_time = time.time()
        entities = self.model.predict_entities(text, labels)
        duration = time.time() - start_time

        return {
            "task": "entity_disambiguation",
            "entities": [{"text": e["text"], "label": e["label"], "score": e["score"]} for e in entities],
            "duration": duration
        }

    def run_all_tests(self) -> Dict:
        """Run all available tests and store results"""
        print("Starting GLiNER comprehensive test suite...")

        self.results = {
            "ner": self.test_ner(),
            "relation_extraction": self.test_relation_extraction(),
            "qa": self.test_qa(),
            "summarization": self.test_summarization(),
            "sentiment_extraction": self.test_sentiment_extraction(),
            "entity_disambiguation": self.test_entity_disambiguation()
        }

        # Save results to JSON file
        with open('gliner_test_results.json', 'w') as f:
            json.dump(self.results, f, indent=4)

        print("\nAll tests completed. Results saved to 'gliner_test_results.json'")
        return self.results


def main():
    # Initialize tester with GPU if available
    try:
        tester = GLiNERTester(device="cuda:0")
        print("Using GPU for testing")
    except:
        tester = GLiNERTester(device="cpu")
        print("Using CPU for testing")

    # Run all tests
    results = tester.run_all_tests()

    # Print summary of results
    print("\nTest Summary:")
    for task, result in results.items():
        print(f"\n{task.upper()}:")
        print(f"Duration: {result['duration']:.2f} seconds")
        print(f"Results: ", result)
        if 'entities' in result:
            print(f"Found {len(result['entities'])} entities")
        elif 'answers' in result:
            print(f"Found {len(result['answers'])} answers")
        elif 'summaries' in result:
            print(f"Generated {len(result['summaries'])} summary segments")
        elif 'sentiments' in result:
            print(f"Found {len(result['sentiments'])} sentiment expressions")


if __name__ == "__main__":
    main()


Test Summary:

NER:
Duration: 0.41 seconds
Results:  {'task': 'ner', 'entities': [{'text': 'Bill Gates', 'label': 'founder', 'score': 0.999995768070221}, {'text': 'Paul Allen', 'label': 'founder', 'score': 0.9999948740005493}, {'text': 'April 4, 1975', 'label': 'date', 'score': 0.9999996423721313}, {'text': 'BASIC interpreters', 'label': 'software', 'score': 0.9999961853027344}, {'text': 'Altair 8800', 'label': 'computer', 'score': 0.9999923706054688}, {'text': 'chairman', 'label': 'position', 'score': 0.9999326467514038}, {'text': 'chief executive officer', 'label': 'position', 'score': 0.9999247193336487}, {'text': 'president', 'label': 'position', 'score': 0.9999806880950928}, {'text': 'chief software architect', 'label': 'position', 'score': 0.9999625086784363}, {'text': 'largest \n        individual shareholder', 'label': 'position', 'score': 0.9741785526275635}], 'duration': 0.4105691909790039}
Found 10 entities

RELATION_EXTRACTION:
Duration: 0.31 seconds
Results:  {'task': 'relation_extraction', 'relations': [{'source': {'start': 9, 'end': 18, 'span': 'Microsoft', 'score': 0.9999996423721313, 'entity': 'organisation'}, 'relation': 'founder', 'target': {'start': 34, 'end': 44, 'span': 'Bill Gates', 'score': 0.9999998211860657, 'entity': 'founder'}, 'score': 0.9999523162841797}, {'source': {'start': 9, 'end': 18, 'span': 'Microsoft', 'score': 0.9999996423721313, 'entity': 'organisation'}, 'relation': 'founder', 'target': {'start': 49, 'end': 59, 'span': 'Paul Allen', 'score': 0.9999998807907104, 'entity': 'founder'}, 'score': 0.999999463558197}, {'source': {'start': 9, 'end': 18, 'span': 'Microsoft', 'score': 0.9999996423721313, 'entity': 'organisation'}, 'relation': 'inception date', 'target': {'start': 63, 'end': 76, 'span': 'April 4, 1975', 'score': 1.0, 'entity': 'date'}, 'score': 0.9999998807907104}, {'source': {'start': 167, 'end': 176, 'span': 'Microsoft', 'score': 0.9999998807907104, 'entity': 'organisation'}, 'relation': 'inception date', 'target': {'start': 63, 'end': 76, 'span': 'April 4, 1975', 'score': 1.0, 'entity': 'date'}, 'score': 0.9999998807907104}, {'source': {'start': 34, 'end': 44, 'span': 'Bill Gates', 'score': 0.9999998211860657, 'entity': 'founder'}, 'relation': 'held position', 'target': {'start': 206, 'end': 214, 'span': 'chairman', 'score': 0.9999998807907104, 'entity': 'position'}, 'score': 0.999997615814209}, {'source': {'start': 34, 'end': 44, 'span': 'Bill Gates', 'score': 0.9999998211860657, 'entity': 'founder'}, 'relation': 'held position', 'target': {'start': 225, 'end': 248, 'span': 'chief executive officer', 'score': 0.9999997019767761, 'entity': 'position'}, 'score': 0.9999843835830688}, {'source': {'start': 34, 'end': 44, 'span': 'Bill Gates', 'score': 0.9999998211860657, 'entity': 'founder'}, 'relation': 'held position', 'target': {'start': 250, 'end': 259, 'span': 'president', 'score': 0.9999998807907104, 'entity': 'position'}, 'score': 0.9999969005584717}, {'source': {'start': 34, 'end': 44, 'span': 'Bill Gates', 'score': 0.9999998211860657, 'entity': 'founder'}, 'relation': 'held position', 'target': {'start': 264, 'end': 288, 'span': 'chief software architect', 'score': 0.9999998807907104, 'entity': 'position'}, 'score': 0.9999908208847046}], 'duration': 0.30675745010375977}

QA:
Duration: 0.48 seconds
Results:  {'task': 'question_answering', 'answers': [{'text': 'Bill Gates', 'score': 0.9978553056716919}], 'duration': 0.4841444492340088}
Found 1 answers

SUMMARIZATION:
Duration: 0.42 seconds
Results:  {'task': 'summarization', 'summaries': [{'text': 'Several studies have reported its pharmacological activities, including anti-inflammatory, \n        antimicrobial, and antitumoral effects.', 'score': 0.8983121514320374}, {'text': 'The effect of E-anethole was studied in the osteosarcoma \n        MG-63 cell line, and the antiproliferative activity was evaluated by an MTT assay.', 'score': 0.7457365393638611}, {'text': '25 μM with apoptosis induction through the mitochondrial-mediated pathway.', 'score': 0.8508360981941223}], 'duration': 0.41564154624938965}
Generated 3 summary segments

SENTIMENT_EXTRACTION:
Duration: 0.36 seconds
Results:  {'task': 'sentiment_extraction', 'sentiments': [{'text': 'impressed', 'label': 'positive sentiment', 'score': 0.7771905660629272}, {'text': 'excellent', 'label': 'positive sentiment', 'score': 0.6963109374046326}, {'text': 'price is a bit high', 'label': 'negative sentiment', 'score': 0.8551780581474304}, {'text': 'amazing', 'label': 'positive sentiment', 'score': 0.6874173879623413}, {'text': 'app could use some improvements', 'label': 'negative sentiment', 'score': 0.7845857739448547}], 'duration': 0.358095645904541}
Found 5 sentiment expressions

ENTITY_DISAMBIGUATION:
Duration: 0.32 seconds
Results:  {'task': 'entity_disambiguation', 'entities': [{'text': 'capital of France', 'label': 'location Paris', 'score': 0.8064324855804443}, {'text': 'Paris Hilton', 'label': 'person Paris', 'score': 0.9987842440605164}, {'text': 'Mercury', 'label': 'planet Mercury', 'score': 0.9934960603713989}, {'text': 'Mercury', 'label': 'planet Mercury', 'score': 0.9940248131752014}, {'text': 'chemical element', 'label': 'element Mercury', 'score': 0.9640767574310303}], 'duration': 0.32335710525512695}
Found 5 entities