[to #42322933] add UT for NER&EL models

Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/10897188

tests/pipelines/test_named_entity_recognition.py

@@ -15,24 +15,260 @@ from modelscope.utils.test_utils import test_level
 
 
 class NamedEntityRecognitionTest(unittest.TestCase, DemoCompatibilityCheck):
+    language_examples = {
+        'zh':
+        '新华社北京二月十一日电（记者唐虹）',
+        'en':
+        'Italy recalled Marcello Cuttitta',
+        'ru':
+        'важным традиционным промыслом является производство пальмового масла .',
+        'fr':
+        'fer à souder électronique',
+        'es':
+        'el primer avistamiento por europeos de esta zona fue en 1606 , '
+        'en la expedición española mandada por luis váez de torres .',
+        'nl':
+        'in het vorige seizoen promoveerden sc cambuur , dat kampioen werd en go ahead eagles via de play offs .',
+        'tr':
+        'köyün pırasa kavurması ve içi yağlama ve akıtma adındaki hamur işleri meşhurdur . ; çörek ekmeği ; '
+        'diye adlandırdıkları mayasız ekmeği unutmamaklazım .',
+        'ko':
+        '국립진주박물관은 1984년 11월 2일 개관하였으며 한국 전통목조탑을 석조 건물로 형상화한 것으로 건축가 김수근 선생의 대표적 작품이다 .',
+        'fa':
+        'ﺞﻤﻋیﺕ ﺍیﻥ ﺎﺴﺗﺎﻧ ۳۰ ﻩﺯﺍﺭ ﻦﻓﺭ ﺎﺴﺗ ﻭ ﻢﻧﺎﺒﻋ ﻢﻬﻣی ﺍﺯ ﺲﻧگ ﺂﻬﻧ ﺩﺍﺭﺩ .',
+        'de':
+        'die szene beinhaltete lenny baker und christopher walken .',
+        'hi':
+        '१४९२ में एक चार्टर के आधार पर, उसके पिता ने उसे वाडोविस के उत्तराधिकारी के रूप में छोड़ दिया।',
+        'bn':
+        'যদিও গির্জার সবসময় রাজকীয় পিউ থাকত, তবে গির্জায় রাজকীয়ভাবে এটিই ছিল প্রথম দেখা।',
+        'multi':
+        '新华社北京二月十一日电（记者唐虹）',
+    }
+
+    all_modelcards_info = [
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_chinese-base-news',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_chinese-base-social_media',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_chinese-base-generic',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_chinese-base-resume',
+            'language': 'zh'
+        },
+        {
+            'model_id': 'damo/nlp_lstm_named-entity-recognition_chinese-news',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_lstm_named-entity-recognition_chinese-social_media',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_lstm_named-entity-recognition_chinese-generic',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_lstm_named-entity-recognition_chinese-resume',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_chinese-base-book',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_chinese-base-finance',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_chinese-base-game',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_chinese-base-bank',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_chinese-base-literature',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_chinese-base-cmeee',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_english-large-news',
+            'language': 'en'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_english-large-social_media',
+            'language': 'en'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_english-large-literature',
+            'language': 'en'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_english-large-politics',
+            'language': 'en'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_english-large-music',
+            'language': 'en'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_english-large-science',
+            'language': 'en'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_english-large-ai',
+            'language': 'en'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_english-large-wiki',
+            'language': 'en'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_chinese-large-generic',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_english-large-generic',
+            'language': 'en'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_multilingual-large-generic',
+            'language': 'multi'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_russian-large-generic',
+            'language': 'ru'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_spanish-large-generic',
+            'language': 'es'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_dutch-large-generic',
+            'language': 'nl'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_turkish-large-generic',
+            'language': 'tr'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_korean-large-generic',
+            'language': 'ko'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_farsi-large-generic',
+            'language': 'fa'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_german-large-generic',
+            'language': 'de'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_hindi-large-generic',
+            'language': 'hi'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_bangla-large-generic',
+            'language': 'bn'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_chinese-base-ecom',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_chinese-base-ecom-50cls',
+            'language': 'zh'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_english-large-ecom',
+            'language': 'en'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_russian-large-ecom',
+            'language': 'ru'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_french-large-ecom',
+            'language': 'fr'
+        },
+        {
+            'model_id':
+            'damo/nlp_raner_named-entity-recognition_spanish-large-ecom',
+            'language': 'es'
+        },
+        {
+            'model_id':
+            'damo/nlp_structbert_keyphrase-extraction_base-icassp2023-mug-track4-baseline',
+            'language': 'zh'
+        },
+    ]
 
     def setUp(self) -> None:
         self.task = Tasks.named_entity_recognition
         self.model_id = 'damo/nlp_raner_named-entity-recognition_chinese-base-news'
-
-    english_model_id = 'damo/nlp_raner_named-entity-recognition_english-large-ecom'
-    chinese_model_id = 'damo/nlp_raner_named-entity-recognition_chinese-large-generic'
-    tcrf_model_id = 'damo/nlp_raner_named-entity-recognition_chinese-base-news'
-    lcrf_model_id = 'damo/nlp_lstm_named-entity-recognition_chinese-news'
-    addr_model_id = 'damo/nlp_structbert_address-parsing_chinese_base'
-    lstm_model_id = 'damo/nlp_lstm_named-entity-recognition_chinese-generic'
-    sentence = '这与温岭市新河镇的一个神秘的传说有关。'
-    sentence_en = 'pizza shovel'
-    sentence_zh = '他 继 续 与 貝 塞 斯 達 遊 戲 工 作 室 在 接 下 来 辐 射 4 游 戏 。'
-    addr = '浙江省杭州市余杭区文一西路969号亲橙里'
-    addr1 = '浙江省西湖区灵隐隧道'
-    addr2 = '内蒙古自治区巴彦淖尔市'
-    ecom = '欧美单 秋季女装时尚百搭休闲修身 亚麻混纺短款 外套西装'
+        self.english_model_id = 'damo/nlp_raner_named-entity-recognition_english-large-ecom'
+        self.chinese_model_id = 'damo/nlp_raner_named-entity-recognition_chinese-large-generic'
+        self.tcrf_model_id = 'damo/nlp_raner_named-entity-recognition_chinese-base-news'
+        self.lcrf_model_id = 'damo/nlp_lstm_named-entity-recognition_chinese-news'
+        self.addr_model_id = 'damo/nlp_structbert_address-parsing_chinese_base'
+        self.lstm_model_id = 'damo/nlp_lstm_named-entity-recognition_chinese-generic'
+        self.sentence = '这与温岭市新河镇的一个神秘的传说有关。'
+        self.sentence_en = 'pizza shovel'
+        self.sentence_zh = '他 继 续 与 貝 塞 斯 達 遊 戲 工 作 室 在 接 下 来 辐 射 4 游 戏 。'
+        self.addr = '浙江省杭州市余杭区文一西路969号亲橙里'
+        self.addr1 = '浙江省西湖区灵隐隧道'
+        self.addr2 = '内蒙古自治区巴彦淖尔市'
+        self.ecom = '欧美单 秋季女装时尚百搭休闲修身 亚麻混纺短款 外套西装'
 
     @unittest.skipUnless(test_level() >= 2, 'skip test in current test level')
     def test_run_tcrf_by_direct_model_download(self):
@@ -222,6 +458,16 @@ class NamedEntityRecognitionTest(unittest.TestCase, DemoCompatibilityCheck):
         pipeline_ins = pipeline(task=Tasks.named_entity_recognition)
         print(pipeline_ins(input=self.sentence))
 
+    @unittest.skipUnless(test_level() >= 2, 'skip test in current test level')
+    def test_run_with_all_modelcards(self):
+        for item in self.all_modelcards_info:
+            model_id = item['model_id']
+            sentence = self.language_examples[item['language']]
+            with self.subTest(model_id=model_id):
+                pipeline_ins = pipeline(Tasks.named_entity_recognition,
+                                        model_id)
+                print(pipeline_ins(input=sentence))
+
     @unittest.skip('demo compatibility test is only enabled on a needed-basis')
     def test_demo_compatibility(self):
         self.compatibility_check()

tests/pipelines/test_sentence_embedding.py

@@ -36,6 +36,16 @@ class SentenceEmbeddingTest(unittest.TestCase):
         'sentences_to_compare': []
     }
 
+    el_model_id = 'damo/nlp_bert_entity-embedding_chinese-base'
+    el_inputs = {
+        'source_sentence': ['宋小宝小品《美人鱼》， [ENT_S] 大鹏 [ENT_E] 上演生死离别，关键时刻美人鱼登场'],
+        'sentences_to_compare': [
+            '董成鹏； 类型： Person； 别名： Da Peng， 大鹏;',
+            '超级飞侠； 类型： Work； 别名： 超飞， 출동!슈퍼윙스， Super Wings;',
+            '王源； 类型： Person； 别名： Roy;',
+        ]
+    }
+
     @unittest.skipUnless(test_level() >= 2, 'skip test in current test level')
     def test_run_by_direct_model_download(self):
         cache_path = snapshot_download(self.model_id)
@@ -77,6 +87,12 @@ class SentenceEmbeddingTest(unittest.TestCase):
         pipeline_ins = pipeline(task=Tasks.sentence_embedding)
         print(pipeline_ins(input=self.inputs))
 
+    @unittest.skipUnless(test_level() >= 2, 'skip test in current test level')
+    def test_run_with_el_model(self):
+        pipeline_ins = pipeline(
+            task=Tasks.sentence_embedding, model=self.el_model_id)
+        print(pipeline_ins(input=self.el_inputs))
+
 
 if __name__ == '__main__':
     unittest.main()

tests/pipelines/test_text_ranking.py

@@ -28,6 +28,19 @@ class TextRankingTest(unittest.TestCase):
         ]
     }
 
+    el_model_id = 'damo/nlp_bert_entity-matching_chinese-base'
+    el_inputs = {
+        'source_sentence': ['我是猫》([日]夏目漱石)【摘要 [ENT_S] 书评 [ENT_E]  试读】'],
+        'sentences_to_compare': [
+            '书评； 类型： Other； 别名： Book review; 三元组: 书评 # 外文名 # Book review $ 书评 # 摘要 # '
+            '书评，即评论并介绍书籍的文章，是以“书”为对象，实事求是的、有见识的分析书籍的形式和内容，探求创作的思想性、学术性、知识性和艺术性，从而在作者、读者和出版商之间构建信息交流的渠道。 $ 书评 # 定义 # '
+            '评论并介绍书籍的文章 $ 书评 # 中文名 # 书评 $ 书评 # 义项描述 # 书评 $ 书评 # 类型 # 应用写作的一种重要文体 $ 书评 # 标签 # 文学作品、文化、出版物、小说、书籍 $',
+            '摘要； 类型： Other； 别名： 摘， abstract， 书评; 三元组: 摘要 # 读音 # zhāi yào $ 摘要 # 外文名 # abstract $ 摘要 # 摘要 # '
+            '摘要又称概要、内容提要，意思是摘录要点或摘录下来的要点。 $  摘要 # 词目 # 摘要 $ 摘要 # 词性 # 动词，名词 $ 摘要 # 中文名 # 摘要 $ 摘要 # 别称 # 概要、内容提要 $ 摘要 '
+            '# 义项描述 # 摘要 $ 摘要 # 标签 # 文化、文学家、行业人物、法律术语、小说 $',
+        ]
+    }
+
     @unittest.skipUnless(test_level() >= 2, 'skip test in current test level')
     def test_run_by_direct_model_download(self):
         for model_id in self.models:
@@ -62,6 +75,12 @@ class TextRankingTest(unittest.TestCase):
         pipeline_ins = pipeline(task=Tasks.text_ranking)
         print(pipeline_ins(input=self.inputs))
 
+    @unittest.skipUnless(test_level() >= 2, 'skip test in current test level')
+    def test_run_with_el_model(self):
+        pipeline_ins = pipeline(
+            task=Tasks.text_ranking, model=self.el_model_id)
+        print(pipeline_ins(input=self.el_inputs))
+
 
 if __name__ == '__main__':
     unittest.main()