Polar_AI integrates several common model services from Alibaba Cloud Model Studio. It abstracts the technical details of model calls to simplify their use. You can perform tasks such as text embedding, text understanding, intelligent conversation, and content creation without moving data out of the database.
You can run the following SQL command to view the built-in models:
SELECT model_seq,model_id,model_name,model_url FROM polar_ai._ai_models; The following output is returned:
-----------+---------------------------------------------
1 | _dashscope/text_embedding/text_embedding_v2 | text-embedding-v2 | https://dashscope.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding
2 | _dashscope/text_embedding/text_embedding_v3 | text-embedding-v3 | https://dashscope.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding
3 | _dashscope/text-classfication/opennlu-v1 | opennlu-v1 | https://dashscope.aliyuncs.com/api/v1/services/nlp/nlu/understanding
(3 rows)The built-in models are primarily integrations of Natural Language Processing (NLP) models from Alibaba Cloud Model Studio. The models and their features are described below:
_dashscope/text_embedding/text_embedding_v2: Text and multimodal embedding. Converts text from languages such as Chinese, English, Spanish, French, Portuguese, Indonesian, Japanese, Korean, German, and Russian into embeddings. The output vector dimension is 1536.
_dashscope/text_embedding/text_embedding_v3: Text and multimodal embedding. Extends the features of text_embedding_v2 with support for over 50 languages. The default output vector dimension is 1024.
_dashscope/text-classfication/opennlu-v1: Text understanding. Performs zero-shot text understanding tasks in Chinese and English, such as information extraction and text classification.