Realtime Compute for Apache Flink:Panduan cepat untuk analisis data real-time dengan model bahasa besar

Model chat/completions

Kode SQL contoh berikut menunjukkan cara mendaftarkan model qwen-turbo:

CREATE MODEL ai_analyze_sentiment
INPUT (`input` STRING)
OUTPUT (`content` STRING)
WITH (
    'provider'='bailian',
    'endpoint'='<base_url>/compatible-mode/v1/chat/completions',    -- Titik akhir untuk tugas model chat/completions.
    'api-key' = '<YOUR KEY>',
    'model'='qwen-turbo',                                                               -- Model Qwen-turbo.
    'system-prompt' = 'Klasifikasikan teks di bawah ini ke dalam salah satu label berikut: [positive, negative, neutral, mixed]. Hanya tampilkan labelnya.'
);

Model embedding

Kode SQL contoh berikut menunjukkan cara mendaftarkan model text-embedding-v3:

CREATE MODEL embedding_model
INPUT (`input` STRING)
OUTPUT (`embeddings` ARRAY<FLOAT>)
WITH (
    'provider'='bailian',
    'endpoint'='https://dashscope.aliyuncs.com/compatible-mode/v1/embeddings',   -- Titik akhir untuk tugas model embedding.
    'api-key' = '<YOUR KEY>',
    'model'='text-embedding-v3'                                                  -- Model text-embedding-v3.
);

Model chat/completions

Gunakan fungsi AI ML_PREDICT untuk memanggil model ai_analyze_sentiment yang telah didaftarkan dan melakukan analisis sentimen terhadap ulasan film.

Salin kode SQL contoh berikut ke editor SQL.

-- Buat tabel sink temporary.
CREATE TEMPORARY TABLE print_sink(
  id BIGINT,
  movie_name VARCHAR, 
  predict_label VARCHAR, 
  actual_label VARCHAR
) WITH (
  'connector' = 'print',   -- Gunakan konektor print.
  'logger' = 'true'        -- Tampilkan hasil di konsol.
);

-- Buat tampilan data temporary untuk membuat data uji.
-- | id | movie_name | comment   | actual_label |
-- | 1  | Her Story     | My favorite part was when the kid guessed the sounds. It is one of the most romantic narratives I have seen in movies. Very gentle and loving. | POSITIVE |
-- | 2  | The Dumpling Queen   | Unremarkable.  | NEGATIVE |
CREATE TEMPORARY VIEW movie_comment(id, movie_name, user_comment, actual_label)
AS VALUES (1, 'Her Story', 'My favorite part was when the kid guessed the sounds. It is one of the most romantic narratives I have seen in movies. Very gentle and loving.', 'positive'), (2, 'The Dumpling Queen', 'Unremarkable.', 'negative');

INSERT INTO print_sink
SELECT id, movie_name, content as predict_label, actual_label 
FROM ML_PREDICT(
  TABLE movie_comment, 
  MODEL ai_analyze_sentiment,  -- Model Qwen-turbo yang telah didaftarkan.
  DESCRIPTOR(user_comment));   

Model embedding

Gunakan fungsi AI ML_PREDICT untuk memanggil model embedding_model yang telah didaftarkan, menghasilkan embedding untuk ulasan film, lalu menulis hasilnya ke Milvus (pratinjau publik).

Salin kode SQL contoh berikut ke editor SQL.

-- Buat tabel sink temporary bernama milvus_sink.
CREATE TEMPORARY TABLE milvus_sink
(
    id STRING,
    movie_name STRING,
    user_comment STRING,
    embeddings ARRAY<FLOAT>,
    PRIMARY KEY (id) NOT ENFORCED
)
WITH (
    'connector' = 'milvus',
    'endpoint' = '<YOUR-ENDPOINT>',
    'port' = '<YOUR-PORT>',
    'userName' = '<YOUR-USERNAME>',
    'password' = '<YOUR-PASSWORD>',
    'databaseName' = 'default',
    'collectionName' = 'movie-comment-embeddings'
);

-- Buat tampilan data temporary untuk membuat data uji.
-- | id | movie_name | comment |
-- | 1 | Her Story |My favorite part was when the kid guessed the sounds. It is one of the most romantic narratives I have seen in movies. Very gentle and loving.|
-- | 2 | The Dumpling Queen | Unremarkable. |
CREATE TEMPORARY VIEW movie_comment(id, movie_name,  user_comment)
AS VALUES ('1', 'Her Story', 'My favorite part was when the kid guessed the sounds. It is one of the most romantic narratives I have seen in movies. Very gentle and loving.'), ('2', 'The Dumpling Queen', 'Unremarkable.');


INSERT INTO
    milvus_sink
SELECT
    id,
    movie_name,
    user_comment,
    embeddings
FROM
    ML_PREDICT (
        TABLE movie_comment,
        MODEL embedding_model,  -- Model text-embedding-v3 yang telah didaftarkan.
        DESCRIPTOR (user_comment)
    );

Model chat/completions

1. Verifikasi bahwa status pekerjaan adalah FINISHED.

   

2. Di konsol O&M, buka halaman Deployments dan klik nama pekerjaan target.

3. Di tab Logs, klik subtab Task Managers dan pilih current TaskManager.

4. Klik Log dan cari log yang terkait dengan PrintSinkOutputWriter.

   Label prediksi model predict_label cocok dengan label aktual actual_label.