Java 类名：com.alibaba.alink.pipeline.regression.BertTextPairRegressor
Python 类名：BertTextPairRegressor
Bert 文本对回归。
| 名称 | 中文名称 | 描述 | 类型 | 是否必须？ | 取值范围 | 默认值 |
|---|---|---|---|---|---|---|
| labelCol | 标签列名 | 输入表中的标签列名 | String | ✓ | ||
| predictionCol | 预测结果列名 | 预测结果列名 | String | ✓ | ||
| textCol | 文本列 | 文本列 | String | ✓ | ||
| textPairCol | 文本对列 | 文本对列 | String | ✓ | ||
| batchSize | 数据批大小 | 数据批大小 | Integer | 32 | ||
| bertModelName | BERT模型名字 | BERT模型名字： Base-Chinese,Base-Multilingual-Cased,Base-Uncased,Base-Cased | String | “Base-Chinese” | ||
| checkpointFilePath | 保存 checkpoint 的路径 | 用于保存中间结果的路径，将作为 TensorFlow 中 Estimator 的 model_dir 传入，需要为所有 worker 都能访问到的目录 |
String | null | ||
| customConfigJson | 自定义参数 | 对应 https://github.com/alibaba/EasyTransfer/blob/master/easytransfer/app_zoo/app_config.py 中的config_json | String | |||
| inferBatchSize | 推理数据批大小 | 推理数据批大小 | Integer | 256 | ||
| intraOpParallelism | Op 间并发度 | Op 间并发度 | Integer | 4 | ||
| learningRate | 学习率 | 学习率 | Double | 0.001 | ||
| maxSeqLength | 句子截断长度 | 句子截断长度 | Integer | 128 | ||
| modelFilePath | 模型的文件路径 | 模型的文件路径 | String | null | ||
| numEpochs | epoch 数 | epoch 数 | Double | 0.01 | ||
| numFineTunedLayers | 微调层数 | 微调层数 | Integer | 1 | ||
| numPSs | PS 角色数 | PS 角色的数量。值未设置时，如果 Worker 角色数也未设置，则为作业总并发度的 1/4（需要取整），否则为总并发度减去 Worker 角色数。 | Integer | null | ||
| numWorkers | Worker 角色数 | Worker 角色的数量。值未设置时，如果 PS 角色数也未设置，则为作业总并发度的 3/4（需要取整），否则为总并发度减去 PS 角色数。 | Integer | null | ||
| overwriteSink | 是否覆写已有数据 | 是否覆写已有数据 | Boolean | false | ||
| predictionDetailCol | 预测详细信息列名 | 预测详细信息列名 | String | |||
| pythonEnv | Python 环境路径 | Python 环境路径，一般情况下不需要填写。如果是压缩文件，需要解压后得到一个目录，且目录名与压缩文件主文件名一致，可以使用 http://, https://, oss://, hdfs:// 等路径；如果是目录，那么只能使用本地路径，即 file://。 | String | "" | ||
| removeCheckpointBeforeTraining | 是否在训练前移除 checkpoint 相关文件 | 是否在训练前移除 checkpoint 相关文件用于重新训练，只会删除必要的文件 | Boolean | null | ||
| reservedCols | 算法保留列名 | 算法保留列 | String[] | null | ||
| modelStreamFilePath | 模型流的文件路径 | 模型流的文件路径 | String | null | ||
| modelStreamScanInterval | 扫描模型路径的时间间隔 | 描模型路径的时间间隔，单位秒 | Integer | 10 | ||
| modelStreamStartTime | 模型流的起始时间 | 模型流的起始时间。默认从当前时刻开始读。使用yyyy-mm-dd hh:mm:ss.fffffffff格式，详见Timestamp.valueOf(String s) | String | null |
** 以下代码仅用于示意，可能需要修改部分代码或者配置环境后才能正常运行！**
url = "http://alink-algo-packages.oss-cn-hangzhou-zmf.aliyuncs.com/data/MRPC/train.tsv"
schemaStr = "f_quality double, f_id_1 string, f_id_2 string, f_string_1 string, f_string_2 string"
data = CsvSourceBatchOp() \
.setFilePath(url) \
.setSchemaStr(schemaStr) \
.setFieldDelimiter("\t") \
.setIgnoreFirstLine(True) \
.setQuoteChar(None)
data = ShuffleBatchOp().linkFrom(data)
regressor = BertTextPairRegressor() \
.setTextCol("f_string_1").setTextPairCol("f_string_2").setLabelCol("f_quality") \
.setNumEpochs(0.1) \
.setMaxSeqLength(32) \
.setNumFineTunedLayers(1) \
.setBertModelName("Base-Uncased") \
.setPredictionCol("pred")
model = regressor.fit(data)
predict = model.transform(data.firstN(300))
predict.print()
import com.alibaba.alink.operator.batch.BatchOperator;
import com.alibaba.alink.operator.batch.dataproc.ShuffleBatchOp;
import com.alibaba.alink.operator.batch.source.CsvSourceBatchOp;
import com.alibaba.alink.pipeline.regression.BertRegressionModel;
import com.alibaba.alink.pipeline.regression.BertTextPairRegressor;
import org.junit.Test;
public class BertTextPairRegressorTest {
@Test
public void test() throws Exception {
String url = "http://alink-algo-packages.oss-cn-hangzhou-zmf.aliyuncs.com/data/MRPC/train.tsv";
String schemaStr = "f_quality double, f_id_1 string, f_id_2 string, f_string_1 string, f_string_2 string";
BatchOperator <?> data = new CsvSourceBatchOp()
.setFilePath(url)
.setSchemaStr(schemaStr)
.setFieldDelimiter("\t")
.setIgnoreFirstLine(true)
.setQuoteChar(null);
data = new ShuffleBatchOp().linkFrom(data);
BertTextPairRegressor regressor = new BertTextPairRegressor()
.setTextCol("f_string_1").setTextPairCol("f_string_2").setLabelCol("f_quality")
.setNumEpochs(0.1)
.setMaxSeqLength(32)
.setNumFineTunedLayers(1)
.setBertModelName("Base-Uncased")
.setPredictionCol("pred");
BertRegressionModel model = regressor.fit(data);
BatchOperator <?> predict = model.transform(data.firstN(300));
predict.print();
}
}
| FIELD1 | f_quality | f_id_1 | f_id_2 | f_string_1 | f_string_2 | |
|---|---|---|---|---|---|---|
| 0 | 1.0 | 2224622 | 2224687 | The indictment ``strikes at one of the very to… | The newly unsealed 32-count indictment alleges… | |
| 1 | 0.0 | 58543 | 58516 | The tech-heavy Nasdaq Stock Markets composite … | The Nasdaq Composite index , full of technolog… | |
| 2 | 1.0 | 1048622 | 1048759 | Rescue workers had to scrounge for boats to re… | Wagner and Wolford both said that rescue worke… | |
| 3 | 1.0 | 2733687 | 2733747 | Asked about the controversy , Bloomberg said ,… | “ I didn ’t see either one today , but if they… | |4 |0.0 |3270421 |3270325 |At 10 p.m. , Odette was centered about 295 mil…|At 10 p.m. Thursday , Odette was about 295 mil… | |… |… |… |… |… |… | |295 |0.0 |2187868 |2187826 |” They were brutally beaten , and it ’s really… |
" It ’s a wonder that Porchia was the only dec… |
| 296 | 0.0 | 257789 | 257726 | Knight had 29 interceptions in his six years w… | Knight has 29 interceptions in six seasons , a… | |
| 297 | 1.0 | 3128160 | 3128278 | It will appear in the next few weeks on the We… | Details of the research will appear in a futur… | |
| 298 | 1.0 | 1125898 | 1125882 | Following the ATP ’s notification by Hewitt ’s… | Following ATP ’s notification by Hewitt ’s law… | |
| 299 | 0.0 | 3041807 | 3041719 | Details of the research appear in the Nov. 5 i… | The results , published in the Journal of the … |