update demo for preset net

demo/text_classification/predict.py

@@ -34,6 +34,7 @@ parser.add_argument("--batch_size",     type=int,   default=1, help="Total examp
 parser.add_argument("--max_seq_len", type=int, default=512, help="Number of words of the longest seqence.")
 parser.add_argument("--use_gpu", type=ast.literal_eval, default=False, help="Whether use GPU for finetuning, input should be True or False")
 parser.add_argument("--use_data_parallel", type=ast.literal_eval, default=False, help="Whether use data parallel.")
+parser.add_argument("--network", type=str, default='bilstm', help="Preset network which was connected after Transformer model, such as ERNIE, BERT ,RoBERTa and ELECTRA.")
 args = parser.parse_args()
 # yapf: enable.
 
@@ -59,7 +60,7 @@ if __name__ == '__main__':
     # Construct transfer learning network
     # Use "pooled_output" for classification tasks on an entire sentence.
     # Use "sequence_output" for token-level output.
-    pooled_output = outputs["pooled_output"]
+    pooled_output = outputs["sequence_output"]
 
     # Setup feed list for data feeder
     # Must feed all the tensor of module need
@@ -79,10 +80,15 @@ if __name__ == '__main__':
         strategy=hub.AdamWeightDecayStrategy())
 
     # Define a classfication finetune task by PaddleHub's API
+    # network choice: bilstm, bow, cnn, dpcnn, gru, lstm
+    # If you wanna add network after ERNIE/BERT/RoBERTa/ELECTRA module,
+    # you must use the outputs["sequence_output"] as the feature of TextClassifierTask
+    # rather than outputs["pooled_output"]
     cls_task = hub.TextClassifierTask(
         data_reader=reader,
         feature=pooled_output,
         feed_list=feed_list,
+        network=args.network,
         num_classes=dataset.num_labels,
         config=config)
 

demo/text_classification/run_classifier.sh

 export FLAGS_eager_delete_tensor_gb=0.0
-export CUDA_VISIBLE_DEVICES=0
+export CUDA_VISIBLE_DEVICES=0,1,2,3
 
 CKPT_DIR="./ckpt_chnsenticorp"
 
@@ -12,7 +12,8 @@ python -u text_classifier.py \
                    --max_seq_len=128 \
                    --warmup_proportion=0.1 \
                    --num_epoch=3 \
-                   --use_data_parallel=True
+                   --use_data_parallel=True \
+                   --network=bilstm
 
 # The sugguested hyper parameters for difference task
 # for ChineseGLUE:

demo/text_classification/run_predict.sh

@@ -7,3 +7,4 @@ python -u predict.py --checkpoint_dir=$CKPT_DIR \
                             --max_seq_len=128 \
                             --use_gpu=True \
                             --batch_size=24 \
+                            --network=bilstm

demo/text_classification/text_classifier.py

@@ -85,11 +85,15 @@ if __name__ == '__main__':
         strategy=strategy)
 
     # Define a classfication finetune task by PaddleHub's API
+    # network choice: bilstm, bow, cnn, dpcnn, gru, lstm
+    # If you wanna add network after ERNIE/BERT/RoBERTa/ELECTRA module,
+    # you must use the outputs["sequence_output"] as the feature of TextClassifierTask
+    # rather than outputs["pooled_output"]
     cls_task = hub.TextClassifierTask(
         data_reader=reader,
         feature=pooled_output,
         feed_list=feed_list,
-        network='dpcnn',
+        network=args.network,
         num_classes=dataset.num_labels,
         config=config,
         metrics_choices=metrics_choices)