Add code

README.txt

@@ -0,0 +1,20 @@
+1. Environment Requirements
+* Ubuntu 16.04
+* Anaconda with Python 3.6.9
+* CUDA 10.0
+* JAVA jdk1.8.0_121 
+* Hadoop 2.9.2-SNAPSHOT
+* Horovod 0.19.5
+Note: The complete python package list of our environment is included in the requirements.txt.
+The installation may need several minutes if there is no environmental conflicts.
+
+2. Hardware requirements
+Needs a server with at least one Tesla V100 GPU.
+
+3. Training and Testing
+* Download datasets and pretrained language models from their original sources
+* Change the path names and data file names. If you have K GPUs, need to split the entire training data into K folds.
+* Execute "sh run.sh"
+
+Note: The logs at the training stage will show the training loss and accuracy. Logs at the test stage will show the test results. The sample codes usually run for a few minutes.
+

config.json

@@ -0,0 +1,25 @@
+{
+  "architectures": [
+    "BertForMaskedLM"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "directionality": "bidi",
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "bert",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "pooler_fc_size": 768,
+  "pooler_num_attention_heads": 12,
+  "pooler_num_fc_layers": 3,
+  "pooler_size_per_head": 128,
+  "pooler_type": "first_token_transform",
+  "type_vocab_size": 2,
+  "vocab_size": 21128
+}