Update README.md

README.md

@@ -43,7 +43,7 @@ When the parameter skip_special_tokens is False:
 
 ## Training data
 
-Contains 700,000 Chinese couplets collected by [couplet-clean-dataset](https://github.com/v-zich/couplet-clean-dataset).
+Training data contains 700,000 Chinese couplets which are collected by [couplet-clean-dataset](https://github.com/v-zich/couplet-clean-dataset).
 
 ## Training procedure
 
@@ -52,14 +52,14 @@ Models are pre-trained by [UER-py](https://github.com/dbiir/UER-py/) on [Tencent
 ```
 python3 preprocess.py --corpus_path corpora/couplet.txt \
 					  --vocab_path models/google_zh_vocab.txt \  
-					  --dataset_path couplet.pt --processes_num 16 \
+					  --dataset_path couplet_dataset.pt --processes_num 16 \
 					  --seq_length 64 --target lm 
 ```
 
 ```
-python3 pretrain.py --dataset_path couplet.pt \
+python3 pretrain.py --dataset_path couplet_dataset.pt \
 				    --vocab_path models/google_zh_vocab.txt \
-					--output_model_path models/couplet_gpt_base_model.bin \  
+					--output_model_path models/couplet_gpt2_base_model.bin \  
 					--config_path models/bert_base_config.json --learning_rate 5e-4 \
 					--tie_weight --world_size 8 --gpu_ranks 0 1 2 3 4 5 6 7 \
 					--embedding word_pos --remove_embedding_layernorm \
@@ -71,7 +71,7 @@ python3 pretrain.py --dataset_path couplet.pt \
 
 Finally, we convert the pre-trained model into Huggingface's format:
 ```
-python3 scripts/convert_gpt2_from_uer_to_huggingface.py --input_model_path couplet_gpt_base_model.bin \
+python3 scripts/convert_gpt2_from_uer_to_huggingface.py --input_model_path couplet_gpt2_base_model.bin-25000 \
 														--output_model_path pytorch_model.bin \
 														--layers_num 12
 ```