Update README.md

README.md

@@ -2,7 +2,7 @@
 language: Chinese
 datasets: CLUECorpusSmall
 widget: 
-- text: "作为电子为主的电商平台，京东绝对是extra0者。如今的刘强extra1已经是身价过extra2的老板。"
+- text: "作为电子extra0的平台，京东绝对是领先者。如今的刘强extra1已经是身价过extra2的老板。"
 
 
 ---
@@ -59,7 +59,7 @@ python3 preprocess.py --corpus_path corpora/cluecorpussmall.txt \
 python3 pretrain.py --dataset_path cluecorpussmall_t5_seq128_dataset.pt \
                     --vocab_path models/google_zh_with_sentinel_vocab.txt \
                     --config_path models/t5/small_config.json \
-                    --output_model_path models/cluecorpussmall_t5_seq128_model.bin \
+                    --output_model_path models/cluecorpussmall_t5_small_seq128_model.bin \
                     --world_size 8 --gpu_ranks 0 1 2 3 4 5 6 7 \
                     --total_steps 1000000 --save_checkpoint_steps 100000 --report_steps 50000 \
                     --learning_rate 1e-3 --batch_size 64 \
@@ -75,20 +75,20 @@ Stage2:
 ```
 python3 preprocess.py --corpus_path corpora/cluecorpussmall.txt \
                       --vocab_path models/google_zh_with_sentinel_vocab.txt \
-                      --dataset_path cluecorpussmall_t5_seq512_dataset.pt \
+                      --dataset_path cluecorpussmall_t5_small_seq512_dataset.pt \
                       --processes_num 32 --seq_length 512 \
                       --dynamic_masking --target t5
 ```
 
 ```
 python3 pretrain.py --dataset_path cluecorpussmall_t5_seq512_dataset.pt \
-                    --pretrained_model_path models/cluecorpussmall_t5_seq128_model.bin-1000000 \
+                    --pretrained_model_path models/cluecorpussmall_t5_small_seq128_model.bin-1000000 \
                     --vocab_path models/google_zh_with_sentinel_vocab.txt \
                     --config_path models/t5/small_config.json \
-                    --output_model_path models/cluecorpussmall_t5_seq512_model.bin \
+                    --output_model_path models/cluecorpussmall_t5_small_seq512_model.bin \
                     --world_size 8 --gpu_ranks 0 1 2 3 4 5 6 7 \
                     --total_steps 250000 --save_checkpoint_steps 50000 --report_steps 10000 \
-                    --learning_rate 1e-3 --batch_size 16 \
+                    --learning_rate 5e-4 --batch_size 16 \
                     --span_masking --span_geo_prob 0.3 --span_max_length 5 \
                     --embedding word --relative_position_embedding --remove_embedding_layernorm --tgt_embedding word \
                     --encoder transformer --mask fully_visible --layernorm_positioning pre --decoder transformer \
@@ -98,7 +98,7 @@ python3 pretrain.py --dataset_path cluecorpussmall_t5_seq512_dataset.pt \
 Finally, we convert the pre-trained model into Huggingface's format:
 
 ```
-python3 scripts/convert_t5_from_uer_to_huggingface.py --input_model_path cluecorpussmall_t5_seq512_model.bin-250000 \
+python3 scripts/convert_t5_from_uer_to_huggingface.py --input_model_path cluecorpussmall_t5_small_seq512_model.bin-250000 \
                                                       --output_model_path pytorch_model.bin \
                                                       --layers_num 6 \
                                                       --type t5