Update README.md

README.md

@@ -7,6 +7,7 @@ widget:
 
 ---
 
+
 # Chinese T5
 
 ## Model description
@@ -54,7 +55,7 @@ python3 preprocess.py --corpus_path corpora/cluecorpussmall.txt \
                       --vocab_path models/google_zh_with_sentinel_vocab.txt \
                       --dataset_path cluecorpussmall_t5_seq128_dataset.pt \
                       --processes_num 32 --seq_length 128 \
-                      --dynamic_masking --target t5 
+                      --dynamic_masking --data_processor t5 
 ```
 
 ```
@@ -65,10 +66,7 @@ python3 pretrain.py --dataset_path cluecorpussmall_t5_seq128_dataset.pt \
                     --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 \
-                    --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 \
-                    --target t5 --tie_weights
+                    --span_masking --span_geo_prob 0.3 --span_max_length 5
 
 ```
 
@@ -79,22 +77,19 @@ python3 preprocess.py --corpus_path corpora/cluecorpussmall.txt \
                       --vocab_path models/google_zh_with_sentinel_vocab.txt \
                       --dataset_path cluecorpussmall_t5_small_seq512_dataset.pt \
                       --processes_num 32 --seq_length 512 \
-                      --dynamic_masking --target t5
+                      --dynamic_masking --data_processor t5
 ```
 
 ```
 python3 pretrain.py --dataset_path cluecorpussmall_t5_seq512_dataset.pt \
-                    --pretrained_model_path models/cluecorpussmall_t5_small_seq128_model.bin-1000000 \
                     --vocab_path models/google_zh_with_sentinel_vocab.txt \
+                    --pretrained_model_path models/cluecorpussmall_t5_small_seq128_model.bin-1000000 \
                     --config_path models/t5/small_config.json \
                     --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 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 \
-                    --target t5 --tie_weights
+                    --span_masking --span_geo_prob 0.3 --span_max_length 5
 ```
 
 Finally, we convert the pre-trained model into Huggingface's format: