Created by: sz547073165

看到这个预训练 (ERNIE 1.0)想用自己的数据进行预训练

执行了这个脚本script/zh_task/pretrain.sh

这个脚本跑不起来 报错 `[root@localhost zh_task]# sh ./pretrain.sh

• export FLAGS_eager_delete_tensor_gb=0
• FLAGS_eager_delete_tensor_gb=0
• export FLAGS_sync_nccl_allreduce=1
• FLAGS_sync_nccl_allreduce=1
• export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
• CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 ++ hostname -i
• python3 /root/ERNIE/pretrain_launch.py --nproc_per_node 8 --selected_gpus 0,1,2,3,4,5,6,7 --node_ips ::1 127.0.0.1 --node_id 0 ./train.py --use_cuda True --is_distributed False --use_fast_executor True --weight_sharing True --in_tokens true --batch_size 8192 --vocab_path ./config/vocab.txt --train_filelist ./data/train_filelist --valid_filelist ./data/valid_filelist --validation_steps 100 --num_train_steps 1000000 --checkpoints ./checkpoints --save_steps 10000 --ernie_config_path ./config/ernie_config.json --learning_rate 1e-4 --use_fp16 false --weight_decay 0.01 --max_seq_len 512 --skip_steps 10 usage: pretrain_launch.py [-h] [--ernie_config_path ERNIE_CONFIG_PATH] [--init_checkpoint INIT_CHECKPOINT] [--checkpoints CHECKPOINTS] [--weight_sharing WEIGHT_SHARING] [--generate_neg_sample GENERATE_NEG_SAMPLE] [--epoch EPOCH] [--learning_rate LEARNING_RATE] [--lr_scheduler {linear_warmup_decay,noam_decay}] [--weight_decay WEIGHT_DECAY] [--num_train_steps NUM_TRAIN_STEPS] [--warmup_steps WARMUP_STEPS] [--save_steps SAVE_STEPS] [--validation_steps VALIDATION_STEPS] [--use_fp16 USE_FP16] [--use_dynamic_loss_scaling USE_DYNAMIC_LOSS_SCALING] [--init_loss_scaling INIT_LOSS_SCALING] [--incr_every_n_steps INCR_EVERY_N_STEPS] [--decr_every_n_nan_or_inf DECR_EVERY_N_NAN_OR_INF] [--incr_ratio INCR_RATIO] [--decr_ratio DECR_RATIO] [--skip_steps SKIP_STEPS] [--verbose VERBOSE] [--train_filelist TRAIN_FILELIST] [--valid_filelist VALID_FILELIST] [--test_filelist TEST_FILELIST] [--vocab_path VOCAB_PATH] [--max_seq_len MAX_SEQ_LEN] [--batch_size BATCH_SIZE] [--in_tokens IN_TOKENS] [--is_distributed IS_DISTRIBUTED] [--use_cuda USE_CUDA] [--use_fast_executor USE_FAST_EXECUTOR] [--do_test DO_TEST] pretrain_launch.py: error: unrecognized arguments: --node_id 0 ./train.py `