DeepSpeed报错

[dancingpipi]

我的执行脚本为:

deepspeed pretrain.py --deepspeed \
    --deepspeed_config "ds_config_test.json" \
    --dataset_path "/home/work/uer_py/data/dataset.pt" \
    --vocab_path "/home/work/uer_py/train_bert/vocab.txt" \
    --output_model_path "/home/work/uer_py/models" \
    --world_size 4 \
    --total_steps 10000000 \
    --save_checkpoint_steps 100000 \
    --batch_size 64 \
    --embedding "word_pos_seg" \
    --encoder "transformer" \
    --mask "fully_visible" \
    --target "bert"

ds_config_test.json为：

{
  "gradient_accumulation_steps": 1,
  "train_micro_batch_size_per_gpu":1,
  "steps_per_print": 100,
  "optimizer": {
    "type": "Adam",
    "params": {
      "lr": 0.00001,
      "weight_decay": 1e-2
    }
  },
  "flops_profiler": {
    "enabled": true,
    "profile_step": 1,
    "module_depth": -1,
    "top_modules": 3,
    "detailed": true
  },
  "fp16": {
    "enabled": true,
    "loss_scale": 0,
    "loss_scale_window": 1000,
    "hysteresis": 2,
    "min_loss_scale": 1
  },
  "zero_optimization": {
      "stage": 2,
      "cpu_offload": true,
      "contiguous_gradients": false,
      "overlap_comm": false,
      "reduce_scatter": false,
      "reduce_bucket_size": 5e7,
      "allgather_bucket_size": 5e7
  },
  "activation_checkpointing": {
      "partition_activations": false,
      "contiguous_memory_optimization": false,
      "cpu_checkpointing": false
  },
  "wall_clock_breakdown": false,
  "zero_allow_untested_optimizer": true
}

执行后会报这个错误：

AssertionError: Client Optimizer (type = <class 'NoneType'> is not instantiated but Client LR Scheduler is instantiated

[hhou435]

您好，我们测试的deepspeed版本是0.3.15，如果使用更高的版本可能会由于deepspeed代码的更新造成一些错误，这个我们后续会进行解决，您可以先尝试降低版本