config_ner.json

{
  "model_name_or_path": "./2sf:10sf",
  "tokenizer_name": "bert-base-uncased",
  "dataset_name": "conll2003",
  "report_to": "all",
  "max_seq_length": 512,
  "preprocessing_num_workers": 8,
  "output_dir": "./tmp/ner",
  "cache_dir": "./.cache",
  "overwrite_output_dir": true,
  "do_train": true,
  "do_eval": true,
  "evaluation_strategy": "epoch",
  "logging_first_step" :true,
  "logging_strategy": "epoch",
  "gradient_accumulation_steps": 1,
  "eval_accumulation_steps": 25,
  "dataloader_num_workers": 8,
  "per_device_train_batch_size": 64,
  "per_device_eval_batch_size": 64,
  "deepspeed": "./ds_zero2_1gpu.json",
  "fp16": true,
  "adam_beta1": 0.9,
  "adam_beta2":0.999,
  "adam_epsilon": 1e-08,
  "learning_rate": 5e-5,
  "weight_decay": 0.01,
  "lr_scheduler_type": "cosine_with_restarts",
  "warmup_ratio": 0,
  "num_train_epochs": 5,
  "save_steps": 1000,
  "save_total_limit": 10,
  "save_on_each_node": true,
  "seed": 42
}