# DISCLAIMER: This is the configuration file for the GPT-NeoX-20B model as it was trained on 96x 40GB A100

# GPUs. Depending on your system configuration, you may need to change some parameters in order to fit

# the model in memory.



{

  # Tokenizer /  checkpoint settings - you will need to change these to the location you have them saved in

  "vocab-file": "./tokenizer/20B_tokenizer.json",

  "save": "./20B_checkpoints",

  "load": "./20B_checkpoints",  

  "use_wandb": false,

  "use_shared_fs": false,

  

  

  # If finetuning, edit the following to the location of your finetuning dataset:

  "data-path": "./data/pile_20B_tokenizer/pile_20B_tokenizer_text_document",

  # parallelism settings ( you will want to change these based on your cluster setup, ideally scheduling pipeline stages

  # across the node boundaries )

  "pipe-parallel-size": 2,

  "model-parallel-size": 4,



  # model settings

  "num-layers": 22,

  "hidden-size": 6144,

  "num-attention-heads": 64,

  "seq-length": 2048,

  "max-position-embeddings": 2048,

  "norm": "layernorm",

  "pos-emb": "rotary",

  "rotary_pct": 0.25,

  "no-weight-tying": true,

  "gpt_j_residual": true,

  "output_layer_parallelism": "column",

  "scaled-upper-triang-masked-softmax-fusion": false,

  "bias-gelu-fusion": false,



  # init methods

  "init_method": "small_init",

  "output_layer_init_method": "wang_init",



  # optimizer settings

  "optimizer": {

    "type": "Adam",

    "params": {

      "lr": 0.97e-4,

      "betas": [0.9, 0.95],

      "eps": 1.0e-8,

      }

      },



  "min_lr": 0.97e-5,



  # for all zero_optimization options, see https://www.deepspeed.ai/docs/config-json/#zero-optimizations-for-fp16-training

  "zero_optimization": {

  "stage": 1,

  "allgather_partitions": True,

  "allgather_bucket_size": 1260000000,

  "overlap_comm": True,

  "reduce_scatter": True,

  "reduce_bucket_size": 1260000000,

  "contiguous_gradients": True,

  },



  # batch / data settings (assuming 96 GPUs)

  "train_micro_batch_size_per_gpu": 4,

  "gradient_accumulation_steps": 64,

  "data-impl": "mmap",

  "split": "995,4,1",



  # activation checkpointing

  "checkpoint-activations": false,

  "checkpoint-selective": true,

  "checkpoint-full-layers": 8,

  "checkpoint-num-layers": 1,

  "partition-activations": false,

  "synchronize-each-layer": true,

  "zero_allow_untested_optimizer": true,



  # regularization

  "gradient_clipping": 1.0,

  "weight-decay": 0.01,

  "hidden-dropout": 0,

  "attention-dropout": 0,



  # precision settings

  "fp16": {

    "fp16": true,

    "enabled": true,

    "loss_scale": 0,

    "loss_scale_window": 1000,

    "initial_scale_power": 32,

    "hysteresis": 2,

    "min_loss_scale": 1

    },



  # misc. training settings

  "train-iters": 150000,

  "lr-decay-iters": 150000,



  "distributed-backend": "nccl",

  "lr-decay-style": "cosine",

  "warmup": 0.02,

  "checkpoint-factor": 500, # this variable previously called `save-interval`

  "eval-interval": 1000,

  "eval-iters": 10,



  # logging

  "log-interval": 1,

  "steps_per_print": 1,

  "wall_clock_breakdown": false,



  ### NEW DATA: ####

  #"tokenizer_type": "HFTokenizer", //gpt-vocab

  "tokenizer_type": "HFTokenizer", //slim:20B-tokenizer

  "tensorboard-dir": "./tensorboard",

  "log-dir": "./logs",



}