shared_resources/benchmark/manifests/deeper_medium.json

{
  "name": "deeper-medium-mirror",
  "data": {
    "kind": "squaring_mod",
    "data_root": "datasets/deeper_medium",
    "batch_size": 512,
    "eval_batch_size": 512,
    "shuffle_train": true,
    "shuffle_eval": false,
    "num_workers": 2,
    "pin_memory": true,
    "drop_last": true,
    "seed": 45
  },
  "runtime": {
    "device": "cuda:0",
    "dtype": "bfloat16",
    "amp": true,
    "compile": false,
    "total_training_time_seconds": 600,
    "max_steps": 1000000,
    "seeds": [
      74
    ],
    "grad_clip": 1,
    "log_every": 100
  },
  "model_state": {
    "maximum_elements": 500000000
  }
}