config.json

{
 "tier": "base",
 "d": 512,
 "heads": 8,
 "e": 128,
 "buckets": 65536,
 "recursions": 6,
 "interact": 2,
 "mlp": 1536,
 "lr": 0.00015,
 "budget": 16384,
 "layout": {
  "max_len": 1024,
  "long_max_len": 2048,
  "long_opts_threshold": 24,
  "head_max_len": 192,
  "max_tok_per_opt": 24
 },
 "temperatures": [
  [
   2.223009569218283,
   3.850024543056701,
   4.047131644703235,
   1.5673085376630793
  ],
  [
   7.009216863860871,
   7.009216863860871,
   7.009216863860871,
   7.009216863860871
  ],
  [
   11.548028041677576,
   11.548028041677576,
   11.548028041677576,
   11.548028041677576
  ]
 ],
 "pretrain": null
}