training_parameters.json

{
  "lora_name": "TinyTR",
  "always_override": false,
  "save_steps": 350.0,
  "micro_batch_size": 6,
  "batch_size": 0,
  "epochs": 0.25,
  "learning_rate": "3e-4",
  "lr_scheduler_type": "linear",
  "lora_rank": 64,
  "lora_alpha": 128,
  "lora_dropout": 0.1,
  "cutoff_len": 768,
  "dataset": "turkish",
  "eval_dataset": "None",
  "format": "Chatml",
  "eval_steps": 0.0,
  "raw_text_file": "None",
  "higher_rank_limit": false,
  "warmup_steps": 120.0,
  "optimizer": "adamw_torch",
  "hard_cut_string": "\\n\\n\\n",
  "train_only_after": "",
  "stop_at_loss": 0,
  "add_eos_token": true,
  "min_chars": 0.0,
  "report_to": "None",
  "precize_slicing_overlap": true,
  "add_eos_token_type": "Every Block",
  "save_steps_under_loss": 0,
  "add_bos_token": true,
  "training_projection": "q-k-v-o",
  "sliding_window": false,
  "warmup_ratio": 0,
  "grad_accumulation": 2,
  "neft_noise_alpha": 6
}