training_log.json

{
  "base_model_name": "jeiku_Rosa_v1_7B",
  "base_model_class": "MistralForCausalLM",
  "base_loaded_in_4bit": false,
  "base_loaded_in_8bit": false,
  "projections": "q, v",
  "loss": 1.4852,
  "learning_rate": 2.727272727272727e-05,
  "epoch": 14.55,
  "current_steps": 465,
  "train_runtime": 254.5375,
  "train_samples_per_second": 15.499,
  "train_steps_per_second": 0.059,
  "total_flos": 4.2428341420032e+16,
  "train_loss": 2.0015284220377603
}