training_log.json

{
  "base_model_name": "Llama2-13B-GPTQ",
  "base_model_class": "LlamaForCausalLM",
  "base_loaded_in_4bit": false,
  "base_loaded_in_8bit": false,
  "projections": "q, v",
  "loss": 1.7243,
  "learning_rate": 6.733428315694157e-07,
  "epoch": 1.0,
  "current_steps": 9599,
  "train_runtime": 3878.3685,
  "train_samples_per_second": 2.479,
  "train_steps_per_second": 0.019,
  "total_flos": 3195128512512000.0,
  "train_loss": 1.8589664967854818
}