training_log.json

{
  "base_model_name": "TheBloke_CodeLlama-34B-Instruct-fp16",
  "base_model_class": "LlamaForCausalLM",
  "base_loaded_in_4bit": true,
  "base_loaded_in_8bit": false,
  "projections": "q, v",
  "loss": 1.3565,
  "learning_rate": 0.0,
  "epoch": 2.53,
  "current_steps": 451,
  "train_runtime": 1472.9079,
  "train_samples_per_second": 1.542,
  "train_steps_per_second": 0.01,
  "total_flos": 4.971878642456986e+16,
  "train_loss": 1.9034520785013835
}