training_log.json

{
  "base_model_name": "openlm-research_open_llama_3b",
  "base_model_class": "LlamaForCausalLM",
  "base_loaded_in_4bit": true,
  "base_loaded_in_8bit": false,
  "projections": "q, v",
  "loss": 2.2909,
  "learning_rate": 0.00026989498249708284,
  "epoch": 0.33,
  "current_steps": 3036,
  "train_runtime": 10929.8782,
  "train_samples_per_second": 10.115,
  "train_steps_per_second": 0.079,
  "total_flos": 6.76424738930688e+16,
  "train_loss": 2.387252706162473
}