training_log.json

{
  "base_model_name": "mpasila_gpt3-finnish-8B-gptq-4bit",
  "base_model_class": "BloomForCausalLM",
  "base_loaded_in_4bit": false,
  "base_loaded_in_8bit": false,
  "projections": "query_key_value",
  "loss": 2.0361,
  "grad_norm": 0.1468171924352646,
  "learning_rate": 3.1034482758620685e-05,
  "epoch": 2.98,
  "current_steps": 1051,
  "train_runtime": 8426.1409,
  "train_samples_per_second": 0.504,
  "train_steps_per_second": 0.004,
  "total_flos": 2618215184203776.0,
  "train_loss": 2.226992231426817
}