zerozeroz/CodeLlama-7b-hf_1000rl_cpp_javascript
028
1{2 "total_flos": 0.0,3 "train_loss": 0.04159710054096379,4 "train_runtime": 3805.4602,5 "train_samples": 456,6 "train_samples_per_second": 0.359,7 "train_steps_per_second": 0.068}