{ "final": { "optimizer": "lion", "final_val_loss": 0.8057956205831156, "final_val_ppl": 2.238476768708403, "final_test_loss": 0.8104495330297178, "final_test_ppl": 2.2489187227268475, "final_test_bleu": 71.78000409525386, "total_tokens": 836214, "total_steps": 3731, "trainable_parameters": 33513984 }, "interval_metrics": [ { "tokens": 83757, "step": 373, "val_loss": 0.8596333521268164, "val_ppl": 2.3622944046846746, "test_bleu": 70.53499805805588, "fraction": 0.1 }, { "tokens": 167424, "step": 746, "val_loss": 0.839582419679536, "val_ppl": 2.3153999094448117, "test_bleu": 67.52653732743481, "fraction": 0.2 }, { "tokens": 251309, "step": 1119, "val_loss": 0.8257771701228493, "val_ppl": 2.2836548641830388, "test_bleu": 69.07988653492116, "fraction": 0.3 }, { "tokens": 335115, "step": 1492, "val_loss": 0.8268005358308448, "val_ppl": 2.285993074477995, "test_bleu": 69.81881395594117, "fraction": 0.4 }, { "tokens": 418695, "step": 1866, "val_loss": 0.8206340236737892, "val_ppl": 2.271939844631793, "test_bleu": 73.8244075245152, "fraction": 0.5 }, { "tokens": 502198, "step": 2239, "val_loss": 0.8146513545858799, "val_ppl": 2.2583881584013854, "test_bleu": 69.81881395594117, "fraction": 0.6 }, { "tokens": 585487, "step": 2612, "val_loss": 0.8131937439929398, "val_ppl": 2.2550987058515752, "test_bleu": 70.53499805805588, "fraction": 0.7 }, { "tokens": 668923, "step": 2985, "val_loss": 0.8099305580214223, "val_ppl": 2.2477518929180085, "test_bleu": 69.81881395594117, "fraction": 0.8 }, { "tokens": 752480, "step": 3358, "val_loss": 0.8081816141717777, "val_ppl": 2.2438241367830165, "test_bleu": 67.52653732743481, "fraction": 0.9 }, { "tokens": 836214, "step": 3731, "val_loss": 0.8057956205831156, "val_ppl": 2.238476768708403, "test_bleu": 73.8244075245152, "fraction": 1.0 } ] }