{ "step": 700, "val_loss": 0.8010137677192688, "mmlu_acc": 0.42578125, "arc_easy_acc": 0.7197265625, "model_config": { "sequence_len": 2048, "vocab_size": 65536, "n_layer": 34, "n_head": 17, "n_kv_head": 17, "n_embd": 2176 } }