{ "step": 999, "val_loss": 2.0692103222040337, "mmlu_acc": 0.2685546875, "arc_easy_acc": 0.251953125, "model_config": { "sequence_len": 1024, "vocab_size": 65536, "n_layer": 16, "n_head": 8, "n_kv_head": 8, "n_embd": 1024 } }