{ "step": 703, "val_loss": 1.1132323741912842, "mmlu_acc": 0.28515625, "arc_easy_acc": 0.30859375, "model_config": { "sequence_len": 2048, "vocab_size": 65536, "n_layer": 20, "n_head": 10, "n_kv_head": 10, "n_embd": 1280 } }