{ "step": 669, "val_loss": 2.058673143386841, "mmlu_acc": 0.263671875, "arc_easy_acc": 0.2470703125, "model_config": { "sequence_len": 2048, "vocab_size": 65536, "n_layer": 4, "n_head": 2, "n_kv_head": 2, "n_embd": 256 } }