{ "architectures": [ "MicroGPT" ], "model_type": "micro_gpt", "arch_name": "micro_gpt_tf", "vocab_size": 74, "block_size": 128, "n_layer": 2, "n_head": 4, "n_embd": 128, "dropout": 0.1, "n_params": 431178, "dataset": "tinystories", "max_steps": 500, "best_val_loss": 2.213334321975708, "training_step": 500, "hidden_act": "gelu", "initializer_range": 0.02, "layer_norm_eps": 1e-05, "use_cache": true, "bos_token_id": 0, "eos_token_id": 0, "pad_token_id": 0 }