{ "architectures": [ "TTTForCausalLM" ], "bos_token_id": 1, "conv_kernel": 4, "dtype": "float32", "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.02, "intermediate_size": 5504, "max_position_embeddings": 2048, "mini_batch_size": 16, "model_type": "ttt", "num_attention_heads": 32, "num_hidden_layers": 24, "pad_token_id": null, "pre_conv": true, "pretraining_tp": 1, "rms_norm_eps": 1e-06, "rope_theta": 10000.0, "scan_checkpoint_group_size": 0, "share_qk": true, "tie_word_embeddings": true, "transformers_version": "5.3.0", "ttt_base_lr": 1.0, "ttt_layer_type": "mlp", "use_cache": false, "use_gate": true, "vocab_size": 32003 }