{ "architectures": [ "MambaForHF" ], "bos_token_id": 1, "d_discr": 256, "d_input": 2048, "d_model": 4096, "d_state": 1024, "eos_token_id": 2, "ker_size": 4, "max_position_embeddings": 131072, "num_layers": 64, "pad_token_id": 0, "parallel": true, "torch_dtype": "float32", "transformers_version": "4.44.2", "vocab_size": 128256 }