| { | |
| "architectures": [ | |
| "MixtureOfRecursions" | |
| ], | |
| "model_type": "mixture_of_recursions", | |
| "vocab_size": 31985, | |
| "d_model": 384, | |
| "hidden_size": 384, | |
| "n_layers": 12, | |
| "num_hidden_layers": 12, | |
| "n_heads": 6, | |
| "num_attention_heads": 6, | |
| "dim_feedforward": 2048, | |
| "intermediate_size": 2048, | |
| "max_seq_len": 128, | |
| "max_position_embeddings": 128, | |
| "max_steps": 4, | |
| "dropout": 0.1, | |
| "router_type": "adaptive", | |
| "padding_idx": 0, | |
| "pos_encoding": "learned", | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.57.0", | |
| "auto_map": { | |
| "AutoConfig": "model_slm.MixtureOfRecursionsConfig", | |
| "AutoModel": "model_slm.MixtureOfRecursions", | |
| "AutoModelForCausalLM": "model_slm.MixtureOfRecursions" | |
| } | |
| } |