| { |
| "architectures": ["RecursiveLanguageModel"], |
| "auto_map": { |
| "AutoConfig": "mixture_of_recursion.RecursiveLanguageModelConfig", |
| "AutoModelForCausalLM": "mixture_of_recursion.RecursiveLanguageModel" |
| }, |
| "model_type": "recursive_language_model", |
| "transformers_version": "5.2.0", |
| "vocab_size": 50260, |
| "embedding_dim": 768, |
| "num_layers": 16, |
| "num_attention_heads": 12, |
| "max_recursion_steps": 5, |
| "max_position_embeddings": 512, |
| "hidden_dropout_prob": 0.1, |
| "attention_dropout_prob": 0.1, |
| "intermediate_size": 3072, |
| "layer_norm_eps": 1e-05, |
| "pad_token_id": 50257, |
| "bos_token_id": 50256, |
| "eos_token_id": 50256, |
| "simple_recursion_steps": 1, |
| "medium_recursion_steps": 3, |
| "complex_recursion_steps": 5, |
| "initializer_range": 0.02, |
| "tie_word_embeddings": false, |
| "dtype": "float32" |
| } |