| { |
| "architectures": [ |
| "RecursiveLanguageModel" |
| ], |
| "auto_map": { |
| "AutoConfig": "mixture_of_recursion.RecursiveLanguageModelConfig", |
| "AutoModel": "mixture_of_recursion.RecursiveLanguageModel", |
| "AutoModelForCausalLM": "mixture_of_recursion.RecursiveLanguageModel" |
| }, |
| "model_type": "recursive_language_model", |
| "transformers_version": "4.35.0", |
| |
| "vocab_size": 50257, |
| "embedding_dim": 512, |
| "num_layers": 6, |
| "num_attention_heads": 8, |
| "max_recursion_steps": 5, |
| "max_position_embeddings": 512, |
| "hidden_dropout_prob": 0.1, |
| "attention_dropout_prob": 0.1, |
| "intermediate_size": 2048, |
| "layer_norm_eps": 1e-5, |
| |
| "pad_token_id": 50256, |
| "bos_token_id": 50256, |
| "eos_token_id": 50256, |
| |
| "simple_recursion_steps": 1, |
| "medium_recursion_steps": 3, |
| "complex_recursion_steps": 5, |
| |
| "confidence_threshold": 0.8, |
| "use_adaptive_stopping": true, |
| "initializer_range": 0.02, |
| |
| "tie_word_embeddings": true |
| } |