| { | |
| "vocab_size": 50261, | |
| "d_model": 384, | |
| "n_layers": 6, | |
| "n_heads": 6, | |
| "d_head": 64, | |
| "d_ff": 720, | |
| "d_memory": 192, | |
| "max_seq_len": 256, | |
| "dropout": 0.1, | |
| "rope_theta": 10000.0, | |
| "memory_decay_init": 0.99, | |
| "surprise_threshold": 0.1, | |
| "rms_norm_eps": 1e-06, | |
| "pad_token_id": 50256, | |
| "bos_token_id": 50256, | |
| "eos_token_id": 50260, | |
| "model_type": "neurothinker", | |
| "architectures": [ | |
| "NeuroThinkerForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_neurothinker.NeuroThinkerConfig", | |
| "AutoModelForCausalLM": "modeling_neurothinker.NeuroThinkerForCausalLM" | |
| }, | |
| "hidden_size": 384, | |
| "num_hidden_layers": 6, | |
| "num_attention_heads": 6, | |
| "max_position_embeddings": 256, | |
| "use_cache": false | |
| } | |