| { | |
| "architectures": [ | |
| "ForgettingTransformerForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "auto_map": { | |
| "AutoConfig": "configuration_forgetting_transformer.ForgettingTransformerConfig", | |
| "AutoModelForCausalLM": "modeling_forgetting_transformer.ForgettingTransformerForCausalLM" | |
| }, | |
| "bos_token_id": null, | |
| "decay_time_max": null, | |
| "decay_time_min": null, | |
| "dtype": "float32", | |
| "elementwise_affine": true, | |
| "eos_token_id": null, | |
| "fgate_bias_init": false, | |
| "fgate_type": "full", | |
| "fuse_cross_entropy": true, | |
| "fuse_norm": true, | |
| "hidden_act": "swish", | |
| "hidden_ratio": 4.0, | |
| "hidden_size": 256, | |
| "initializer_range": 0.02, | |
| "intermediate_size": null, | |
| "max_position_embeddings": null, | |
| "model_type": "forgetting_transformer", | |
| "norm_eps": 1e-06, | |
| "num_heads": 4, | |
| "num_hidden_layers": 2, | |
| "num_kv_heads": null, | |
| "ogate_act": "sigmoid", | |
| "qk_norm": false, | |
| "qk_norm_share_param_across_head": false, | |
| "rope_base": 500000.0, | |
| "tie_word_embeddings": false, | |
| "transformers_version": "4.57.1", | |
| "use_cache": true, | |
| "use_k_shift": false, | |
| "use_output_gate": false, | |
| "use_output_norm": false, | |
| "use_rope": false, | |
| "use_v_shift": false, | |
| "vocab_size": 50277, | |
| "window_size": null | |
| } | |