{ "architectures": [ "TransformerForCausalLM" ], "attention_bias": false, "auto_map": { "AutoConfig": "configuration_transformer.TransformerConfig", "AutoModelForCausalLM": "modeling_transformer.TransformerForCausalLM" }, "bos_token_id": null, "dtype": "float32", "elementwise_affine": true, "eos_token_id": null, "fuse_cross_entropy": true, "fuse_norm": true, "hidden_act": "swish", "hidden_ratio": 4, "hidden_size": 384, "initializer_range": 0.02, "intermediate_size": null, "max_position_embeddings": null, "model_type": "transformer", "norm_eps": 1e-06, "num_heads": 6, "num_hidden_layers": 4, "num_kv_heads": null, "rope_base": 500000.0, "tie_word_embeddings": false, "transformers_version": "4.57.1", "use_cache": true, "use_rope": true, "vocab_size": 50277, "window_size": null }