{ "architectures": [ "SDARForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "auto_map": { "AutoConfig": "configuration_sdar.SDARConfig", "AutoModel": "modeling_sdar.SDARForCausalLM", "AutoModelForCausalLM": "modeling_sdar.SDARForCausalLM" }, "block_size": 4, "bos_token_id": 2, "debug": false, "eos_token_id": 3, "ep_size": 1, "fuse_cross_entropy": true, "head_dim": 128, "hidden_act": "silu", "hidden_size": 768, "initializer_range": 0.02, "intermediate_size": 3072, "mask_token_id": 2196, "max_position_embeddings": 2048, "max_window_layers": 24, "micro_forward": false, "model_type": "sdar", "num_attention_heads": 12, "num_hidden_layers": 12, "num_key_value_heads": 2, "rms_norm_eps": 1e-06, "rope_scaling": null, "rope_theta": 1000000, "skip_checkpoint": false, "sliding_window": 2048, "tie_word_embeddings": true, "torch_dtype": "bfloat16", "transformers_version": "4.52.4", "use_cache": false, "use_deepep": false, "use_sliding_window": false, "vocab_size": 2200 }