{ "architectures": [ "Gemma4ForCausalLM" ], "model_type": "gemma4_text", "vocab_size": 32768, "hidden_size": 640, "intermediate_size": 1792, "num_hidden_layers": 28, "num_attention_heads": 10, "num_key_value_heads": 2, "head_dim": 64, "global_head_dim": 128, "per_layer_config": { "3": { "head_dim": 128 }, "7": { "head_dim": 128 }, "11": { "head_dim": 128 }, "15": { "head_dim": 128 }, "19": { "head_dim": 128 }, "23": { "head_dim": 128 }, "27": { "head_dim": 128 } }, "hidden_activation": "gelu_pytorch_tanh", "max_position_embeddings": 8192, "rms_norm_eps": 1e-06, "sliding_window": 512, "layer_types": [ "sliding_attention", "sliding_attention", "sliding_attention", "full_attention", "sliding_attention", "sliding_attention", "sliding_attention", "full_attention", "sliding_attention", "sliding_attention", "sliding_attention", "full_attention", "sliding_attention", "sliding_attention", "sliding_attention", "full_attention", "sliding_attention", "sliding_attention", "sliding_attention", "full_attention", "sliding_attention", "sliding_attention", "sliding_attention", "full_attention", "sliding_attention", "sliding_attention", "sliding_attention", "full_attention" ], "final_logit_softcapping": 15.0, "tie_word_embeddings": true, "hidden_size_per_layer_input": 0, "vocab_size_per_layer_input": 32768, "num_kv_shared_layers": 0, "use_double_wide_mlp": false, "attention_k_eq_v": false, "enable_moe_block": false, "attention_bias": false, "attention_dropout": 0.0, "use_cache": true, "rope_parameters": { "sliding_attention": { "rope_type": "default", "rope_theta": 10000.0 }, "full_attention": { "rope_type": "proportional", "partial_rotary_factor": 1.0, "rope_theta": 1000000.0 } }, "pad_token_id": 0, "eos_token_id": 1, "bos_token_id": 2, "dtype": "bfloat16" }