{ "architectures": [ "DogeForSequenceClassification" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": null, "dtype": "float32", "eos_token_id": null, "hidden_act": "silu", "hidden_dropout": 0.0, "hidden_size": 32, "initializer_range": 0.02, "intermediate_size": 64, "is_decoder": false, "is_moe": false, "keep_window_size": 2048, "max_position_embeddings": 512, "mlp_bias": false, "model_type": "doge", "norm_topk_prob": false, "num_attention_heads": 4, "num_experts": 16384, "num_experts_per_tok": 64, "num_hidden_layers": 2, "num_key_value_heads": 4, "output_router_logits": false, "pad_token_id": 2, "rms_norm_eps": 1e-06, "rope_parameters": { "rope_theta": 10000.0, "rope_type": "default" }, "router_aux_loss_coef": 0.001, "sliding_window": null, "tie_word_embeddings": false, "transformers_version": "5.16.0.dev0", "type_vocab_size": 16, "use_cache": true, "vocab_size": 32768 }