{ "vocab_size": 151936, "hidden_size": 2048, "intermediate_size": 6144, "num_hidden_layers": 49, "num_attention_heads": 16, "num_key_value_heads": 4, "head_dim": 128, "attention_dropout": 0.0, "sliding_window_size": 512, "compress_block_size": 16, "num_experts": 25, "num_experts_per_tok": 7, "expert_intermediate_size": 640, "use_shared_expert": true, "router_aux_loss_coef": 0.001, "max_position_embeddings": 4096, "rope_theta": 10000.0, "rms_norm_eps": 1e-06, "hidden_act": "silu", "initializer_range": 0.02, "use_cache": false, "output_router_logits": true, "return_dict": true, "output_hidden_states": false, "dtype": "bfloat16", "chunk_size_feed_forward": 0, "is_encoder_decoder": false, "architectures": [ "AETHERV27wayForCausalLM" ], "id2label": { "0": "LABEL_0", "1": "LABEL_1" }, "label2id": { "LABEL_0": 0, "LABEL_1": 1 }, "problem_type": null, "_name_or_path": "", "transformers_version": "5.3.0", "pad_token_id": 151643, "tie_word_embeddings": false, "model_type": "aether_v2_7way", "output_attentions": false, "auto_map": { "AutoConfig": "configuration_aether_v2_7way.AETHERV27wayConfig", "AutoModelForCausalLM": "modeling_aether_v2_7way.AETHERV27wayForCausalLM", "AutoModel": "modeling_aether_v2_7way.AETHERV27wayModel" } }