{ "architectures": [ "MamLAForCausalLM" ], "bos_token_id": 1, "d_model": 2048, "dtype": "float32", "eos_token_id": 2, "mamba_d_conv": 4, "mamba_d_state": 64, "mamba_expand": 2, "mamba_layers_per_cycle": 7, "mamba_n_heads": 16, "mla_kv_lora_rank": 128, "mla_layers_per_cycle": 1, "mla_n_heads": 16, "mla_q_lora_rank": 128, "mla_qk_nope_head_dim": 32, "mla_qk_rope_head_dim": 32, "mla_v_head_dim": 32, "model_type": "mamla", "moe_d_ff": 1024, "moe_num_experts": 4, "moe_top_k": 2, "n_layer": 24, "num_experts": 4, "num_experts_per_tok": 2, "num_layers": 16, "pad_token_id": 0, "pad_vocab_size_multiple": 8, "tie_word_embeddings": true, "transformers_version": "5.0.0", "vocab_size": 32000 }