fix config

config.json

@@ -8,25 +8,26 @@
     "AutoModel": "modeling_openpangu_moe.PanguUltraMoEModel",
     "AutoModelForCausalLM": "modeling_openpangu_moe.PanguUltraMoEForCausalLM"
   },
-  "num_dense_layers": 3,
+  "first_k_dense_replace": 3,
   "hidden_act": "silu",
   "hidden_size": 7680,
   "initializer_range": 0.02,
   "intermediate_size": 18432,
-  "attention_kv_lora_dim": 512,
+  "kv_lora_rank": 512,
   "max_position_embeddings": 131072, 
   "model_type": "pangu_ultra_moe",
   "moe_intermediate_size": 2048,
-  "num_routed_experts": 256,
-  "num_shared_experts": 1,
+  "n_routed_experts": 256,
+  "n_shared_experts": 1,
+  "norm_topk_prob": true,
   "num_attention_heads": 128,
   "num_experts_per_tok": 8,
   "num_hidden_layers": 61,
   "num_key_value_heads": 128,
-  "num_mtp_layers": 1,
-  "attention_q_lora_dim": 1536,
-  "attention_qk_dim": 128,
-  "attention_qk_rope_dim": 64,
+  "num_nextn_predict_layers": 1,
+  "q_lora_rank": 1536,
+  "qk_nope_head_dim": 128,
+  "qk_rope_head_dim": 64,
   "rms_norm_eps": 1e-05,
   "rope_theta": 25600000,
   "routed_scaling_factor": 2.5,
@@ -35,6 +36,6 @@
   "torch_dtype": "bfloat16",
   "transformers_version": "4.48.2",
   "use_cache": true,
-  "attention_v_dim": 128,
+  "v_head_dim": 128,
   "vocab_size": 153600
-}
+}