Remove _base_model field with Qwen2.5 reference

config.json

@@ -11,7 +11,6 @@
   "max_position_embeddings": 131072,
   "thinking_tokens": 1000000,
   "_name_or_path": "zenlm/zen-next-80b-instruct",
-  "_base_model": "Qwen/Qwen2.5-72B-Instruct",
   "hidden_size": 8192,
   "num_hidden_layers": 80,
   "num_attention_heads": 64,