Update model type in config.json

The model type is currently set to qwen2_5_vl_text, which causes a ModuleNotFoundError: No module named 'transformers.models.qwen2_5_vl_text' when loading any variant of Qwen2.5-7b-Instruct available in the Unsloth mapper. The incorrect model type qwen2_5_vl_text gets passed to the model_types variable in the unsloth-zoo compiler, which then triggers the error when exec(f"import {model_location}", globals()) is executed.
Change the model type from qwen2_5_vl_text to qwen2_5_vl to match the correct Transformers module path.

config.json

@@ -159,7 +159,7 @@
     ],
     "max_position_embeddings": 128000,
     "max_window_layers": 28,
-    "model_type": "qwen2_5_vl_text",
+    "model_type": "qwen2_5_vl",
     "num_attention_heads": 28,
     "num_hidden_layers": 28,
     "num_key_value_heads": 4,