Use positional unpacking for mlx_lm.load() tuple

mlx_lm.load() returns a tuple (model, tokenizer), not a dict.
The example incorrectly used ** (keyword unpacking) instead of * (positional unpacking).

Author: Anri-Lombard

docs/guide/getting_started.md

@@ -125,7 +125,7 @@ For a quick start, you can find below an example of how to initialize all suppor
     # Create an MLXLM model with the output of mlx_lm.load
     # The model will be downloaded from the HuggingFace hub
     model = outlines.from_mlxlm(
-        **mlx_lm.load("mlx-community/SmolLM-135M-Instruct-4bit")
+        *mlx_lm.load("mlx-community/SmolLM-135M-Instruct-4bit")
     )
     ```